LiCS: Navigation using Learned-imitation on Cluttered Space

TL;DR

LiCS is a robust, fast navigation system for indoor UGVs using learned imitation with Transformer networks, demonstrating superior performance and safety in cluttered environments through simulation and hardware tests.

Contribution

The paper introduces a novel navigation approach combining behavior cloning with Transformer neural networks for UGVs in cluttered spaces, enhancing robustness and speed.

Findings

Outperforms baseline algorithms in navigation performance

Maintains robustness in highly cluttered environments

Operates safely at speeds up to 1.5 m/s during hardware tests

Abstract

In this letter, we propose a robust and fast navigation system in a narrow indoor environment for UGV (Unmanned Ground Vehicle) using 2D LiDAR and odometry. We used behavior cloning with Transformer neural network to learn the optimization-based baseline algorithm. We inject Gaussian noise during expert demonstration to increase the robustness of learned policy. We evaluate the performance of LiCS using both simulation and hardware experiments. It outperforms all other baselines in terms of navigation performance and can maintain its robust performance even on highly cluttered environments. During the hardware experiments, LiCS can maintain safe navigation at maximum speed of $1.5\ m/s$.