Agile Autonomous Driving using End-to-End Deep Imitation Learning

TL;DR

This paper introduces an end-to-end deep imitation learning system for agile off-road autonomous driving that uses low-cost sensors and outperforms existing methods by effectively handling covariate shift and generalization.

Contribution

It presents a novel online imitation learning approach that eliminates the need for state estimation and planning, validated through real-world high-speed off-road driving experiments.

Findings

Policies trained with online imitation learning outperform batch methods.

The system achieves high-speed off-road driving matching state-of-the-art performance.

Experimental validation confirms the effectiveness of the approach.

Abstract

We present an end-to-end imitation learning system for agile, off-road autonomous driving using only low-cost sensors. By imitating a model predictive controller equipped with advanced sensors, we train a deep neural network control policy to map raw, high-dimensional observations to continuous steering and throttle commands. Compared with recent approaches to similar tasks, our method requires neither state estimation nor on-the-fly planning to navigate the vehicle. Our approach relies on, and experimentally validates, recent imitation learning theory. Empirically, we show that policies trained with online imitation learning overcome well-known challenges related to covariate shift and generalize better than policies trained with batch imitation learning. Built on these insights, our autonomous driving system demonstrates successful high-speed off-road driving, matching the state-of-the-art performance.