Embedding Synthetic Off-Policy Experience for Autonomous Driving via Zero-Shot Curricula
This addresses the issue of safety-critical performance in autonomous driving by improving robustness through data prioritization, though it is incremental as it builds on existing imitation learning methods.
The paper tackled the problem of robust motion planning for autonomous driving by predicting the difficulty of driving situations from fleet data and using it to create a zero-shot curriculum for imitation learning. The result was a 15% reduction in collisions and a 14% increase in route adherence while using only 10% of the training data.
ML-based motion planning is a promising approach to produce agents that exhibit complex behaviors, and automatically adapt to novel environments. In the context of autonomous driving, it is common to treat all available training data equally. However, this approach produces agents that do not perform robustly in safety-critical settings, an issue that cannot be addressed by simply adding more data to the training set - we show that an agent trained using only a 10% subset of the data performs just as well as an agent trained on the entire dataset. We present a method to predict the inherent difficulty of a driving situation given data collected from a fleet of autonomous vehicles deployed on public roads. We then demonstrate that this difficulty score can be used in a zero-shot transfer to generate curricula for an imitation-learning based planning agent. Compared to training on the entire unbiased training dataset, we show that prioritizing difficult driving scenarios both reduces collisions by 15% and increases route adherence by 14% in closed-loop evaluation, all while using only 10% of the training data.