Core Insights - The conference titled "Empowering New Energy, Driving the Future" focused on the transformation of scientific achievements and high-quality development in embodied intelligence, gathering over a hundred top young scientists and leaders from listed companies to discuss the future of intelligence [1] Group 1: Research Breakthroughs - The research led by Dong Huixu addresses industry pain points in unstructured environments such as factories, logistics, and homes, which are complex and dynamic [3] - Dong's team has made significant breakthroughs in dexterous hands, multi-modal perception, and embodied intelligence, which are crucial for robotic applications in these environments [3] Group 2: Applications of Dexterous Hands - The development of a single-motor six-degree-of-freedom hand has achieved seven times the load capacity of similar hands, enabling effective object recognition and grasping in logistics scenarios [5] - In agricultural applications, a flexible picking hand has been designed to harvest fruits in windy conditions, ensuring complete collection [5] - The team has also created a three-fingered dexterous hand for library applications, capable of complex grasping tasks with fewer fingers [5] - In traditional Chinese medicine, a massage robot hand has been developed to perform various massage techniques [5] - The team has innovated a variable stiffness hand for diverse everyday grasping tasks, addressing the need for versatility in object handling [5] - Multiple humanoid five-fingered dexterous hands have been developed for complex operations involving both flexible and rigid objects [5] Group 3: Multi-Modal Perception and Planning - Dong's team has developed a three-dimensional tactile sensor using light waveguides and visual-tactile methods, enhancing dexterous operations in dynamic scenes [7] - The integration of visual-tactile perception in robotic hands has overcome challenges in large-scale integration and object perception in three-dimensional dynamic environments [7] - The concept of "embodied cerebellum" has emerged, enabling high-precision assembly and optimized sequential grasping in constrained spaces through advanced learning networks [7] - The advancements in embodied intelligence are paving the way for practical applications in unstructured environments, indicating a clear path for future intelligent terminals [7]