Lapasi, Manikandan (Indian Institute of Technology Madras), Ghosh, Satadal (Indian Institute of Technology Madras)

Online Hybrid Motion Planning for Unmanned Aerial Vehicles in Planar Environments

Scheduled for presentation during the Regular Session "Airspace Management" (FrB4), Friday, September 4, 2020, 12:50−13:10, Naousa

2020 International Conference on Unmanned Aircraft Systems (ICUAS), September 1-4, 2020 (Postponed from June 9-12, 2020), Athens, Greece

This information is tentative and subject to change. Compiled on April 25, 2024

Abstract

Motion planning for unmanned aerial vehicles (UAVs) in an obstacle-cluttered environment for target interception/rendezvous has been a problem of growing importance. Several offline deterministic and sampling-based planners have been presented in literature to address this issue. However, online motion planners gained paramount importance due to their inherent feedback mechanism. Following this, several guidance algorithms and collision avoidance algorithms have been presented in literature. But, very few literature have merged them to come up with a holistic online motion planner. Even among them, realistic turn capabilities of UAVs have been largely ignored. Also, they rely on the near-collision-course geometry between the UAV and the target. To obviate these limitations, this paper presents an online hybrid motion planner for a constant-speed UAV in a 2-D obstacle-cluttered environment that combines a global planner with local planners for online collision avoidance with obstacles. Specifically, this paper investigates the performance of combinations of the most widely studied guidance, proportional navigation, as global planner with three seminal local planners for collision avoidance based on the potential field, velocity obstacle and collision cone concepts. Extensive simulation studies help elucidate the performance comparison of these three hybrid motion planning algorithms in terms of four crucial measures of effectiveness (MOEs) - target interception time, minimum obstacle distance, overall control effort and computational cycle time. Discussion about the effectiveness of these three online hybrid algorithms under different considerations of these MOEs exhibits their potential in real-time motion planning and control for real-life UAV applications.