Exploring the Limitations of Behavior Cloning for Autonomous Driving
This work addresses the scalability of imitation learning for autonomous driving, but it is incremental as it primarily benchmarks and confirms known limitations.
The paper tackles the problem of scaling behavior cloning for autonomous driving by proposing a new benchmark to investigate its limitations, showing it achieves state-of-the-art results in unseen environments but confirming issues like dataset bias and generalization that hinder real-world application.
Driving requires reacting to a wide variety of complex environment conditions and agent behaviors. Explicitly modeling each possible scenario is unrealistic. In contrast, imitation learning can, in theory, leverage data from large fleets of human-driven cars. Behavior cloning in particular has been successfully used to learn simple visuomotor policies end-to-end, but scaling to the full spectrum of driving behaviors remains an unsolved problem. In this paper, we propose a new benchmark to experimentally investigate the scalability and limitations of behavior cloning. We show that behavior cloning leads to state-of-the-art results, including in unseen environments, executing complex lateral and longitudinal maneuvers without these reactions being explicitly programmed. However, we confirm well-known limitations (due to dataset bias and overfitting), new generalization issues (due to dynamic objects and the lack of a causal model), and training instability requiring further research before behavior cloning can graduate to real-world driving. The code of the studied behavior cloning approaches can be found at https://github.com/felipecode/coiltraine .