Beyond Planning: Differentiable Control-in-the-loop End-to-End Autonomous Driving
Beyond Planning: Differentiable Control-in-the-loop End-to-End Autonomous Driving
This research introduces DIVA, a differentiable control-in-the-loop framework for end-to-end autonomous driving that integrates an identified vehicle model into the training loop. By enabling gradient flow through the control system, DIVA closes the gap between planning and control, achieving a driving score of 86.7 on the Bench2Drive benchmark, surpassing state-of-the-art methods with reduced jerk and brake switching.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
driving score
86.7
View original evidence
ss training-time distribution shift and deployment-time model mismatch respectively. On the closed-loop Bench2Drive benchmark, DIVA achieves a driving score of 86.7, surpassing the widely adopted state-of-the-art approach by 3.3 points while nearly halving mean jerk and reducing brake switching by 64%, and maintains superior path tracking accuracy under significant vehicle model mismatch at deployment.
Open source S5control-only cycle
3.8
View original evidence
rception to vehicle states. A learning-based control decoder replaces the external controller and is supervised on the rolled-out vehicle states, bypassing the aliased control-input labels. A rate-consistent control-input parameterization reconciles the heterogeneous sampling rates and enables asynchronous execution with a 3.8 ms control-only cycle. On-policy re-identification and an online disturbance observer addre
Open source S4
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ss training-time distribution shift and deployment-time model mismatch respectively. On the closed-loop Bench2Drive benchmark, DIVA achieves a driving score of 86.7, surpassing the widely adopted state-of-the-art approach by 3.3 points while nearly halving mean jerk and reducing brake switching by 64%, and maintains superior path tracking accuracy under significant vehicle model mismatch at deployment.
Open source S5
rception to vehicle states. A learning-based control decoder replaces the external controller and is supervised on the rolled-out vehicle states, bypassing the aliased control-input labels. A rate-consistent control-input parameterization reconciles the heterogeneous sampling rates and enables asynchronous execution with a 3.8 ms control-only cycle. On-policy re-identification and an online disturbance observer addre
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org