Collision Avoidance

Ask about this topic

Overview

Collision avoidance studies how to detect potential impacts and choose safe actions for systems ranging from road vehicles and drones to robots, ships, and spacecraft. Common approaches include geometric methods such as velocity obstacles, model predictive control, control barrier functions, and reachability analysis, alongside reinforcement learning and learned perception. Work increasingly accounts for uncertainty in motion, sensing, communication, and hidden obstacles, and tests methods in dense multi-agent settings as well as real-world deployments. A central challenge is balancing strong safety guarantees with efficient, timely decisions that are not so conservative that they block useful motion.

Latest papers

CardsList

91 papers11 papers in the last four weeks, up 83% on the four weeks before. 0.1% of all new papers.

Mar 17, 2024

N-dimensional Convex Obstacle Avoidance using Hybrid Feedback Control (Extended version)

This paper addresses the autonomous robot navigation problem in a priori unknown n-dimensional environments containing disjoint convex obstacles of arbitrary shapes and sizes, with pairwise distances strictly greater than the robot's diameter. We propose a hybrid feedback control scheme that guarantees safe and global asymptotic convergence of the robot to a predefined target location. The proposed control strategy relies on a switching mechanism allowing the robot to operate either in the move-to-target mode or the obstacle-avoidance mode, based on its proximity to the obstacles and the availability of a clear straight path between the robot and the target. In the obstacle-avoidance mode, the robot is constrained to move within a two-dimensional plane that intersects the obstacle being avoided and the target, preventing it from retracing its path. The effectiveness of the proposed hybrid feedback controller is demonstrated through simulations in two-dimensional and three-dimensional environments.

Mayur Sawant, Ilia Polushin, Abdelhamid Tayebi

Department of Electrical and Computer Engineering, Lakehead University, Thunder Bay, ON P7B 5E1, Canada · Department of Electrical and Computer Engineering, Western University, London, ON N6A 3K7, Canada

Collision AvoidanceSafe Robot NavigationFeedback ControlRobot Collision Avoidance
AskPDF