Problem
Research questions and friction points this paper is trying to address.
end-to-end autonomous driving
imitation learning
execution error accumulation
long-horizon objectives
reliability
Post-Training in End-to-End Autonomous Driving
🤖 AI Summary
End-to-end autonomous driving models often struggle to ensure reliability in safety-critical, highly interactive scenarios due to error accumulation, limited recovery capability, and the absence of long-horizon objectives, making imitation learning alone insufficient. This work presents the first unified framework that systematically organizes post-training methods for such models, categorizing them into four classes based on the form of supervision: reinforcement learning, online fine-tuning, closed-loop simulation-based optimization, and human feedback. It provides a thorough analysis of the capabilities and fundamental challenges inherent to each approach. By clearly delineating the scope of post-training research, this study offers a roadmap for enhancing the reliability and efficiency of driving policies and identifies key directions for future investigation.
📝 Abstract
End-to-end models that map multimodal inputs directly to future trajectories/maneuvers have emerged as an increasingly prominent research paradigm in autonomous driving. This class of models includes both Vision-Language-Action models and trajectory-generative planners. Unlike classic machine learning applications, autonomous vehicles operate in safety-critical and interaction-intensive environments where traditional open-loop imitation of expert demonstrations is not sufficient to ensure reliability. In particular, small execution errors can accumulate over time, while recovery behaviors are scarce in training data. In addition, long-horizon objectives such as safety and driving comfort are not captured by pointwise labels either. These limitations have motivated a shift toward post-training techniques, which further refine driving policies beyond pure imitation. This survey presents a unified view of post-training for autonomous driving by defining its scope and organizing the existing literature into four major families based on the form of supervision they use. For each family, we discuss its capabilities, limitations, and open challenges. We aim to facilitate a systematic understanding of this emerging area and stimulate future research on reliable and efficient post-training for autonomous driving.
Innovation
Methods, ideas, or system contributions that make the work stand out.
post-training
end-to-end autonomous driving
trajectory refinement
driving policy optimization
supervision signals
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
