MomADv2: Reliable Temporal Memory for End-to-End Autonomous Driving
2026-08-24 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionRobotics
AI summaryⓘ
The authors developed MomADv2, a system to help self-driving cars plan their routes better over longer times. They improve how the system remembers past decisions by carefully choosing which past information is useful and matches the current driving command. They also added a way to fix small trajectory errors smoothly to keep the plan stable. Tests show their method reduces collisions and improves planning consistency compared to the previous version.
long-horizon planningstate-space memorytemporal continuitycommand consistencytrajectory refinementautonomous drivingend-to-end planningcollision ratetrajectory residualplanning stability
Authors
Ziying Song, Shengkai Zhang, Lin Liu, Peiliang Wu, Lei Yang, Dongyang Xu, Bin Sun, Li Wang, Shaoqing Xu, Caiyan Jia, Yadan Luo
Abstract
Long-horizon planning is critical for safe autonomous driving in complex scenarios. Existing methods improve planning continuity with temporal memory, but such memory may become invalid and mislead decisions when the driving command changes. Thus, selectively leveraging useful history while suppressing command-inconsistent memory remains a key challenge. To address this issue, we propose MomADv2, a reliable state-space memory framework for long-horizon end-to-end autonomous driving. At its core, MomADv2 introduces a Selective State-Space Planning Memory Query Module, which filters historical planning queries based on temporal continuity and command consistency, selects planning modes relevant to the current command, and models the evolution of planning intentions through a selective state-space mechanism. To further alleviate local trajectory deviations and error accumulation in long-horizon planning, we design a Flow-Matching Trajectory Residual Refiner. It learns a continuous residual correction field from the refined planning output to the expert trajectory, enabling fine-grained trajectory refinement while preserving the stability of anchor-based planning. Extensive experiments on closed-loop NAVSIM and Bench2Drive, as well as open-loop nuScenes, demonstrate that MomADv2 improves long-horizon planning consistency and reduces the average collision rate by 15.6% over MomAD under 6-second planning.