Learning Stability Attention in Vision-based End-to-end Driving Policies
This addresses stability issues in autonomous driving systems, which is an incremental improvement for safety-critical applications.
The paper tackled the problem of ensuring stability and robustness in vision-based end-to-end driving policies by proposing stability attention in control Lyapunov functions (att-CLFs) and an uncertainty propagation technique, demonstrating effectiveness in simulations and on a real autonomous vehicle.
Modern end-to-end learning systems can learn to explicitly infer control from perception. However, it is difficult to guarantee stability and robustness for these systems since they are often exposed to unstructured, high-dimensional, and complex observation spaces (e.g., autonomous driving from a stream of pixel inputs). We propose to leverage control Lyapunov functions (CLFs) to equip end-to-end vision-based policies with stability properties and introduce stability attention in CLFs (att-CLFs) to tackle environmental changes and improve learning flexibility. We also present an uncertainty propagation technique that is tightly integrated into att-CLFs. We demonstrate the effectiveness of att-CLFs via comparison with classical CLFs, model predictive control, and vanilla end-to-end learning in a photo-realistic simulator and on a real full-scale autonomous vehicle.