论文/arXiv 2609.28969·1 个月前

Sim-to-Real 感知的微出行端到端学习环境

针对微出行车辆仿真器难捕捉特定运动学的问题，构建面向WHILL Model CR的Sim-to-Real端到端学习环境，并在仿真避障与真机走廊中验证。

AI 整理

• 本体 SDK 与控制
• 感知与导航
• 仿真与训练

原文链接：Sim-to-Real Aware End-to-End Learning Environment for Micromobility（arXiv 2609.28969 · 发布日期：2026-09-24）

Sim-to-Real Aware End-to-End Learning Environment for Micromobility

针对微出行车辆仿真器难捕捉特定运动学的问题，构建面向WHILL Model CR的Sim-to-Real端到端学习环境，并在仿真避障与真机走廊中验证。

核心要点

• 现有端到端自动驾驶用于微出行时，受限于仿真器难刻画差速驱动、全向轮等运动学。
• 环境基于AWSIM和ROS 2，针对WHILL Model CR进行车辆特定设计。
• 用真实数据和贝叶斯优化校准物理参数，以降低多种驾驶场景的轨迹误差。
• DreamerV3策略在仿真避障中完成任务，并无需微调迁移到真车绕过纸箱。

验证范围与局限：验证包括AWSIM中的仿真障碍物避让，以及WHILL Model CR在真实走廊复现环境中绕过纸箱。

英文摘要 (Abstract)

While end-to-end autonomous driving systems show promise, their application to micromobility vehicles is hindered by simulators failing to capture specific kinematics, such as differential drives and omni-wheels. This paper pro- poses a sim-to-real-aware, vehicle-specific end-to-end learning environment for the WHILL Model CR on AWSIM and ROS 2. To minimize the sim-to-real gap, physical parameters are optimized via Bayesian optimization using real-world data, reducing trajectory errors across various driving scenarios. Additionally, this study introduces a synchronized architecture tailored for the stable training of world model-based agents. An end-to-end policy trained with DreamerV3 exhibited learning progress and achieved task completion in a simulated obstacle avoidance setting. Furthermore, this policy demonstrated direct sim-to-real transfer to the physical vehicle, enabling the vehicle to navigate around a cardboard box in a real-world corridor replica without fine-tuning. This paper provides a practical foundation for sim-to-real micromobility policy studies.

作者：Shouma Amano, Takuya Azumi

来源

本页事实对应的公开页面，以原文为准。

1. arXiv 2609.28969：Sim-to-Real Aware End-to-End Learning Environment for Micromobility访问 2026-10-21