Biomechatronics & Intelligent Robotics Lab

Robotics postdoc position at University of Tulsa #robotics #AI #jobs

Workshop July 15 in Boston IEEE/ASME International Conference on Advanced Intelligent Mechatronics conference. https://www.aim2024.org

Join us July 15th 9AM in Boston for Future of Work in the Age of Robotics and AI workshop (half-day) at IEEE/ASME International Conference on Advanced Intelligent Mechatronics (https://www.aim2024.org) Thanks for Minghui Zheng, Tan Chen, Hao Su, Jingang Yi, and Ellen Mazumdar for organizing this workshop IEEE and ASME (The American Society of Mechanical Engineers) Workshop website: https://lnkd.in/eFwVmG7c

It is a great pleasure for Hao Su's team Biomechatronics & Intelligent Robotics Lab to work with Jie Yin and his team including Yanbin Li to lead this effort on mechanical computing metastructure platform published in Science Advances. North Carolina State University NC State College of Engineering NC State Mechanical & Aerospace Engineering Y. Li, S. Yu, H. Qing, Y. Hong, Y. Zhao, F. Qi, H. Su, J. Yin "Reprogrammable and reconfigurable mechanical computing metastructures with stable and high-density memory", Science Advances, 2024 Paper PDF: https://lnkd.in/gmtG8WRK Video: https://lnkd.in/gvkp8y-G

We are very privileged to share our research paper published in Nature titled "Experiment-free Exoskeleton Assistance via Learning in Simulation," The full paper is available: https://rdcu.be/dKCZV Video: https://lnkd.in/gdFD7nnh NCSU news: https://lnkd.in/gCsswrAF Can wearable robots be intelligent, versatile, and easy to use to improve human health and quality of life? We presented a “learning-in-simulation” framework to rapidly develop robot controllers to assist multi-locomotion activities without relying on any human-involved experiments - one of the grand challenges of wearable robots that have significantly hindered their widespread adoption. Our reinforcement learning-based method does not require any human tests or handcrafting of control rules. The learned controller can be directly deployed on a portable exoskeleton which immediately reduces energy consumption during walking, running, and stair climbing. Unlike previous achievements of reinforcement learning that primarily focused on simulation or board games (e.g., AlphaGo) or robots only, we proposed a new method, data-driven and physics-informed deep #reinforcementlearning, to control wearable robots to directly benefit humans. To bridge the sim2real gap, the key idea is to build the high-fidelity simulation of human, robot, and human-robot interaction, then simultaneously learn 3 neural networks in a closed-loop fashion with NVIDIA GPU to identify the controller policy network.?We will further personalize this framework with methods like human-in-the-loop. This work represents a milestone for wearable robots, human-centered AI, which may offer a generalizable and scalable strategy for rapid development and adoption of assistive robots for able-bodied and mobility-impaired individuals. This research would have never been possible without the amazing support of many collaborators! This work is led by Shuzhen Susan Luo, Megan Jiang, Sainan Zhang, Junxi Zhu, Israel Dominguez Silva?Photos by Weibo Gao. We are grateful for the effort and contribution of co-authors Xianlian Alex Zhou, Bolei Zhou, Elliott Rouse Hyunwoo Yuk. We thank National Science Foundation (NSF), The National Institutes of Health for the continuous support of our research throughout these years! We also thank the support by North Carolina State University NC State Mechanical & Aerospace Engineering NC State College of Engineering University of North Carolina at Chapel Hill Joint Department of Biomedical Engineering | UNC & NC State Reference: S. Luo, M. Jiang, S. Zhang, J. Zhu, S. Yu, I. Dominguez, E. Rouse, B. Zhou, H. Yuk, X. Zhou, and H. Su. Experiment-free exoskeleton assistance via learning in simulation. Nature, 630, 353–359 (2024) DOI: 10.1038/s41586-024-07382-4 View and comment by Nature and Prof. Alexandra (Sasha) Voloshina. Thanks for your great summary. https://lnkd.in/ggAHuZZm

Ph.D. Student and now postdoc Sainan Zhang presented her work to make wearable robots compact and versatile in real world at Dynamic Walking. https://lnkd.in/e6VQP72g

Proud to share our exoskeleton work here at North Carolina State University, NC State Mechanical & Aerospace Engineering! A great experience with my colleagues Antonio Di Lallo, Ivan Alonso Lopez Sanchez, and Sainan Zhang. Special thanks to our advisor Professor Hao Su In our most recent work, we tackle the comfort challenges in current hip exoskeletons: Hip-assistance exoskeletons have shown the potential to improve human mobility, but they are still far from widespread adoption. Comfort is a current challenge: unattractive, cumbersome wearable robots will not make an impact. This includes excessive human-exoskeleton interaction forces leading to undesired movements (wobbling) and likely hindering their assistive benefits. In this talk, we introduce our comfort-centered mechatronic design for portable hip exoskeletons, which significantly improves comfort: the exoskeleton, with neural network capabilities, reduced wobbling by 74%, while the metabolic cost of walking was reduced by 18% and got an excellent System Usability Scale (SUS) score of 74. [1] D. Rodríguez-Jorge, S. Zhang, N. Srinivasan, M. Jiang, I. Dominguez, X. Zhou, Q. Zhang, H. Su, Comfort-Centered Design of Lightweight Hip Exoskeleton with Real-Time Machine Learning Capability (to be submitted)

PhD student Sainan Zhang work on optimization of actuators for pediatric rehabilitation to make robots lightweight and efficient is published in journal Mechatronics https://lnkd.in/etthEESM S. Zhang, J. Zhu, T.-H. Huang, S. Yu, J. S. Huang, I. Lopez-Sanchez, T. Devine, M. Abdelhady, M. Zheng, T. C. Bulea, and H. Su, “Actuator optimization and deep learning-based control of pediatric knee exoskeleton for community-based mobility assistance,” Mechatronics, vol. 97, p. 103109, Feb. 2024.