Towards Learning Abstract Representations for Locomotion Planning in High-dimensional State Spaces

TL;DR

This paper introduces a CNN-based cost function for abstract representations in high-dimensional locomotion planning, significantly improving planning efficiency for ground robots navigating complex terrains.

Contribution

It proposes a novel CNN-based cost function trained on artificial data that generalizes to real scenes, enhancing search-based planning for hybrid locomotion.

Findings

CNN-based cost function accelerates planning by multiple orders of magnitude.

Abstract representation reduces the complexity of high-dimensional state spaces.

Method effectively generalizes from artificial to real-world environments.

Abstract

Ground robots which are able to navigate a variety of terrains are needed in many domains. One of the key aspects is the capability to adapt to the ground structure, which can be realized through movable body parts coming along with additional degrees of freedom (DoF). However, planning respective locomotion is challenging since suitable representations result in large state spaces. Employing an additional abstract representation---which is coarser, lower-dimensional, and semantically enriched---can support the planning. While a desired robot representation and action set of such an abstract representation can be easily defined, the cost function requires large tuning efforts. We propose a method to represent the cost function as a CNN. Training of the network is done on generated artificial data, while it generalizes well to the abstraction of real world scenes. We further apply our method to the problem of search-based planning of hybrid driving-stepping locomotion. The abstract representation is used as a powerful informed heuristic which accelerates planning by multiple orders of magnitude.