ISE Graduates Won Best Paper at IROS 2025 for Breakthrough in Human-Robot Collaboration

ISE graduates Benhua Gao, Tian Wang, Zeyuan Ren won the Best Paper Award at the Workshop on Generative AI for Robotics and Smart Manufacturing, held during the 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025). The awarded paper is titled “InstructTODG: A Multimodal LLMs-driven Approach for Task-oriented Dexterous Grasping in Unstructured Human-Robot Collaborative Manufacturing,” supervised by Professor Zheng Pai.

As one of the most influential international conferences in the field of robotics, IROS 2025 was held under the theme of "Human-Robotics Frontier," focusing on how artificial intelligence systems and intelligent machines can transform industries, enhance productivity, and address global challenges. The conference featured multiple workshops, including one organized by researchers from EPFL, which focused on the challenges of deploying robots supported by generative AI and foundational models at an industrial scale to drive the development of manufacturing. The workshop encouraged cutting-edge and forward-looking research on generative AI in the fields of robotics and manufacturing. The awarded paper centers on human-robot collaborative manufacturing, highlighting the critical need for robots to achieve not only stable and precise grasps but also seamless alignment with task-specific human instructions during dexterous manipulation. To tackle this challenge, the awarded paper introduces InstructTODG, an innovative framework powered by multimodal large language model (MLLM) aimed at achieving task-oriented dexterous grasping in unstructured HRC environments, significantly enhancing robot manipulation capabilities and advancing human-robot collaboration in complex manufacturing tasks.

ISE graduates Benhua Gao, Tian Wang, Zeyuan Ren won the Best Paper Award at the Workshop on Generative AI for Robotics and Smart Manufacturing, held during the 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025). The awarded paper is titled “InstructTODG: A Multimodal LLMs-driven Approach for Task-oriented Dexterous Grasping in Unstructured Human-Robot Collaborative Manufacturing,” supervised by Professor Zheng Pai.

As one of the most influential international conferences in the field of robotics, IROS 2025 was held under the theme of "Human-Robotics Frontier," focusing on how artificial intelligence systems and intelligent machines can transform industries, enhance productivity, and address global challenges. The conference featured multiple workshops, including one organized by researchers from EPFL, which focused on the challenges of deploying robots supported by generative AI and foundational models at an industrial scale to drive the development of manufacturing. The workshop encouraged cutting-edge and forward-looking research on generative AI in the fields of robotics and manufacturing. The awarded paper centers on human-robot collaborative manufacturing, highlighting the critical need for robots to achieve not only stable and precise grasps but also seamless alignment with task-specific human instructions during dexterous manipulation. To tackle this challenge, the awarded paper introduces InstructTODG, an innovative framework powered by multimodal large language model (MLLM) aimed at achieving task-oriented dexterous grasping in unstructured HRC environments, significantly enhancing robot manipulation capabilities and advancing human-robot collaboration in complex manufacturing tasks.