Simultaneous Navigation and Construction Benchmarking Environments

TL;DR

This paper introduces a benchmarking environment for simultaneous navigation and construction tasks in grid worlds, highlighting the challenges of coupling localization and environment manipulation without GPS, and evaluating existing methods.

Contribution

It presents a new benchmark for mobile construction tasks modeled as POMDPs, comparing handcrafted policies with deep RL methods to identify current limitations.

Findings

Coupling localization and manipulation makes the task very challenging for existing methods.

Deep RL methods struggle with the coupled navigation and construction problem.

The benchmark highlights the need for novel, task-specific solutions.

Abstract

We need intelligent robots for mobile construction, the process of navigating in an environment and modifying its structure according to a geometric design. In this task, a major robot vision and learning challenge is how to exactly achieve the design without GPS, due to the difficulty caused by the bi-directional coupling of accurate robot localization and navigation together with strategic environment manipulation. However, many existing robot vision and learning tasks such as visual navigation and robot manipulation address only one of these two coupled aspects. To stimulate the pursuit of a generic and adaptive solution, we reasonably simplify mobile construction as a partially observable Markov decision process (POMDP) in 1/2/3D grid worlds and benchmark the performance of a handcrafted policy with basic localization and planning, and state-of-the-art deep reinforcement learning (RL) methods. Our extensive experiments show that the coupling makes this problem very challenging for those methods, and emphasize the need for novel task-specific solutions.