Rectify, Don't Regret: Avoiding Pitfalls of Differentiable Simulation in Trajectory Prediction
This addresses a critical issue for autonomous driving systems by improving robustness and safety in trajectory prediction, though it is an incremental advance over existing simulation methods.
The paper tackles the problem of trajectory prediction in autonomous driving, where current models suffer from compounding errors and shortcut learning in differentiable simulators, and introduces a detached receding horizon rollout method that reduces target collisions by up to 33.24% compared to fully differentiable training and up to 27.74% compared to open-loop baselines.
Current open-loop trajectory models struggle in real-world autonomous driving because minor initial deviations often cascade into compounding errors, pushing the agent into out-of-distribution states. While fully differentiable closed-loop simulators attempt to address this, they suffer from shortcut learning: the loss gradients flow backward through induced state inputs, inadvertently leaking future ground truth information directly into the model's own previous predictions. The model exploits these signals to artificially avoid drift, non-causally "regretting" past mistakes rather than learning genuinely reactive recovery. To address this, we introduce a detached receding horizon rollout. By explicitly severing the computation graph between simulation steps, the model learns genuine recovery behaviors from drifted states, forcing it to "rectify" mistakes rather than non-causally optimizing past predictions. Extensive evaluations on the nuScenes and DeepScenario datasets show our approach yields more robust recovery strategies, reducing target collisions by up to 33.24% compared to fully differentiable closed-loop training at high replanning frequencies. Furthermore, compared to standard open-loop baselines, our non-differentiable framework decreases collisions by up to 27.74% in dense environments while simultaneously improving multi-modal prediction diversity and lane alignment.