Learning to Drive by Watching YouTube Videos: Action-Conditioned Contrastive Policy Pretraining
This work addresses the bottleneck of high interaction costs in real-world applications like autonomous driving, offering a novel pretraining approach that is incremental but impactful for the domain.
The paper tackles the problem of reducing the need for extensive online interactions in deep visuomotor policy learning for driving tasks by pretraining policy representations from uncurated YouTube videos, resulting in action-conditioned features that substantially improve downstream reinforcement and imitation learning tasks, outperforming previous unsupervised methods and ImageNet pretrained weights.
Deep visuomotor policy learning, which aims to map raw visual observation to action, achieves promising results in control tasks such as robotic manipulation and autonomous driving. However, it requires a huge number of online interactions with the training environment, which limits its real-world application. Compared to the popular unsupervised feature learning for visual recognition, feature pretraining for visuomotor control tasks is much less explored. In this work, we aim to pretrain policy representations for driving tasks by watching hours-long uncurated YouTube videos. Specifically, we train an inverse dynamic model with a small amount of labeled data and use it to predict action labels for all the YouTube video frames. A new contrastive policy pretraining method is then developed to learn action-conditioned features from the video frames with pseudo action labels. Experiments show that the resulting action-conditioned features obtain substantial improvements for the downstream reinforcement learning and imitation learning tasks, outperforming the weights pretrained from previous unsupervised learning methods and ImageNet pretrained weight. Code, model weights, and data are available at: https://metadriverse.github.io/ACO.