OCROMay 30, 2019

Fast Reciprocal Collision Avoidance Under Measurement Uncertainty

arXiv:1905.12875v211 citations
Originality Incremental advance
AI Analysis

This addresses a practical challenge in robotics for teams operating in uncertain environments, though it is incremental as it builds on existing collision avoidance methods.

The paper tackles the problem of distributed collision avoidance for mobile robots with noisy on-board sensors and no communication, presenting a convex optimization-based algorithm that guarantees mutual collision avoidance under mild conditions, with numerical simulations showing all agents avoid collisions and reach goals in 2D and 3D despite uncertainty.

We present a fully distributed collision avoidance algorithm based on convex optimization for a team of mobile robots. This method addresses the practical case in which agents sense each other via measurements from noisy on-board sensors with no inter-agent communication. Under some mild conditions, we provide guarantees on mutual collision avoidance for a broad class of policies including the one presented. Additionally, we provide numerical examples of computational performance and show that, in both 2D and 3D simulations, all agents avoid each other and reach their desired goals in spite of their uncertainty about the locations of other agents.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes