The 2026 IEEE International Conference on Robotics and Automation (ICRA 2026), one of the world’s premier conferences in robotics and automation, was held in Vienna, Austria, from June 1 to 5. Hesheng Wang, Dean of Shanghai Jiao Tong University Global College (SJTUGC, abbreviated as GC), was invited to deliver a keynote speech titled Learning to Navigate: From Scene Understanding to Decision Making, sharing the latest advances from his research team in robot navigation and embodied intelligence.
Accurate self-localization and environmental understanding are fundamental prerequisites for autonomous robot operation. Traditionally, navigation systems have been developed under the assumption that environments are static and structurally stable. As embodied intelligence technologies move toward real-world deployment, however, dynamic and continuously changing environments are posing new challenges to conventional Simultaneous Localization and Mapping (SLAM) techniques. In applications such as autonomous driving, service robotics, and medical robotics, moving objects, occlusions, and environmental deformations have become major obstacles for traditional navigation approaches.
Addressing these challenges, Wang systematically introduced a complete technical framework spanning environmental perception, map construction, localization, and decision-making. At the perception level, his team has developed a multimodal sensing framework that integrates LiDAR and vision sensors, combining optical flow, scene flow, and 4D reconstruction techniques to enable robust understanding of dynamic environments. For mapping, the team proposed environment representation methods based on Dynamic Gaussian SLAM and deformable 3D Gaussian maps, allowing continuous modeling of moving objects and deformable scenes.
To tackle localization and mapping challenges in highly deformable environments such as biological tissues, the team has also conducted research for surgical robot applications, exploring precise perception and navigation capabilities under complex dynamic conditions. Furthermore, to enhance robots’ long-term environmental understanding and autonomous decision-making abilities, the researchers introduced Neural Radiance Field (NeRF)-based memory mechanisms and visual-language model reasoning capabilities. These innovations provide robots with long-term environmental memory and semantic understanding, enabling them not only to localize and navigate, but also to make decisions based on historical experience and scene semantics.
According to Wang, these technologies have already been deployed in a variety of real-world applications, including autonomous mining trucks, warehouse robots, automated parking systems, mobile communication platforms, and robotic lawn mowers. As robots increasingly encounter dynamic changes, complex semantics, and environmental deformations in real-world settings, navigation technologies are evolving beyond traditional SLAM toward embodied intelligence infrastructures that integrate world models, large-model reasoning, and deformable-environment perception.
ICRA is the flagship conference of the IEEE Robotics and Automation Society and has grown into one of the most influential international conferences in robotics and automation since its inception in 1984. This year’s conference under the theme “Robots for All,” attracted more than 8,000 scholars, researchers, and industry representatives from 86 countries and regions to discuss the latest advances and future directions in robotics and automation.
Personal Introduction
Hesheng Wang is a Tang Junyuan Chair Professor at Shanghai Jiao Tong University and Dean of Shanghai Jiao Tong University Global College. He serves as Vice Chair of the Technical Committee on Hybrid Intelligence of the Chinese Association of Automation and Vice Chair of the Intelligent Vehicles and Robotics Branch of the China Instrument and Control Society.
Professor Wang currently serves or has previously served on the editorial boards of several leading international journals, including IEEE Transactions on Robotics (TRO), IEEE/ASME Transactions on Mechatronics (TMECH), IEEE Transactions on Automation Science and Engineering (TASE), IEEE Robotics and Automation Letters (RAL), International Journal of Humanoid Robotics (IJHR), and Robotica. He is Senior Editor of IEEE/ASME Transactions on Mechatronics, Advisory Editorial Board Member of Advanced Intelligent Systems, and Editor-in-Chief of Robot Learning.
As principal investigator, he has led numerous major research projects, including China’s National Key Research and Development Program, the National Science Fund for Distinguished Young Scholars, the National Science Fund for Excellent Young Scholars, and key projects under the Joint Fund of the National Natural Science Foundation of China. His honors include the Baosteel Outstanding Teacher Award, the Shanghai Rising-Star Program for Young Scientists, and the Shanghai Shuguang Program.
Professor Wang has served as General Chair of the 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025), General Chair of the 2016 International Conference on Real-time Computing and Robotics (RCAR 2016) and the 2022 IEEE International Conference on Robotics and Biomimetics (ROBIO 2022).
