Probabilistic Collision Constraint for Motion Planning in Dynamic Environments

TL;DR

This paper introduces a probabilistic collision constraint for real-time motion planning in dynamic environments, accounting for uncertainties in robot and obstacle states to improve collision avoidance.

Contribution

It presents a novel tight upper bound for collision probability, integrated as a real-time solvable constraint in motion planning algorithms.

Findings

Successful collision avoidance in simulation for mobile robots and quadrotors.

Outperforms several state-of-the-art methods in dynamic environments.

Real-time applicability demonstrated through simulation results.

Abstract

Online generation of collision free trajectories is of prime importance for autonomous navigation. Dynamic environments, robot motion and sensing uncertainties adds further challenges to collision avoidance systems. This paper presents an approach for collision avoidance in dynamic environments, incorporating robot and obstacle state uncertainties. We derive a tight upper bound for collision probability between robot and obstacle and formulate it as a motion planning constraint which is solvable in real time. The proposed approach is tested in simulation considering mobile robots as well as quadrotors to demonstrate that successful collision avoidance is achieved in real time application. We also provide a comparison of our approach with several state-of-the-art methods.