Functional Path Optimisation for Exploration in Continuous Occupancy Maps

TL;DR

This paper introduces a novel variational approach for autonomous exploration that directly optimizes continuous paths using functional gradients, leveraging Hilbert maps for efficient information gain computation, resulting in smooth, safe, and informative trajectories.

Contribution

It formulates exploration as a variational problem enabling direct trajectory optimization with functional gradients, utilizing Hilbert maps for efficient information-based path planning.

Findings

The method finds smooth, safe exploration paths.

It outperforms other exploration strategies in experiments.

The approach efficiently maximizes information gain along trajectories.

Abstract

Autonomous exploration is a complex task where the robot moves through an unknown environment with the goal of mapping it. The desired output of such a process is a sequence of paths that efficiently and safely minimise the uncertainty of the resulting map. However, optimising over the entire space of possible paths is computationally intractable. Therefore, most exploration methods relax the general problem by optimising a simpler one, for example finding the single next best view. In this work, we formulate exploration as a variational problem which allows us to directly optimise in the space of trajectories using functional gradient methods, searching for the Next Best Path (NBP). We take advantage of the recently introduced Hilbert maps to devise an information-based functional that can be computed in closed-form. The resulting trajectories are continuous and maximise safety as well as mutual information. In experiments we verify the ability of the proposed method to find smooth and safe paths and compare these results with other exploration methods.