外企招聘,北京朝阳区工作,具备博士后科研工作站资质
【职位描述】
□ 工作内容
- 主导专为机器人操作、移动或全身运动定制的深度强化学习(RL)算法的设计与实现。
- 构建可扩展的仿真到现实迁移(sim-to-real transfer)及领域自适应流水线。
- 负责基于强化学习的策略在实际场景中的部署,包括安全评估、系统集成和故障恢复。
- 推动动态环境中通用化、样本高效且鲁棒的强化学习策略的研发。
□ 工作经验
- 主导从模型设计到实际部署的端到端强化学习机器人项目。
- 具备将强化学习策略从仿真环境迁移至物理机器人的丰富实践经验。
- 熟悉动态足式机器人移动(如双足/四足机器人)和/或使用多自由度手臂/手部的灵巧操作。
- 具备通过领域自适应、策略迁移或现实微调弥合仿真与现实差距的经验。
□ 专业知识
- 深入理解强化学习理论及连续控制算法。
- 掌握机器人动力学、控制策略及富接触交互知识。
- 熟练使用强化学习模型实现的机器学习框架。
- 熟悉全身控制、平衡或基于力的操作(优先考虑)。
□ 任职要求
- 机器人学、机器学习、计算机科学或相关领域博士学位,或具备同等强化学习机器人控制研发/工程经验。
- 具备开发和部署用于操作或足式移动的强化学习策略的成功案例。
- 精通Python、C++编码与调试,具备大规模强化学习训练流水线开发经验。
□ Task
- Lead the design and implementation of deep RL algorithms tailored for robotic manipulation, locomotion, or whole body motion
- Architect scalable pipelines for sim-to-real transfer, domain adaptation
- Field deployment of RL-based policies, including safety evaluation, system integration, and failure recovery
- Drive research and development toward generalizable, sample-efficient, and robust RL policies in dynamic environments.
□ Job Experience
- Led end-to-end RL-based robotic projects from model design to real-world deployment
- Extensive hands-on experience transferring RL policies from simulation to physical robots
- Experience with dynamic legged robot locomotion (e.g., bipedal/quadrupeds) and/or dexterous manipulation using multi-DOF arms or hands
- Experience bridging the sim-to-real gap through domain adaptation, policy transfer, or real-world fine-tuning.
□ Knowledge
- Deep understanding of reinforcement learning theory and continuous control algorithms.
- Knowledge of robot dynamics, control strategies, and contact-rich interactions.
- Proficiency in ML frameworks for RL model implementation
- Familiarity with whole-body control, balance or force-based manipulation is a plus
□ Requirements
- Ph.D. in Robotics, Machine Learning, Computer Science, or a related field, or equivalent research/engineering experience in RL-based robotic control.
- Proven track record of developing and deploying RL policies for manipulation or legged locomotion.
- Strong coding and debugging skills in Python, C++ and experience working with large-scale RL training pipelines.