Keywords: Learning, Autonomous systems, Mechanical systems/robotics
Abstract: Deep Reinforcement Learning (DRL) has demonstrated excellent performance in various domains. However, learning continuous control is a challenging problem as it requires a relatively more samples for training. Many DRL methods aim to improve sample efficiency through various exploration methods and state-of-the-art actor-critic and policy gradient methods. However, this often comes at the cost of a high amount of computational resources. Motivated by the success of value-based methods for approximating state-action values, such as the radial basis function deep Q-network (RBF-DQN), we explore the potential of value-based DRL for learning continuous control for mobile robot navigation using sparse LiDAR findings. Our comparative study against the state-of-the-art actor-critic algorithm reveals the potential of the value-based DRL for learning continuous robotic navigation tasks through action maximization and accurate value function approximation. The findings highlight the RBF-DQN model's superior performance, comparable to state-of-the-art actor-critic methods, and its enhanced sample efficiency in achieving successful navigation and goal completion in both simulated and real-time mapless navigation.

Keywords: Mechanical systems/robotics, Adaptive systems, Learning
Abstract: Model-free reinforcement learning (RL) has enabled adaptable and agile quadruped locomotion; however, policies often converge to a single gait, leading to suboptimal performance. Traditionally, Model Predictive Control (MPC) has been extensively used to obtain task-specific optimal policies but lacks the ability to adapt to varying environments. To address these limitations, we propose an optimization framework for real-time gait adaptation in a continuous gait space, combining the Model Predictive Path Integral (MPPI) algorithm with a Dreamer module to produce adaptive and optimal policies for quadruped locomotion. At each time step, MPPI jointly optimizes the actions and gait variables using a learnt Dreamer reward that promotes velocity tracking, energy efficiency, stability, and smooth transitions, while penalizing abrupt gait changes. A learned value function is incorporated as terminal reward, extending the formulation to an infinite-horizon planner. We evaluate our framework in simulation on the Unitree Go1, demonstrating an average reduction of up to 36.48 % in energy consumption across varying target speeds, while maintaining accurate tracking and adaptive, task-appropriate gaits.

Keywords: Machine learning, Modeling and simulation, Mechanical systems/robotics
Abstract: This paper presents a reinforcement learning-based control method for a human biomechanical system with exoskeletons that can be used in applications such as musculoskeletal rehabilitation, safety and ergonomics analysis, and industrial task simulation. We apply Reinforcement Learning (RL) to control a full-body biomechanical model consisting of two arms, two legs and a torso with a pivoted pelvis. It consists of a total of 212 muscles and 220 tendons. We also apply this method to concurrently control four motor-actuated exoskeletons located at the elbows and knees, that collectively control 107 joints. As this is a highly overactuated system, uncorrelated exploration strategies for RL are unable to converge to controllers with acceptable tracking performance in a computationally efficient manner. To mitigate this, we apply Differential Extrinsic Plasticity for generating correlated noise during RL exploration. We incorporate this RL-based control method into a motion capture pipeline, allowing real-time tracking of human body movements during industrial and daily activities at a granular musculoskeletal level, going beyond kinematic tracking typical for motion capture studies. We evaluate the position tracking errors for motion capture data in conditions with and without the exoskeletons attached, and compare the corresponding activations in major muscle groups.

Keywords: Agents-based systems, Neural networks, Simulation
Abstract: Safe navigation in unknown and dynamic environments remains a key challenge for autonomous multi-robot systems. Traditional methods often depend on global maps or fixed heuristics, limiting scalability and adaptability in complex, unstructured settings. In this work, we propose a risk-aware multi-agent deep reinforcement learning (MADRL) framework for mapless navigation through static and dynamic obstacles. Each robot learns an independent policy using local LiDAR, goal, and velocity data, with centralized training and decentralized execution. To enhance safety, we introduce a sequence-based risk critic that predicts continuous risk for each action from recent state-action histories. Instead of binary collision labels, we generate soft risk labels using a two-threshold scheme on future LiDAR data, allowing agents to anticipate and avoid imminent danger. Integrating this risk critic into the reward function fosters nuanced, proactive avoidance. Experiments in ROS2-Gazebo with moving obstacles and unseen layouts show improved success and lower collision rates versus MADDPG. The lightweight networks support real-time onboard inference and sim-to-real deployment.

Keywords: Flight control, Nonlinear systems, Machine learning
Abstract: Loss-of-control (LOC) remains a leading cause of fixed-wing aircraft accidents, especially in post-stall and flat-spin regimes where conventional gain-scheduled or logic-based recovery laws may fail. This study formulates spin-recovery as a continuous-state, continuous-action Markov Decision Process and trains a Proximal Policy Optimization (PPO) agent on a high-fidelity six-degree-of-freedom F-18/HARV model that includes nonlinear aerodynamics, actuator saturation and rate coupling. A two-phase potential-based reward structure first penalizes large angular rates and then enforces trimmed flight. After 6,000 simulated episodes, the policy generalities to unseen upset initializations. Results show that the learned policy successfully arrests the angular rates and stabilizes the angle of attack. The controller performance is observed to be satisfactory for recovery from spin condition which was compared with a state-of-the-art sliding mode controller. The findings demonstrate that deep reinforcement learning can deliver interpretable, dynamically feasible manoeuvres for real-time loss of control mitigation and provide a pathway for flight-critical RL deployment.

Keywords: Learning, Control education, Machine learning
Abstract: The Deep Deterministic Policy Gradient (DDPG) controller for an Arduino-based second-order RC series circuit is designed and implemented in this study. Gaussian reward shaping was implemented to improve learning performance. Initially, Bayesian Optimization was used to optimize the parameters scaling factor (λ) and standard deviation (σ). The value of σ was further adjusted using the Lyapunov stability criteria in order to guarantee stability and enhance convergence. After being trained in a simulation environment and implemented on real hardware, the reinforcement learning(RL) based controller showed dynamic response and efficient voltage management. The suggested controller’s performance was contrasted with that of reference H-infinity (H∞) and Proportional Integral Derivative (PID) controllers. According to the results, the DDPG controller with reward shaping achieves better tracking performance and robustness under varying input and disturbances. For real-time embedded systems, this work shows a novel integration of control theory, optimization, and reward shaping.

Keywords: Cooperative control, Optimization, Agents-based systems
Abstract: Existing distributed optimization (DO) algorithms, primarily effective for linear dynamics, often fall short when attempting to accurately represent systems or datasets displaying multiplicative dynamics, scale invariance, or power-law relationships. We address these limitations by proposing a novel adaptive distributed optimization (DO) algorithm, enabling multi-agent systems (MAS) on weight-unbalanced directed networks to minimize a global convex function. Leveraging logarithmic transformation, we prove the asymptotic convergence of agents' states to the unique optimum. Since the convergence depends on neither network structures nor cost functions, the proposed DO dynamics offers high scalability across diverse network configurations and convex cost functions that obey power-laws. Numerical simulations validate our theoretical advancements, showing the algorithm's superior performance compared to state-of-the-art, especially when minimizing cost functions with power-law relationships.

Keywords: Learning, Machine learning, Optimization
Abstract: This paper investigates Federated Learning (FL) within the framework of Stochastic Approximation (SA). FL is a decentralized approach to training machine learning models collaboratively across multiple clients, eliminating the need for centralized data storage. Each client updates a local model using its own data and periodically transmits the model parameters to a central server. The server aggregates these parameters and redistributes the aggregated model to the clients, who then continue training from the updated state. SA is an iterative optimization technique that employs noisy or approximate gradient estimates along with a diminishing step size to converge to a minimizer of a cost function. In the proposed approach, clients update their local model parameters using a stochastic approximation scheme. It is shown that the evolution of the aggregated model parameters closely follows the trajectory of an autonomous ordinary differential equation (ODE). Numerical experiments are conducted to evaluate the performance of the method, with comparisons made to baseline algorithms such as FedAvg and FedProx. The results demonstrate that the proposed algorithm exhibits enhanced robustness and yields more accurate parameter estimates.

Keywords: Agents-based systems, Nonlinear systems
Abstract: We propose a continuous-time nonlinear model of opinion dynamics with utility-maximizing agents connected via a social influence network. A distinguishing feature of the proposed model is the inclusion of an opinion-dependent resource-penalty term in the utilities, which limits the agents from holding opinions of large magnitude. This model is applicable in scenarios where the opinions pertain to the usage of resources, such as money, time, computational resources, etc. Each agent myopically seeks to maximize its utility by revising its opinion in the gradient ascent direction of its utility function, thus leading to the proposed opinion dynamics. We show that for any arbitrary social influence network, opinions are ultimately bounded. For networks with weak antagonistic relations, we show that there exists a globally exponentially stable equilibrium using contraction theory. We establish conditions for the existence of consensus equilibrium and analyze the relative dominance of the agents at consensus. We also conduct a game-theoretic analysis of the underlying opinion formation game, including on Nash equilibria and on prices of anarchy in terms of satisfaction ratios. In addition, we also investigate the oscillatory behavior of opinions in a two-agent scenario. Finally, simulations illustrate our findings.

Keywords: Decentralized control, Optimal control, Control applications
Abstract: We consider a susceptible-infected-susceptible (SIS) epidemic model in which a large group of individuals decide whether to adopt partially effective protection without being aware of their individual infection status. Each individual receives a signal which conveys noisy information about its infection state, and then decides its action to maximize its expected utility computed using its posterior probability of being infected conditioned on the received signal. We first derive the static signal which minimizes the infection level at the stationary Nash equilibrium under suitable assumptions. We then formulate an optimal control problem to determine the optimal dynamic signal that minimizes the aggregate infection level along the solution trajectory. We compare the performance of the dynamic signaling scheme with the optimal static signaling scheme, and illustrate the advantage of the former through numerical simulations.

Keywords: Machine learning, Agents-based systems, Stochastic systems
Abstract: We consider a nonlinear version of the Kolmogorov forward equation in probability theory, which is shown to subsume many popular dynamic models in applied probability and stochastic algorithms. These range from urn models to mean field dynamics. We argue that this is a convenient template for learning algorithms based on `stateless' observations (i.e., without an underlying `state' process given by a dynamical system), particularly for multi-agent learning. We list several known instances of it, along with some accompanying analysis. end{abstract}

Keywords: Autonomous systems, Control applications, Optimal control
Abstract: In this paper, we present a dynamic game-theoretic framework for analyzing multi-agent interactions in military and defense applications. Specifically, we study a variation of the active target-attacker-defender differential game involving a target, an attacker, and multiple defenders. The framework enables defenders to autonomously switch operational modes—from rescuers (rendezvousing with the target) to interceptors (intercepting the attacker), and vice versa. Using the Games of a Degree approach, we model the interaction within each mode as a linear-quadratic differential game and derive feedback Nash equilibrium strategies. We then adopt a receding horizon approach to enable mode switching and compute the agents’ switching strategies. The framework also accommodates scenarios in which the number of agents changes over the course of the game. We illustrate our results using agents modeled as differential-drive mobile robots.

Keywords: Estimation, Control education, Autonomous systems
Abstract: Accurate alignment is critical for stable physical contact operations such as wall perching or inspection using quadrotors. Misalignment at the moment of contact can induce rotation about the contact point, often leading to perching failure or damage. To address this, we present a minimal-sensing control framework that uses a single forward-facing range sensor to estimate the wall angle in real time via a yaw sweep maneuver. The estimated angle is then used to regulate the quadrotor's yaw and approach velocity, ensuring perpendicular alignment before contact. The method is implemented on a micro quadrotor and validated through docking experiments on vertical surfaces. Results show a significant improvement in the estimation of yaw angle and stable contact with the vertical surface. The proposed method achieves over 90% success rate and maintains estimation errors within ±2° compared to open-loop alignment. This work demonstrates that combining simple sensing with active control enables robust physical interaction in unstructured environments.

Keywords: Fault detection/accomodation, Supervisory control, Pattern recognition and classification
Abstract: Leakage in urban Water Distribution Networks (WDNs) contributes significantly to Non-Revenue Water (NRW), leading to economic losses, infrastructure degradation, and service disruption. Accurate and timely leak detection is essential for sustainable water management and efficient utility operation. In this study, we propose a hybrid meta model framework combining a linear Auto-Regressive with eXogenous input (ARX) model, a nonlinear residual learning module using Multi-Layer Perceptrons (MLP), and a meta-learning layer to enhance flow prediction and anomaly detection or leak detection. The proposed methodology is validated on the SmartWaterDistributionNetwork(SWaDNet) testbed, an experimental water distribution setup replicating real-world dynamics. While the ARX model performs well on open-loop data, its prediction accuracy degrades under closed-loop conditions. The hybrid framework mitigates this limitation, achieving R^2 values exceeding 0.9 across flow channels. For leak detection, residuals from the meta-model are analyzed using a robust Median Absolute Deviation (MAD) measure. Controlled leak events were successfully detected with high temporal accuracy, and very less false positives, and an average delay less than a minute. The results demonstrate the effectiveness of data-driven hybrid model in capturing nonlinear behaviors and enabling real time abnormalities in WDNs.

Keywords: Estimation, Filtering, Control applications
Abstract: Accurate state estimation in large-scale lithium-ion battery packs is critical for ensuring safety, longevity, and optimal performance, particularly in electric vehicle applications. However, most existing estimators rely on simplified lumped models that overlook the complexities arising from cell-to-cell variations and interconnection effects in parallel-connected packs. This paper addresses these challenges by proposing a generalized equivalent circuit model formulated as a nonlinear discrete-time descriptor system. The model explicitly incorporates contact resistances to capture the uneven current distribution among cells. To enable state estimation under a reduced sensing scenario, where only pack-level voltage and current measurements are available, we propose an extended Kalman filter based on the maximum correntropy criterion. This enhancement improves robustness to non-Gaussian sensor noise commonly encountered in practice. The proposed filter accurately estimates both differential states (cell state-of-charges) and algebraic states (local cell currents), making it a promising solution for cell-level monitoring in parallel-connected lithium-ion battery packs under realistic operating conditions.

Keywords: Kalman filtering, Estimation, Spacecraft control
Abstract: Autonomous turret systems are crucial to modern defence and security technology, with target tracking and attack capability needed. Turret mechanisms are subject to noise interference, high-speed targets, and interference, causing loss of accuracy and low reaction time. This paper presents a hybrid approach combining Savitzky-Golay smoothing algorithms with Kalman filtering for tracking stability improvement and attacking accuracy enhancement in autonomous turrets. The framework combines object detection via the light YOLOv3-tiny model, motion prediction via the Kalman filter, and trajectory smoothing via Savitzky-Golay filters. Extensive 2D and 3D space simulations are found to achieve high tracking precision for various target motions with up to 99.5% accuracy under general conditions and robust performance with excessive noise levels.

Keywords: Kalman filtering, Control of communication networks, Multivehicle systems
Abstract: The sim-to-real gap is a crucial challenge that exists in the literature of multi-agent robotic systems. Our work is motivated to increase the communication reliability among the agents sharing critical information. To address it, this paper discusses the handling of communication uncertainties (i.e., noises, delays, etc.) via the well-established Kalman filter aided with the data-driven Koopman operator. We further propose a modified Koopman-Kalman filter aided with the information of communication fault to carefully handle the delayed communication data. To show the effectiveness of the proposed method, a case study of two Mecanum-wheeled robots coordinating with minimal communication is considered.

Keywords: Optimal control, Optimization, Optimization algorithms
Abstract: Recent successful lunar landings have renewed interest in developing next-generation lunar landers capable of achieving pinpoint landing accuracy. The confirmed presence of water ice near the Moon’s south pole has further motivated multiple space agencies to plan missions targeting this scientifically and strategically important region. However, the south polar environment poses significant challenges due to extremely low sun elevation angles and hazardous terrain, resulting in limited or no visual navigation support during the primary deceleration phase. This leads to substantial divert requirements and increased navigation uncertainty in both vertical (altitude) and horizontal dimensions near the landing site. Additionally, to ensure continuous surface tracking and real-time hazard awareness, it is essential to maintain lander attitude within altimeter operational limits throughout the terminal descent. This constraint imposes further requirements on the lander’s permissible attitude maneuvers, particularly during large lateral divert actions to cater for navigation (horizontal) errors. In this paper, we propose a near-fuel-optimal terminal descent strategy coupled with analytical guidance formulation with state and control constraint to allow precise landings with high divert capability. The terminal descent is divided into two stages: (i) a minimum vertical thrust phase, during which the lander leverages lunar gravity to increase vertical velocity, thereby mitigating

Keywords: Spacecraft control, Stability of nonlinear systems, Aerospace
Abstract: Electric propulsion is used to maximize payload capacity in communication satellites. These orbit raising maneuvers span several months and hundreds of revolutions, making trajectory design a complex challenge. The literature typically addresses this problem using feedback laws, with Q-law being one of the most prominent approaches. However, Q-law suffers from closed-loop stability issues, limiting its suitability for real-time on-board implementation. In this work, we focus on closed-loop orbit raising rather than offline trajectory planning and address the stability limitations of the Q-law through a Lyapunov based control design. A Lyapunov-guided modification of the classical Q-law is proposed to ensure closed-loop stability and enable real-time implementation. The effectiveness of the proposed method is demonstrated through closed-loop orbit transfers across various scenarios, including co-planar transfers, equatorial to polar orbit transfers, and geostationary transfer orbit (GTO) to geostationary earth orbit (GEO) transfers.

Keywords: Aerospace, Autonomous systems, Multivehicle systems
Abstract: This work addresses the problem of trajectory planning for a receiver aircraft executing a rendezvous with a tanker aircraft in a three-dimensional airspace. To ensure a wake-free rendezvous, two altitude-separated influence zones are defined around the tanker, encapsulating the geometric constraints governing the approach and terminal phases of the receiver’s trajectory. A polynomial-based guidance law is developed to regulate the relative line-of-sight angle between the aircraft as a function of their relative separation, ensuring compliance with the constraints imposed by the defined influence regions. The receiver’s commanded speed profile is formulated using a polynomial function dependent on the inter-aircraft range and the instantaneous speed of the tanker. The effectiveness of the proposed approach is demonstrated through numerical simulations, including validation against a high-fidelity six-degree-of-freedom nonlinear autopilot model representative of a modern high-performance fighter aircraft. The proposed guidance framework offers a computationally efficient and implementable solution with enhanced spatial adaptability for rendezvous operations.

Keywords: Aerospace, Control applications, Nonlinear systems
Abstract: This paper studies the problem of target circumnavigation by an Unmanned Aerial Vehicle (UAV) while confining its trajectories within a specified region. Our solution approach relies on controlling the relative side-bearing angle of the UAV and the radial distance to the target to fulfill these control objectives. Under some mild assumption on initial states, we propose the lateral acceleration command applied on the UAV by relying on a Barrier Lyapunov Function (BLF)-based formulation. Using Lyapunov stability theory, we prove that, under the proposed acceleration command, UAV follows the desired circumnavigation path and its trajectories remain bounded within the specified region. We also characterize the feasible initial conditions and present simulations to illustrate the proposed strategy.

Keywords: Aerospace, Autonomous systems, Control applications
Abstract: This paper addresses the problem of guiding an interceptor to intercept a stationary target at a prescribed impact angle while respecting input constraints arising from actuator limitations. Although classical guidance laws ensure target interception, imposing a specific impact angle and input constraints can significantly improve the effectiveness of the guidance strategy. To this end, we propose a guidance strategy that employs nested saturation functions to ensure that the interceptor's lateral acceleration remains within bounded limits. The guidance law was developed for the case of a zero-degree impact angle, which can be extended to any arbitrary impact angle through a coordinate transformation. Numerical simulations under various initial conditions validate the effectiveness of the proposed approach.

Keywords: Aerospace, Control applications, Autonomous systems
Abstract: This paper presents a nonlinear guidance strategy to neutralize a target, which is moving with a constant speed, at a prescribed impact time. The proposed approach leverages a trigonometric function-based formulation that ensures the convergence of relevant error variables to zero at a user-defined time (strictly earlier than the desired impact time). This convergence time can be preassigned independently of the initial engagement geometry, which is an alluring feature of the proposed solution. The guidance law is inspired from deviated pursuit guidance, known for its implementation simplicity. The effectiveness of the proposed strategy is demonstrated through numerical simulations across a diverse range of engagement scenarios.

Keywords: Optimization algorithms, Machine learning, Computational methods
Abstract: In this paper, we present a unified algorithm for stochastic optimization that makes use of a ``momentum'' term; in other words, the stochastic gradient depends not only on the current true gradient of the objective function, but also on the true gradient at the previous iteration. Our formulation includes the Stochastic Heavy Ball (SHB) and the Stochastic Nesterov Accelerated Gradient (SNAG) algorithms as special cases. In addition, in our formulation, the momentum term is allowed to vary as a function of time (i.e., the iteration counter), and both the momentum term and the step size can be random. The assumptions on the stochastic gradient are the most general in the literature, in that it can be biased, and have a conditional variance that grows in an unbounded fashion as a function of time. This last feature is crucial in order to make the theory applicable to ``zero-order'' methods, where the gradient is estimated using just two function evaluations.

We present a set of sufficient conditions for the convergence of the unified algorithm. These conditions are natural generalizations of the familiar Robbins-Monro and Kiefer-Wolfowitz-Blum conditions for standard stochastic gradient descent. We also analyze another method from the literature for the SHB algorithm with a time-varying momentum parameter, and show that it is impractical.

Keywords: Optimal control, Neural networks, Computational methods
Abstract: We study the application of a physics-informed neural network (PINN) to solve the two-point boundary value problem (BVP) arising from the necessary conditions postulated by Pontryagin's Minimum Principle for optimal control. Such BVPs are known to be numerically difficult to solve by traditional shooting methods due to extremely high sensitivity to initial guesses. In the light of recent successes in applying PINNs for solving high-dimensional differential equations, we develop a PINN to solve the problem of finding trajectories with minimum exposure to a spatiotemporal threat for a vehicle kinematic model. First, we implement PINNs that are trained to solve the BVP for a given pair of initial and final states for a given threat field. Next, we implement a PINN conditioned on the initial state for a given threat field, which eliminates the need for retraining for each initial state. We demonstrate that the PINN outputs satisfy the necessary conditions with low numerical error.

Keywords: Optimization, Optimization algorithms, Multivehicle systems
Abstract: We present a graph-theoretic framework for static rake-link design in timetabled suburban railway networks. By constructing a Link Feasibility DAG over scheduled services and solving a Minimum Path Cover via bipartite matching, our method computes the minimal fleet size in polynomial time. We further explore key operational trade-offs—deadhead distance, platform slack, and load balance—through a multi-objective parametric study on real-world data, revealing dominating solutions and insight into decision-space structure. The resulting tool offers timetable planners an efficient, flexible offline design routine that is independent of timetable planning. Finally, we outline a roadmap toward an online dynamic linking controller based on Model Predictive Control, which will leverage the current approach with a receding-horizon feedback loop to handle delays, insertions, and resource failures in live operations.

Keywords: Optimal control, Neural networks, Machine learning
Abstract: This paper presents a physics-informed machine learning approach for synthesizing optimal feedback control policy for infinite-horizon optimal control problems by solving the Hamilton–Jacobi–Bellman (HJB) partial differential equation(PDE). The optimal control policy is derived analytically for affine dynamical systems with separable and strictly convex control costs, expressed as a function of the gradient of the value function. The resulting HJB-PDE is then solved by approximating the value function using the Extreme Theory of Functional Connections (X-TFC)—a hybrid approach that combines the Theory of Functional Connections (TFC) with the Extreme Learning Machine (ELM) algorithm. This approach ensures analytical satisfaction of boundary conditions and significantly reduces training cost compared to traditional Physics-Informed Neural Networks (PINNs). We benchmark the method on linear and non-linear systems with known analytical solutions as well as demonstrate its effectiveness on control tasks such as spacecraft optimal de-tumbling control.

Keywords: Optimal control, Optimization, Numerical algorithms
Abstract: India achieved a significant milestone on August 23^{text{rd}} 2023, becoming the fourth country to accomplish a soft landing on the Moon. This paper presents the powered descent trajectory design for the Chandrayaan-3 mission. The optimization framework is based on pseudospectral Radau collocation, and controllability-based waypoint refinement is employed to further enhance the robustness of the trajectory against state and control perturbations. Furthermore, the trade-off between fuel consumption and robustness is explicitly quantified, providing insights into the practical considerations of mission planning.

Keywords: Optimization, Optimization algorithms, Numerical algorithms
Abstract: In this work, we propose a framework to allocate hubs and determine multi-hop routes in an intermodal freight transportation network that is relevant to the Indian Railways. In contrast to the existing methods that assume a small discrete set of candidate hub locations beforehand and select a subset of these as hubs in the transportation network, we develop a framework that allocates hubs in a continuous domain. This extends the space of feasible solutions; thus, paving the way for a more optimized network. We provide a stage-wise viewpoint of the paths in the transportation network and use the Law of Optimality to significantly reduce the number of decision variables required to determine the optimal paths, which results in the second highlight of the proposed framework - computational scalability. The objective function of the resulting optimization problem is riddled with multiple poor local minima. Thus, we develop a Maximum Entropy Principle (MEP) based method to deal with this non-convexity, and design our algorithms to avoid poor local minima. We demonstrate the efficacy of our proposed framework on a commodity transport dataset that is relevant to the Indian Railways.

Keywords: Agents-based systems, Optimal control, Machine learning
Abstract: Q-learning and double Q-learning are well-known sample-based, off-policy reinforcement learning algorithms. However, Q-learning suffers from overestimation bias, while double Q-learning suffers from underestimation bias. To address these issues, this letter proposes a weighted smooth Q-learning (WSQL) algorithm. The proposed algorithm employs a weighted combination of the mellowmax operator and the log-sum-exp operator in place of the maximum operator. Firstly, a new stochastic approximation based result is derived and as a consequence the almost sure convergence of the proposed WSQL is presented. Further, a sufficient condition for the boundedness of WSQL algorithm is obtained. Numerical experiments are conducted on benchmark examples to validate the effectiveness of the proposed weighted smooth Q-learning algorithm.

Keywords: Stochastic systems, Optimization, Simulation
Abstract: We present a family of generalized simultaneous perturbation-based gradient search (GSPGS) estimators that use noisy function measurements. The number of function measurements required by each estimator is guided by the desired level of accuracy. We first present in detail unbalanced generalized simultaneous perturbation stochastic approximation estimators and later present the balanced versions of these. We extend this idea further and present the generalized smoothed functional and generalized random directions stochastic approximation estimators, respectively, as well as their balanced variants. We show that estimators within any specified class requiring more number of function measurements result in lower estimator bias. We present a detailed analysis of both the asymptotic and non-asymptotic convergence of the resulting stochastic approximation schemes. We further present a series of experimental results with the various GSPGS estimators on the Rastrigin and quadratic function objectives. Our experiments are seen to validate our theoretical findings.

Keywords: Agents-based systems, Learning, Optimal control
Abstract: In cooperative decentralized multi-agent reinforcement learning (MARL), the presence of even a single greedy adversarial agent can significantly disrupt the convergence towards optimality. This paper provides both theoretical insights and empirical evidence illustrating this phenomenon. Leveraging a variant of the ClippedGossip algorithm for consensus, we propose a novel approach to neutralize the disruptive influence of greedy adversaries. Through rigorous analysis, we establish the convergence of off-policy actor-critic decentralized MARL in environments containing non-cooperating agents. Experiments across diverse scenarios validate the efficacy of our approach, demonstrating its ability to maintain cooperation and achieve convergence even in the presence of adversarial behavior.

Keywords: Learning, Optimal control, Autonomous systems
Abstract: Maximum-entropy reinforcement learning (MaxEnt RL) methods designed for continuous action spaces generally employ actor-critic framework, which involves alternating between policy evaluation and policy improvement stages. The normalized advantage function (NAF) simplifies continuous control by modeling a quadratic action-value function, but its deterministic policy limits exploration capabilities. In this paper, a MaxEnt RL framework, termed as soft normalized advantage function (SNAF), is introduced. SNAF extends NAF by incorporating a stochastic policy, thereby enhancing exploration and mitigating the risk of converging to local optima without requiring additional sampling approximations. The performance of SNAF is evaluated on benchmark optimal control problems, which encompass pendulum stabilization, the Van der Pol oscillator, and a target control task, comparing its performance with soft actor-critic and the original NAF. Experimental results demonstrate that SNAF achieves improved sample efficiency, accelerated convergence, and enhanced robustness while maintaining computational simplicity. These findings underscore its potential for addressing high-dimensional continuous control tasks.

Keywords: Optimal control, Iterative learning control, Optimization
Abstract: In a recent work, we proposed Reliable Policy Iteration (RPI), that restores policy iteration's monotonicity-of-value-estimates property to the function approximation setting. Here, we assess the robustness of RPI's empirical performance on two classical control tasks---CartPole and Inverted Pendulum---under changes to neural network and environmental parameters. Relative to DQN, Double DQN, DDPG, TD3, and PPO, RPI reaches near-optimal performance early and sustains this policy as training proceeds. Because deep RL methods are often hampered by sample inefficiency, training instability, and hyperparameter sensitivity, our results highlight RPI’s promise as a more reliable alternative.

Keywords: Optimal control, Stochastic systems, Delay systems
Abstract: This paper investigates a general class of two player non-zero-sum mean-field stochastic differential game under partial information, incorporating both discrete and distributed delays. The state dynamics are described by mean-field stochastic differential equations, where the state and control processes of both players are subject to delays. Utilizing the Hamiltonian and the generalized anticipated mean-field backward stochastic differential equations, we derive a necessary condition for optimality. Furthermore, under suitable concavity assumptions, a sufficient verification theorem is established. Finally, as an application of the theoretical framework, we examine a portfolio optimization problem accounting for delayed market responses.

Keywords: Air traffic management, Numerical algorithms, Computational methods
Abstract: This paper addresses a mine dispensing problem, wherein multiple Unmanned Aerial Vehicles (UAVs) transport mines from a base to a set of desired target locations. Each UAV operates on a shuttle basis, returning to the base after each delivery to collect the next mine. The primary objective is to minimize the total mission time. Since this problem involves allocation, which is NP-hard, we propose a heuristic algorithm, Farther Target First (FTF), to efficiently generate feasible solutions. To benchmark the quality of the heuristic solutions, we also formulate the problem as a binary integer program and solve it using standard optimization tools. Further to allocation, we introduce a scheduling algorithm to regulate safe UAV movements during mine dispensing. The algorithm adopts an Air Traffic Control-like approach, dividing the battlefield into sectors and altitude layers or corridors. The scheduling algorithm ensures that no two UAVs occupy the same altitude corridor within the same sector concurrently. Together with the FTF allocation heuristic, this yields a two-stage solution, where we ‘first allocate and then schedule’. Finally, we give an integrated allocation and scheduling algorithm to solve the mine dispensing problem. Comprehensive simulations demonstrate that the integrated approach consistently outperforms the two-stage method across various mission scenarios.

Keywords: Autonomous systems, Control applications
Abstract: This paper addresses the problem of converging to and maintaining a desired formation of a group of multiple UAVs and their coordinated reference path following in a constant altitude planar environment. A novel hybrid guidance strategy is proposed that integrates a virtual structure framework with vector field-based path following, ensuring that the UAVs achieve and maintain a desired formation w.r.t. the virtual leader's trajectory, while the constant speed virtual leader is guided by a vector field-based approach from existing literature to follow the reference path asymptotically. The constant speed of the virtual leader provides a reference state for the UAVs, which is unaffected by sensor noise or communication delays. Each follower UAV control its heading and speed using a combination of vector field and Lyapunov-based guidance method, guaranteeing finite-time convergence of heading errors and asymptotic convergence of formation errors. A strategy for collision-free motion of the agents is devised from geometric conditions. Theoretical analysis and numerical simulation are presented to validate the effectiveness of the proposed strategy highlighting its ability to attain highly precise formation path following and safe inter-UAV separation.

Keywords: Networked control systems, Stability of nonlinear systems, Robust control
Abstract: This paper proposes a methodology that ensures prescribed-time resilient consensus in a heterogeneous multi-agent system (HMAS). The methodology has two components: (a) a reconfiguration protocol that ensures robust connectivity and (b) a prescribed time control design that achieves the desired formation. The reconfiguration protocol is based on a weighted mean subsequence reduced algorithm (W-MSR),and the prescribed time control is designed using the Barrier Lyapunov function for flocking and tracking the desired formation even in the presence of non-cooperative agents.The simulation experiments validate the effectiveness of the proposed methodology, which can be used in further detail for real-world applications.

Keywords: Networked control systems, Autonomous systems, Control applications
Abstract: This paper proposes an event-triggered polynomial model predictive control method for collision-free point-to-point multi-agent navigation. In this control method, each control input to each agent is a polynomial whose coefficients are updated in an event-triggered manner. For each agent, we design an event-triggering rule that guarantees non-Zeno behavior of inter-event times. At each event, the controller updates the coefficients of the polynomial control law corresponding to a subset of agents by solving one or more finite horizon optimization problems. We also ensure feasibility of the optimization problems solved at each event. Through numerical simulations, we illustrate the results and compare the proposed method with other existing methods.

Keywords: Autonomous systems, Control applications, Flight control
Abstract: This paper presents a differential game-based framework for defending a stationary target in three- dimensional (3D) space using autonomous unmanned aerial vehicles (UAVs). The scenario involves a single attacker attempting to reach the target while multiple defenders coordinate to intercept it. Each agent is modeled us- ing position-level single-integrator dynamics for high-level control design. The interaction is cast as a continuous- time zero-sum game with a quadratic cost functional, where optimal strategies are derived using a time-varying Riccati differential equation. To ensure safe coordination during close-proximity operations, collision avoidance is incorporated via repulsion-based augmentation of defender controls. Furthermore, the high-level control inputs are tracked using a low-level proportional-derivative (PD) controller within a realistic 6-DOF UAV dynamic model implemented in Simulink. Simulation results validate the effectiveness of the proposed framework in achieving successful interception while maintaining defender safety and realistic trajectory tracking.

Keywords: Agents-based systems, Intelligent systems, Mechanical systems/robotics
Abstract: Coordinating multiple robots in decentralized path planning scenarios with only local sensing and communication requires careful information exchange. In this connection, learning-based approaches using Graph Neural Networks (GNN) or Graph Convolution Networks (GCN) are promising proponents, but they rely on predefined, fixed communication ranges, also called K-hops. Consequently, this may not be optimal in diverse situations. Effectively, this limitation is addressed in this paper by incorporating a mechanism called Adaptive Diffusion Convolution (ADC) into an established decentralized Multi-Robot Path Planning (MRPP) framework based on diffusion generative models, which are trained with imitation learning. ADC replaces the fixed-hop GNN message passing with a diffusion process whose effective range is automatically learned during the training process. Adaptive control of the information diffusion radius helps the system perform like an expert planner, thereby showing the benefits of adaptive neighbourhood learning in complex multi-robot coordination. Extensive simulation results across varying situations and a case study have corroborated the effectiveness of the proposed methods of the work.

Keywords: Reduced order modeling, Modeling and simulation
Abstract: We study the model-reduction problem by moment matching for linear and nonlinear systems in a data-driven setting. We show that reduced-order models can be directly computed from input-output data without requiring the knowledge of the structure of the signal generator or its internal state. The reduced-order models thus obtained match the moments of the unknown underlying system asymptotically. Our formulation provides a simple way to enforce additional constraints on the structure of the reduced-order model, which could be used to incorporate prior knowledge about the underlying system. In addition, we show that our method can be directly applied to a large class of linear and nonlinear time-delay systems with minimal modifications. Finally, we provide a simple algorithmic formulation that can be used directly with data and demonstrate its effectiveness on a benchmark example---a nonlinear RC ladder circuit.

Keywords: Modeling and simulation, Identification, Nonlinear systems
Abstract: Dynamic Mode Decomposition with control (DMDc) is a popular data-driven system identification approach for obtaining plant models of systems with control input exhibiting nonlinear dynamics. Hankel DMDc is an extension of DMDc, which is used for systems having limited outputs and delayed responses. The standard Hankel DMDc algorithm uses the least squares approach in estimating the system matrices. This approach cannot handle different noise levels in the output and input data. In this article, we propose a modified algorithm that computes the system matrices via noise-normalized total least squares within the Hankel DMDc framework. The efficacy of the proposed approach is numerically evaluated using simulated data of a forced van der Pol oscillator, wherein it is shown to perform better than the traditional Hankel DMDc approach.

Keywords: Estimation, Kalman filtering, Sensor fusion
Abstract: Sensor selection is an often under-appreciated aspect of state estimator or Kalman filter design. The basic minimum requirement for the choice of a sensor set while designing Kalman filters is that all states are observable. In addition, the sensors should be chosen with a view towards estimating the states with a desired accuracy. Often observability is treated as true/false check during filter design. Beyond observability, the observability degree, which measures how observable the states are, has been used as the metric of choice to for sensor selection or placement applications. The higher the degree of observability, the better the possibility of designing Kalman filters that achieve the desired state estimation accuracy and consistency requirements. When a wide variety of sensors are available, sometimes with cost and physical constraints involved, sensor selection plays a crucial role in filter design. In such situations it is important to know the expected contribution of each sensor towards observability degree. Shapley values, developed in cooperative game theory for fair allocation of the payout of a multi-player game to individual players, are widely used in machine learning to assess feature importance. This paper shows that Shapley values can indeed be leveraged to quantify the expected marginal contribution of each sensor in any given sensor set towards the observability degree. This has potential applications in filter design and sensor selection/placement.

Keywords: Estimation, Identification, Optimization
Abstract: Learning the edge connectivity structure of networked systems from limited data is a fundamental challenge in many critical infrastructure domains, including power, traffic, and finance. Such systems obey steady-state conservation laws: x = L*y, where x and y represent injected flows (inputs) and potentials (outputs), respectively. The sparsity pattern of the pxp Laplacian L* encodes the underlying edge structure. In a stochastic setting, the goal is to infer this sparsity pattern from zero-mean i.i.d. samples of y.

Recent work by Rayas et al. [1] has established statistical consistency results for this learning problem by considering an l1-regularized maximum likelihood estimator. However, their approach did not develop a scalable algorithm but relies on solving a convex program via the CVX package. To address this gap, we propose an alternating direction method of multipliers (ADMM), which is transparent and fast. A key contribution is to demonstrate the role of an algebraic matrix Riccati equation in the primal update step of ADMM. Numerical experiments on a host of synthetic and benchmark networks, including power and water systems, show the efficiency of our method.

Keywords: Quantum control, Estimation, Identification
Abstract: This paper presents an adaptive observer-based framework for the simultaneous state and parameter estimation of a closed two-level quantum system governed by the Liouville–von Neumann equation. The system, driven by an external laser field with unknown atom-field coupling strength, is reformulated in the Bloch sphere representation, wherein the nonlinear dynamics of the spin variables are described. To estimate the unmeasured internal states and the unknown coupling parameter, we propose a cascaded strategy combining a nonlinear adaptive observer with the Dynamic Regressor Extension and Mixing (DREM) technique. The observer reconstructs the full Bloch vector from partial measurement (population difference), and the resulting regressors enable a scalar DREM formulation for the unknown parameter. Unlike traditional gradient-based identification, the DREM-based update ensures improved noise robustness and accelerated convergence even for scalar parameters. Numerical simulations confirm the accuracy and robustness of the proposed estimation strategy, demonstrating effective reconstruction of both the quantum state and the Hamiltonian parameter under realistic noisy measurements.

Keywords: Observers for nonlinear systems, LMIs, Stability of nonlinear systems
Abstract: A novel observer proposal, whose design conditions are in the form of linear matrix inequalities, is developed in this work for a class of nonlinear descriptor systems. It is shown that a suitable rearrangement of the system expressions can overcome former difficulties on the factorization of the error signals on the left-hand side of equations, at the price of turning the problem into an H-infinity-attenuation-level minimization. Examples of practical interest are shown that illustrate the effectiveness of the novel scheme.

Keywords: Robust adaptive control, Vision-based control, Nonlinear systems
Abstract: Most existing approaches that deploy the super-twisting control (STC) policy for regulating a first-order sliding mode (FOSM) rely on assuming the perturbation and its derivative are bounded by a known constant, and/or achieve exact finite-time convergence in the constraint-free case. In contrast, this brief considers the development of a novel adaptive STC policy that guarantees exact finite-time convergence of the FOSM subjected to a state-dependent uncertainty arising from the presence of a time-varying, asymmetric output constraint. This is achieved by first initiating a smooth invertible nonlinear transformation that explicitly accounts for the output constraint, which is followed by the design of a novel adaptive STC policy that guarantees exact finite-time convergence by explicitly accounting for the state-dependent structure of the perturbation term of the resulting unconstrained error system. Formation tracking experiments under visibility constraints, including performance comparisons with leading alternative designs, are used to verify the efficacy and demonstrate the advantages of the proposed scheme.

Keywords: Control applications, Constrained control, Nonholonomic systems
Abstract: This paper introduces a novel Barrier Lyapunov Guidance Vector Field (BLGVF)-based path following controller for a unicycle robot that integrates the Barrier Lyapunov function (BLF) in the vector field formulation to ensure that its trajectory converges to a desired path while remaining strictly confined within a predefined region. While conventional guiding vector field (GVF)-based approaches primarily focus on asymptotic convergence to a desired path, these do not address the practical need for spatial motion constraints on a robotic system deployed in a regulated environment. Under a natural assumption on the robot's initial position and the scalar field characterizing the desired curve, we propose a BLGVF-based formulation that not only ensures asymptotic convergence to the desired curve but also ensures that its trajectories remain bounded within an annular region of interest at all times. Simulation results are provided to demonstrate the proposed approach, followed by a comparison with the conventional method from the existing literature.

Keywords: Algebraic/geometric methods, Nonlinear systems, Stability of nonlinear systems
Abstract: The use of Morse functions on the Lie group SO of rigid body rotations as Lyapunov functions for rigid body attitude control leads to almost globally asymptotically stable (AGAS) control laws. A Morse function on SO has four critical points. When used as a Lyapunov function, it leads to a globally continuous state-dependent feedback control scheme with the minimum as the AGAS equilibrium of the feedback system. It is known that Morse functions give the largest possible domain of convergence with state-dependent and globally continuous feedback control laws. In this work, we explore the use of time-varying Morse functions for rigid body attitude control. This strategy can be used to parametrically stabilize desired equilibria or destabilize undesired equilibria of the feedback system. The positive gains of this Morse function are each the sum of a constant and a sinusoidal function of time. Linearization of the dynamics around the equilibria of the feedback system leads to decoupled Mathieu oscillators. Careful selection of these control gains can give parametric stability of the equilibria while varying these gains can lead to interesting behavior like period-doubling bifurcations. Numerical simulations of the nonlinear feedback system with control gains selected according to the linearized analysis show that the trajectory for an initial state on the stable manifold of an unstable equilibrium of this feedback system can be successfully destabilized.

Keywords: Stability of nonlinear systems, Switched systems, Large scale systems
Abstract: This paper deals with the problem of incremental stability in interconnected systems, specifically focusing on nonlinear switched systems connected in cascade. Guaranteeing stability in an overall interconnected nonlinear system is challenging, even if the individual subsystems are stable independently. In this context, the paper establishes a set of sufficient conditions that ensures incremental stability in the interconnection. To achieve this, the contraction theory as a tool is utilised to achieve incremental convergence, which ensures a formal guarantee. The paper uses computationally feasible matrix-measure-based conditions. Two case studies are presented to validate the proposed results.

Keywords: Stability of nonlinear systems, Feedback linearization, Nonlinear systems
Abstract: Incremental stability is a property of dynamical systems that ensures the convergence of trajectories with respect to each other rather than a fixed equilibrium point or a fixed trajectory. In this paper, we introduce a related stability notion called incremental input-to-state practical stability (delta-ISpS), ensuring safety guarantees. We also present a feedback linearization based control design scheme that renders a partially unknown system incrementally input-to-state practically stable and safe with formal guarantees. To deal with the unknown dynamics, we utilize Gaussian process regression to approximate the model. Finally, we implement the controller synthesized by the proposed scheme on a manipulator example.

Keywords: Constrained control, Uncertain systems, Robust control
Abstract: The article addresses the problem of tracking control for an unknown nonlinear system with time-varying bounded disturbance subjected to prescribed performance and input constraints (PIC). Since the simultaneous prescription of PIC involves a tradeoff, we propose an analytical feasibility condition to prescribe a feasible PIC, which also results in a feasible initial state set as a corollary. In addition, we propose an approximation-free controller to ensure the tracking performance meets the prescribed PIC. The effectiveness of the proposed approach is demonstrated through numerical examples.

Statement on Differences with Related Work: A preliminary version of this work was presented at the 2022 Indian Control Conference [1], which proposed a low-complexity approximation-free controller for scalar strict-feedback nonlinear systems under prescribed performance and input constraints. The current journal paper significantly extends this by generalizing the design and analysis to general n-dimensional strict-feedback systems, enabling broader applicability. In addition, this version introduces precise analytical feasibility conditions for simultaneously prescribing PPC and PIC, and derives a closed-form characterization of feasible initial state sets.

[1] P. K. Mishra and P. Jagtap, "On Controller Design for Unknown Nonlinear Systems with Prescribed Performance and Input Constraints," Proc. of the Eighth Indian Control Conference (ICC), 2022, pp. 212–217.

Keywords: Flight control, Robust control, Robust adaptive control
Abstract: The existing control design for the tilt quadrotor often assumes that the derivative of the disturbance is zero and that the moment of inertia (MoI) is diagonal. However, in general the disturbance may be bounded but not constant. Also, due to the tilting of the rotors, the MoI may not be a diagonal matrix. To address this, we design a robust controller which handles the MoI matrix uncertainties and is capable of rejecting varying external disturbances. This proposed controller guarantees convergence to a desired bounded region in the existence of bounded external disturbances. In this work, we develop the first small-sized, over-actuated tilt quadrotor using 3D-printed parts. Using the proposed controller, experiments are performed on both conventional and tilt quadrotors. From these experiments, the tilt quadrotor demonstrates better response time, flight time, and lower energy consumption compared to the conventional quadrotor. Finally, the proposed controller’s robustness is validated for both position and attitude setpoints of the tilt quadrotor when subjected to wind.

Keywords: Robust control
Abstract: Linear time-invariant (LTI) output feedback controllers are known to suffer from inherent robustness limitations when applied to plants with unstable poles and non-minimum phase (NMP) zeros. To address this, recently, a high-frequency continuous-time periodic output feedback control scheme was proposed for strictly proper plants. In this paper, we extend this result to single-input single-output (SISO) non-strictly proper plants and also show that the desirable robustness properties of the optimal full-state feedback loop are exactly emulated using periodic output feedback controller in an averaged sense. The closed-form expressions for controller gains, that ensure the above, are also obtained and found to be very computationally straightforward. Simulations show that the method is effective and results in insignificant ripples in the steady state although oscillations are still present in the transient phase.

Keywords: Robust control, Optimization algorithms, Linear systems
Abstract: In this paper we study the H_infinity-norm full-state feedback controller gain synthesis problem for linear time-invariant systems. We first analytically prove shortcoming in the gradient computation procedure using the H_infinity-norm algebraic Riccati equation (ARE). We then propose an approximate H_infinity-norm gradient which can be efficiently computed using the H_infinity-norm ARE. We then propose a scalable gradient-based controller gain synthesis procedure which uses the proposed approximate gradient. Finally, we compare our approach with existing methods and the MATLAB robust control toolbox.

Keywords: Robust adaptive control, Discrete event systems
Abstract: The primary concern of this research work is to design an adaptive super-twisting controller (ASTC) for uncertain nonlinear systems under resource-constrained scenario with a simplification in gain-tuning parameters. To address the resource-constrained issue, we introduce an event-triggering mechanism in the closed-loop setup, which transmits the output (sliding variable) information to the controller and then to actuator at irregular sampling intervals only. Here, the triggering rule is constructed with fewer triggering parameters compared to the existing event-based ASTC design approaches (cite{{9901685}, {li2020event}, {dou2024adaptive}}). Also, we have shown that the proposed approach reduces the practical sliding motion band size. Furthermore, the gain adaptation law is constructed here based on the sampled output information to determine the dynamic gain, which addresses the over-estimation problem in designing the control gain for unknown disturbance bound. Finally, an academic example is simulated to demonstrate the effectiveness of the proposed approach, and the comparison study is also included with the periodic sampling strategy.

Keywords: Robust control, PID control, Linear systems
Abstract: A novel graphical strategy to tune the decentralized PI controllers for a two-input two-output (TITO) system is proposed in this paper. The method starts with constructing the root crossing boundaries to map the entire region of stability in the controller parameter space. Robust performance and effective disturbance rejection are ensured by considering the phase margin and maximum sensitivity function. Steps for obtaining the curves representing the desired values of phase margin and maximum sensitivity are derived. Intersection of the obtained curves within the complete region of stability yields the required controller gains. Simulation results for two benchmark problems of TITO systems are given to evaluate the efficacy of the proposed control methodology.

Keywords: Mechanical systems/robotics, Robust adaptive control, Nonlinear systems
Abstract: Time domain restrictions on a dynamical system arise when tracking objectives to be accomplished in a finite time. The objective becomes challenging when state and input constraints are also present. Furthermore, a continuous control policy necessitates faster controller updates in a sampling-based actuation of a given system, leading to redundant use of communication and computational resources. In this regard, the contribution of this study is twofold: (i) an adaptive barrier-function-based control policy to achieve robust local prescribed time convergence while adhering to state and input constraints, (ii) a periodic event-triggering control (PETC) alongside the adaptive barrier control policy to achieve desired performance with fewer control updates. In particular, the proposed approaches use a time-based generator function to prespecify settling time embedded within a filtered tracking error, which confines the states within the state constraints using time-varying inequalities and limits the control action through a saturation function. Moreover, the PETC is designed to tackle the problem of continuously monitoring triggering conditions, while avoiding Zeno behaviour. Finally, an experimental study has been undertaken to demonstrate the efficacy of the proposed scheme.

Keywords: Cooperative control
Abstract: This work presents the experimental evaluation of multiple consensus protocols using Crazyflie drone swarm. Unlike prior studies limited to simulations, we implement and compare classic consensus, fractional-order consensus, double-integrator schemes, and formation control in real-world conditions. We analyze convergence speed, stability under disturbances, oscillatory behavior, and formation accuracy. The results reveal distinct trade-offs: fractional-order dynamics provide improved oscillation damping, double-integrator protocols enhance formation stability, and advanced formation control achieves precise dynamic coordination. By addressing noise, communication delays, and real-time execution, this study provides novel insights into adaptive and scalable swarm consensus, bridging theoretical models with practical deployment.

Keywords: Manufacturing systems
Abstract: This work presents a AI-enabled waste clas- sification system combining with affordable hardware to ensure automated, accurate, and cost-effective sorting of different types of solid waste. Hardware centers on a Raspberry Pi 4, utilizing a smartphone via DroidCam for wireless video input and dual servo motors for precise mechanical sorting. Employing YOLOv8x for object detec- tion and categorizing waste into seven types. This model achieves 89.2% mAP at 0.5 IoU, 91% precision, 87% recall, and an F1-score of 89%, with real-time inference at 5-7 FPS and 97% sorting accuracy in tests exceeding 500 items. This model combines high-performance AI with practical deployment on low-cost components, suitable for small- scale applications in developing regions.

Keywords: Flexible structures, Control education, Mechanical systems/robotics
Abstract: This work presents a bioinspired crab robot using single servo actuation per leg and simplified drag gait to enable robust locomotion. A 3D printed thermo- plastic polyurethane (TPU) frame and a dual controller setup for gait exceution and sensing. Field tests on even and uneven surfaces shows acceptable heading stability, displacement linearity (R2 > 0.9) and gait stability in- dex (GSI > 0.83). Experiments under windy outdoor conditions validates real-world application. The proposed design offers a low-cost, easy-to-maintain platform for lightweight robotic exploration.

Keywords: Mechanical systems/robotics
Abstract: Upper limb amputations, particularly below-elbow amputations are prevalent globally, especially in India, recording the highest number of cases. The increasing demand for low-cost prosthetics in the lower and middle-income countries is unfulfilled by the expensive and often low-functionality prosthetics. This study introduces GRASP (Gesture and Voice Responsive Smart Prosthetic), which is a human hand-like low-cost prosthetic hand that works on the voice commands of the user. The prosthetic hand is capable of lifting heavy weights of upto 2.9 Kg with a single finger. The fingers are made from composite material(TPU + PETG), which gives them a human-finger-like feel in motion as well as in appearance.

Keywords: Autonomous systems, Robust control
Abstract: Autonomous navigation in structured indoor environments demands accurate and lightweight localization methods, particularly when operating under constrained onboard resources. While Visual SLAM has become a standard solution, its performance on embedded hardware for scaled autonomous platforms remains underexplored. In this work, we implement ORB-SLAM3 on an NVIDIA Jetson Orin Nano within a 1/10th scale electric vehicle testbed (DEFT) and conduct controlled experiments on an indoor traffic track. To benchmark the estimated trajectories against both the ZED-mini stereo camera’s built-in odometry as ground truth. Results demonstrate that ORB-SLAM3 achieves competitive accuracy in linear trajectories but suffers from drift in curved maneuvers, highlighting its sensitivity to motion dynamics and scene structure. Beyond providing a quantitative baseline for future research, this study identifies key challenges in deploying V-SLAM for scaled autonomous platforms. It motivates extensions through visual-inertial fusion and lane-constrained SLAM.

Keywords: Agents-based systems
Abstract: Information design has gained in importance as sellers (data aggregators) are able to incentivize certain behaviors from competitive Buyers (firms) to increase social welfare or to sell this information to increase their profits. However, in both situations, optimal design is limited by the firms’ private information about their payoffs. To elicit such private information, sellers design mechanisms (e.g., auctions) whereby the firms are incentivized to both participate and to provide truthful information to the seller. This combined Information and Mechanism Design problem, which we refer to as Information Markets sits at the heart of many interesting applications involving smart infrastructures where information intermediaries regulate a physical layer through coordination.

The challenge in creating such an information marketplace stems from the very nature of information as an asset: (i) it can be replicated at zero marginal cost; (ii) it can be versioned with noise, (iii) its value to a firm is dependent on which other firms get access to such information (externality); and (iv) its value to a firm is heterogeneous.

We present a general framework involving N competing private firms (agents) and a monopolistic information seller. The asymmetry of information creates an opportunity to coordinate the actions of these firms in order to maximize social welfare. To illustrate the power of such formulations, we highlight applications in (i) electric vehicle charging, and ...

Keywords: Nonlinear systems, Mechanical systems/robotics, Aerospace
Abstract: Conjunction analysis is often used in space applications to assess collision risk between satellites or of satellites with debris. However, in a more general sense, conjunction analysis is applicable to guidance or path planning of autonomous vehicles where two or more vehicles move in space. Depending on the application, it would be desirable to predict the occurrence of collision, and subsequently devise control strategies to either achieve it, as in problems of docking or interception, or avoid it, as in mobile robotics. In this talk, we will describe the use of the notion of collision cones to carry on conjunction analysis in Euclidean spaces as well as its generalization on spherical manifolds. The applications covered will encompass problems ranging from robotics to astrodynamics.