The replication package contains processed real-world emission datasets and Python scripts used to reproduce the main analyses and figures presented in this study. The datasets include overall trip-average emissions, operating mode-specific emission results, and SCR upstream/downstream NOx data for China V and China VI heavy-duty diesel trucks tested using PEMS under real-road conditions. All analyses were conducted using Python 3.9 in PyCharm Community Edition 2023.1. Due to confidentiality restrictions, only processed and aggregated datasets are provided.

This replication package supports the manuscript “RESPOND: Risk-Enhanced Structured Pattern for LLM-driven Online Node-level Decision-making in Autonomous Driving” submitted to Communications in Transportation Research. The package includes source code, configuration files, structured memory files, simulation scripts, HighD evaluation scripts, visualization tools, and baseline implementation. It supports reproduction of the 5 × 3 risk-pattern encoding, Layer-1 exact memory, Layer-2 sub-pattern memory, hybrid Rule+LLM decision-making, reflection learning, personalization, highway-env simulation, and HighD real-world validation.

Reinforcement learning (RL) is promising for end-to-end driving, but suffers from low sample efficiency and limited semantic interpretability. Vision-language models (VLMs) provide rich knowledge, yet their high inference cost hinders integration into RL training. We propose Found-RL, a platform for enhancing AD RL with foundation models. Its core is an asynchronous batch inference framework that decouples VLM reasoning from the simulation loop, reducing latency and enabling learning from VLM feedback. On this platform, we use VMR and AWAG to distill VLM action guidance into the policy, and adopt CLIP-based reward shaping with Conditional Contrastive Action Alignment for dense supervision. Experiments show that lightweight RL policies with millions of params can approach billion-parameter VLMs while maintaining real-time speed (~500 FPS). Code, data, and models are available at https://github.com/ys-qu/found-rl.