上海期智研究院PI,清华大学交叉信息研究院助理教授。
在清华大学取得学士学位,在加州大学伯克利分校取得博士学位,师从美国国家工程院院士、机电控制学科先驱Masayoshi Tomizuka教授。近年来在机器人与人工智能的交叉领域从事前沿研究。研究目标是构建出具备高性能、高智能的高端机器人软硬件系统。他在机器人、人工智能、控制、交通等领域的国际顶级会议和期刊上发表了四十余篇论文,部分论文入围L4DC 2022、IEEE IV 2021、IFAC MECC 2021等国际会议优秀论文奖。
个人荣誉
福布斯中国30位30岁以下精英(科学榜,2021年)
人形机器人:构建人形通用智能机器人的本体及其运动控制能力
具身通用人工智能:构建有手有腿,能听会说,能装载在实体机器人上与真实物理世界进行交互的通用人工智能
成果2:人形通用智能机器人
构建像人一样的智能机器人是人类一直以来的梦想。人形机器人在软硬件方面都是所有机器人中集成度最高,最复杂的。它能去到所有人能去的地方,干所有人能干的事。由于其完美适用于人类环境,人形机器人是通用人工智能的最佳载体,也具备极其广阔的市场空间。陈建宇团队致力于构建人形通用智能机器人,包括通用本体及其通用智能。通用本体方面,团队研发了基于本体感知驱动器的高性能、低成本人形机器人硬件本体“小星”机器人,搭配具备高扭矩密度电机以及低减速比减速器的一体化关节模组。运动控制算法上,“小星”在全球范围内首次实现了人形机器人端到端强化学习野外雪地行走,包括雪地上下坡,以及上下楼梯。该过程不需要依赖于预先编程的行走模式,而是完全通过AI自主学习实现的。这使得机器人能够自主地适应不同的地面条件,从而在复杂的雪地环境中稳定行走。
在具身智能方面,团队提出了DoReMi框架,这是世界第一篇用大语言模型赋能人形机器人决策的文章。该工作改进了上层语言模型规划与下层强化学习策略的对齐问题,将视觉语言模型、大语言模型与人形机器人算法进行整合,用大型语言模型指导小星的上层任务规划,用强化学习来获取小星的底层控制器,构成的框架可以增强其执行任务的智能性和泛化性。
------------------------------------------------------------------------------------------------------------------------------
成果1:安全强化学习
强化学习对解决复杂的机器人问题具有巨大的前景,包括自动驾驶、机械臂操作等。然而将强化学习应用于上述实体机器人系统上时却有很大的安全隐患。无论是训练过程中还是训练收敛后,基于神经网络的机器人控制策略都可能出现问题并造成与环境之间的危险碰撞。
陈建宇团队首先从强化学习的约束保障出发,基于控制理论中的前向不变性,提出了针对机器人动力系统的通用安全保障机制,以及耦合该机制的强化学习框架。在理论允许的范围中,该方法能保障系统的安全性,并在实际算法验证中得到零安全约束违反的结果。同时,针对约束无法避免的情况,团队通过整合冗余自由度机械臂的零空间控制、变阻抗控制、以及基于视觉的强化学习等理论与算法,提出了一套保障接触安全的强化学习框架。该框架能不仅能保障真实机器人应用强化学习时,其末端执行器的接触力比较柔顺,同时在机械臂身遇到意外的接触与碰撞时能及时推断出来并进行柔顺处理。
22. Prediction with Action: Visual Policy Learning via Joint Denoising Process, Yanjiang Guo∗, Yucheng Hu*, Jianke Zhang, Yen-Jen Wang, Xiaoyu Chen, Chaochao Lu, Jianyu Chen†, https://sites.google.com/view/pad-paper, NeurIPS 2024.
21. HiRT: Enhancing Robotic Control with Hierarchical Robot Transformers, Jianke Zhang∗, Yanjiang Guo∗, Xiaoyu Chen,Yen-Jen Wang, Yucheng Hu, Chengming Shi, Jianyu Chen†, https://arxiv.org/abs/2410.05273, CoRL 2024.
20. Whleaper: A 10-DOF High-Performance Bipedal Wheeled Robot, Yinglei Zhu, Sixiao He, Zhenghao Qi, Zhuoyuan Yong, Yihua Qin, Jianyu Chen†, https://rasevents.org/presentation?id=146224, IROS 2024.
19. DoReMi: Grounding Language Model by Detecting and Recovering from Plan-Execution Misalignment, Yanjiang Guo*, Yen-Jen Wang*, Lihan Zha*, Jianyu Chen †, https://sites.google.com/view/doremi-paper, IROS 2024.
18. Advancing Humanoid Locomotion: Mastering Challenging Terrains with Denoising World Model Learning, Xinyang Gu, Yen-Jen Wang, Xiang Zhu, Chengming Shi, Yanjiang Guo, Yichen Liu, Jianyu Chen, RSS 2024.
17. Yanjiang Guo, Zheyuan Jiang, Yen-Jen Wang, Jingyue Gao, Jianyu Chen, Decentralized Motor Skill Learning for Complex Robotic Systems, ICRA 2024
16. Yanjiang Guo, Zheyuan Jiang, Yen-Jen Wang, Jingyue Gao, Jianyu Chen, Decentralized Motor Skill Learning for Complex Robotic Systems, IEEE Robotics and Automation Letters (RA-L), 2023 查看PDF
15. Yujie Yang, Yuxuan Jiang, Jianyu Chen, Shengbo Eben Li, Ziqing Gu, Yuming Yin, Qian Zhang, Kai Yu, Belief State Actor-Critic Algorithm from Separation Principle for POMDP, American Control Conference (ACC), 2023 查看PDF
14. Yujie Yang, Yuxuan Jiang, Yichen Liu, Jianyu Chen, Shengbo Eben Li, Model-Free Safe Reinforcement Learning Through, IEEE Robotics and Automation Letters (RA-L), 2023 查看PDF
13. Hai Zhong, Yutaka Shimizu, Jianyu Chen, Chance-Constrained Iterative Linear-Quadratic Stochastic Games, IEEE Robotics and Automation Letters (RA-L), 2022 查看PDF
12. Yanjiang Guo, Jingyue Gao, Zheng Wu, Chengming Shi, Jianyu Chen, Reinforcement learning with Demonstrations from Mismatched Task under Sparse Reward, International Conference on Robots Learning (CORL), 2022 查看PDF
11. Zheyuan Jiang, Jingyue Gao, Jianyu Chen, Unsupervised Skill Discovery via Recurrent Skill Training, Conference on Neural Information Processing Systems (NeurIPS), 2022 查看PDF
10. Xiang Zhu, Shucheng Kang, Jianyu Chen, A Contact-Safe Reinforcement Learning Framework for Contact-Rich Robot Manipulation, International Conference on Intelligent Robots and Systems (IROS), 2022 查看PDF
9. Xiaoyu Chen, Yao Mark Mu, Ping Luo, Shengbo Eben Li, Jianyu Chen, Flow-based Recurrent Belief State Learning for POMDPs, International Conference on Machine Learning (ICML), 2022 查看PDF
8. Dongjie Yu, Haitong Ma, Shengbo Eben Li, Jianyu Chen, Reachability Constrained Reinforcement Learning, International Conference on Machine Learning (ICML), 2022 查看PDF
7. Haitong Ma, Changliu Liu, Shengbo Eben Li, Sifa Zheng, Jianyu Chen, Joint Synthesis of Safety Certificate and Safe Control Policy using Constrained Reinforcement Learning, Conference on Learning for Dynamics and Control (L4DC), 2022 查看PDF
6. Yuheng Lei, Jianyu Chen, Shengbo Eben Li, Sifa Zheng, Performance-Driven Controller Tuning via Derivative-Free Reinforcement Learning, IEEE Conference on Decision and Control (CDC), 2022 查看PDF
5. Yujie Yang, Jianyu Chen, Shengbo Li, Learning POMDP Models with Similarity Space Regularization: a Linear Gaussian Case Study, Conference on Learning for Dynamics and Control (L4DC), 2022 查看PDF
4. Baiyu Peng, Jingliang Duan, Jianyu Chen, Shengbo Eben Li, Genjin Xie, Congsheng Zhang, Yang Guan, Yao Mu, Enxin Sun, Model-Based Chance-Constrained Reinforcement Learning via Separated Proportional-Integral Lagrangian, IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2022 查看PDF
3. Haitong Ma, Jianyu Chen,Shengbo Eben, Ziyu Lin,Yang Guan, Yangang Ren, Sifa Zheng, Model-based Constrained Reinforcement Learning using Generalized Control Barrier Function, International Conference on Intelligent Robots and Systems (IROS), 2021 查看PDF
2. Jianyu Chen, Yutaka Shimizu, Liting Sun, Masayoshi Tomizuka, Wei Zhan, Constrained Iterative LQG for Real-Time Chance-Constrained Gaussian Belief Space Planning, International Conference on Intelligent Robots and Systems (IROS), 2021 查看PDF
1. Jianyu Chen, Shengbo Eben Li, Masayoshi Tomizuka, Interpretable end-to-end urban autonomous driving with latent deep reinforcement learning, IEEE Transactions on Intelligent Transportation Systems (TITS), 2021 查看PDF