AI and Robotics Research Lab
Embodied intelligence, as a cutting-edge field that deeply integrates artificial intelligence and robotics, represents a crucial direction for the future development of intelligent systems. The AI and Robotics Research Lab is equipped with a series of advanced platforms consisting of robots (such as seven-axis collaborative robotic arms and mobile robots), vision perception systems, speech interaction modules, and a central computing unit.
Functionally, by deploying state-of-the-art large models such as DeepSeek within the central computing unit, and integrating technologies like force control feedback and human-computer interaction, the lab’s systems are capable of understanding complex instructions, performing logical reasoning, interpreting user intent, and autonomously planning and executing tasks. This marks a fundamental distinction between embodied intelligent robots and traditional robotic systems.
The AI and Robotics Research Lab not only supports research and development activities but also provides innovative solutions for a wide range of applications, including intelligent manufacturing, healthcare services, and commercial services.
Explore humanoid robots, mobile collaborators, and seven-axis arms
Humanoid & Intelligent Robotics Research
Embodied Intelligence & Autonomous Decision-Making
With self-developed seven-axis humanoid arms, integrated with vision, speech, and sensor technologies, the system builds a perception–decision–execution loop. Large models enable robots to understand complex instructions, perform reasoning, and autonomously plan and execute tasks.
Bimanual Coordination & Control
Featuring a “one-controller-two-arms” system with reinforcement learning, the platform supports precise bimanual coordination and conflict avoidance, simulating natural human movements. With electronic skin and large-model decision-making, it achieves intelligent motion planning and seamless human-robot collaboration.
Reinforcement Learning & Self-Evolution
By applying hierarchical reinforcement learning and cross-domain transfer learning, robots continuously adapt to complex environments and optimize strategies. Through ongoing interaction, they evolve autonomously and move closer to becoming universal intelligent agents.
Multimodal Perception & Fusion
Equipped with multimodal sensors including depth vision, speech recognition, LiDAR, and electronic skin, the system forms a comprehensive perception network. It supports natural language parsing, speech interaction, and gesture recognition, enabling intuitive human-robot interaction and AI applications.
Future Lab Demonstrations
