Reactive Collision Avoidance for Fixed-Wing MAVs Flying in Urban Terrain
Rajnikant Sharma, Jeffery B. Saunders, Clark N. Taylor, Randal W. Beard · 2009
Flying flxed-wing miniature air vehicles (FW-MAVs) in urban terrain is a challenging task because of unstructured urban environments, the minimum speed requirement for ∞ight of flxed-wing vehicles, and the non-holonomic characteristics of FW-MAVs. While traditional path-planning algorithms can efiectively overcome many of these constraints, if the FW-MAV detects new obstacles in-∞ight that were not included in the initial pathplanning, immediate action must be taken to avoid collision. Re-planning with traditional path-planning algorithms is not feasible due to the time required to plan an entire path. Therefore, a reactive collision avoidance system must be used to guarantee obstacle avoidance until a new path can be planned which includes the newly sensed information. In this paper we develop a reactive nonlinear collision avoidance guidance law for FWMAVs ∞ying in urban terrain. Our guidance law is designed for the scenario where a wall or other large object is observed. In this case, we would like the FW-MAV to maintain a specifled safe distance away from the obstacle over time. The algorithm is developed using Lyapunov stability theory, relative geometry between the FW-MAV and the obstacle. The FW-MAV reacts given the range information between the FW-MAV and the obstacle at a constant bearing. We show that our guidance law guarantees the convergence of distance between the FW-MAV and the obstacle to the desired safe distance. We also derive a sliding mode controller to take into account the uncertainty in wall orientation. Simulation results show that the guidance law is successful in navigating the FW-MAV between zigzag walls, non parallel walls, and curved walls.