BeTAIL: Behavior Transformer Adversarial Imitation Learning From Human Racing Gameplay

摘要

Autonomous racing poses a significant challenge for control, requiring planning minimum-time trajectories under uncertain dynamics and controlling vehicles at their handling limits. Current methods requiring hand-designed physical models or reward functions specific to each car or track. In contrast, imitation learning uses only expert demonstrations to learn a control policy. Imitated policies must model complex environment dynamics and human decision-making. Sequence modeling is highly effective in capturing intricate patterns of motion sequences but struggles to adapt to new environments or distribution shifts that are common in real-world robotics tasks. In contrast, Adversarial Imitation Learning (AIL) can mitigate this effect, but struggles with sample inefficiency and handling complex motion patterns. Thus, we propose BeTAIL: Behavior Transformer Adversarial Imitation Learning, which combines a Behavior Transformer (BeT) policy from human demonstrations with online AIL. BeTAIL adds an AIL residual policy to the BeT policy to model the sequential decision-making process of human experts and correct for out-of-distribution states or shifts in environment dynamics. We test BeTAIL on three challenges with expert-level demonstrations of real human gameplay in the high-fidelity racing game Gran Turismo Sport. Our proposed BeTAIL reduces environment interactions and improves racing performance and stability, even when the BeT is pretrained on different tracks than downstream learning.