A Super-human Vision-based Reinforcement Learning Agent for Autonomous Racing in Gran Turismo

TL;DR

This paper presents a novel vision-based reinforcement learning agent for autonomous racing in Gran Turismo, capable of outperforming human drivers using only onboard visual and sensor data, with global features used solely during training.

Contribution

It introduces the first super-human racing agent that relies solely on local visual and sensor inputs, eliminating the need for external instrumentation during operation.

Findings

Outperforms top human drivers in time trial races.

Relies primarily on visual inputs for decision making.

Effective across multiple tracks and car models.

Abstract

Racing autonomous cars faster than the best human drivers has been a longstanding grand challenge for the fields of Artificial Intelligence and robotics. Recently, an end-to-end deep reinforcement learning agent met this challenge in a high-fidelity racing simulator, Gran Turismo. However, this agent relied on global features that require instrumentation external to the car. This paper introduces, to the best of our knowledge, the first super-human car racing agent whose sensor input is purely local to the car, namely pixels from an ego-centric camera view and quantities that can be sensed from on-board the car, such as the car's velocity. By leveraging global features only at training time, the learned agent is able to outperform the best human drivers in time trial (one car on the track at a time) races using only local input features. The resulting agent is evaluated in Gran Turismo 7 on multiple tracks and cars. Detailed ablation experiments demonstrate the agent's strong reliance on visual inputs, making it the first vision-based super-human car racing agent.