← Back to all news
arXiv Computer Vision By Li-Heng Chen, Qifeng Chen, Chengwei Wu, Yingjie Zhang, Sheng Yang, Hongbo Fu, Shaoqing Ren

FPD-Drive: Future Policy Distillation for End-to-End Autonomous Driving

Read the original on arXiv Computer Vision →

FPD-Drive is an end‑to‑end autonomous driving planner that uses future‑privileged multi‑target distillation to transfer diverse, high‑quality trajectory targets from a teacher model with access to future observations to a student policy. It also introduces safety‑aware score correction to improve candidate selection by adjusting rankings between safe and unsafe trajectories. The approach achieves state‑of‑the‑art performance on NAVSIM v1 and v2 benchmarks and demonstrates strong zero‑shot closed‑loop results on HUGSIM.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computer Vision.

Related stories

arXiv Machine Learning
Sep 18

OPTED: On-Policy Fine-Tuning for End-to-End Driving using a Render-Free Teacher

OPTED is a method for on‑policy fine‑tuning of end‑to‑end driving models that separates reinforcement learning from the policy update. A privileged teacher trained with RL on vectorized inputs (HD‑maps and bounding boxes) supervises the pre‑trained student during closed‑loop post‑training. Applied to the camera‑based models TransFuser and VaVAM in AlpaSim, OPTED boosts driving scores by 1.6× and 9.5×, respectively, while requiring roughly three orders of magnitude fewer simulator interactions than direct RL post‑training.

By Damiano Da Col, Maximilian Igl, Peter Karkus, Kashyap Chitta, Boris Ivanovic, Marco Pavone, Konrad Schindler, Christos Sakaridis
agentsreinforcement-learningroboticsfine-tuningsafety
More like this →
arXiv Computer Vision
Sep 3

Designing Versatile Samples for Learned Trajectory Scoring

The paper proposes a new training dataset that generates more informative positive and negative samples for trajectory scoring in autonomous driving. By perturbing logged human trajectories laterally toward the drivable boundary and longitudinally toward a leading vehicle, the dataset provides richer supervision than the planner’s default proposal pool. Using a transformer-based scorer trained on this dataset, the authors achieve improved EPDMS scores on two frozen planners, DiffusionDrive and MeanFuser, when evaluated on the NAVSIM navtrain dataset.

By Yaguang Li, Jiaru Zhang, Chuheng Wei, Can Cui, Ziran Wang
llms
More like this →
arXiv AI
Jun 6

Drive-KD: Multi-Teacher Distillation for VLMs in Autonomous Driving

arXiv:2601. 21288v2 Announce Type: replace Abstract: Autonomous driving is an important and safety-critical task, and recent advances in LLMs/VLMs have opened new possibilities for reasoning and planning in this domain.

By Weitong Lian, Zecong Tang, Haoran Li, Tianjian Gao, Yifei Wang, Zixu Wang, Lingyi Meng, Tengju Ru, Zhejun Cui, Yichen Zhu, Hangshuo Cao, Qi Kang, Tianxing Chen, Kaixuan Wang, Yu Zhang
llmsagentsroboticsfine-tuningefficiencysafety
More like this →