Extended Abstract: Motion Planners Learned from Geometric Hallucination

TL;DR

This paper introduces a novel approach called hallucination to generate obstacle geometries that make a given motion plan optimal, enabling training of effective motion planners with less data in cluttered environments.

Contribution

It proposes two methods to hallucinate obstacle spaces that preserve plan optimality, facilitating data-efficient learning of motion planners for real-world robots.

Findings

Hallucination methods successfully generate obstacle geometries for plan optimization.

Trained motion planner performs well in real-world cluttered environments.

Approach reduces data requirements for learning collision-free navigation.

Abstract

Learning motion planners to move robot from one point to another within an obstacle-occupied space in a collision-free manner requires either an extensive amount of data or high-quality demonstrations. This requirement is caused by the fact that among the variety of maneuvers the robot can perform, it is difficult to find the single optimal plan without many trial-and-error or an expert who is already capable of doing so. However, given a plan performed in obstacle-free space, it is relatively easy to find an obstacle geometry, where this plan is optimal. We consider this "dual" problem of classical motion planning and name this process of finding appropriate obstacle geometry as hallucination. In this work, we present two different approaches to hallucinate (1) the most constrained and (2) a minimal obstacle space where a given plan executed during an exploration phase in a completely safe obstacle-free environment remains optimal. We then train an end-to-end motion planner that can produce motions to move through realistic obstacles during deployment. Both methods are tested on a physical mobile robot in real-world cluttered environments.