Leju and Hunan University Establish Joint R&D Center for Industrial Embodied AI
Leju Robotics and Hunan University have officially inaugurated a joint research and development center for industrial embodied AI. The center's establishment marks a new stage in their collaboration, aiming to advance the application of robotics in industrial settings. Embodied AI refers to artificial intelligence systems that interact with the physical world through a robotic body, and industrial embodied AI specifically targets manufacturing environments.
The center will focus on real-world industrial needs, prioritizing joint research in industrial embodied AI technologies and visual perception large models. The goal is to accelerate the deployment of embodied intelligent robots in production lines, enhancing manufacturing intelligence and efficiency. By combining Leju's engineering capabilities with Hunan University's research strengths, the center seeks to address key technical challenges.
The center has outlined three major research directions. The first is wide-area multimodal active perception for embodied robots, targeting large-space industrial scenarios. It aims to overcome challenges in precise perception and dynamic adaptation in complex environments, improving robots' operational accuracy and environmental robustness. This involves developing sensors and algorithms that can fuse data from multiple modalities, such as vision, lidar, and touch, to enable robots to navigate and manipulate objects in cluttered spaces.
The second direction is industrial embodied AI large models, integrating task planning, imitation learning, reinforcement learning, and tactile sensing. This aims to break through bottlenecks such as weak autonomous decision-making, slow response, limited task variety, and low operation success rates. By leveraging large-scale pre-trained models and fine-tuning them with industrial data, the center hopes to create AI systems that can understand complex tasks and adapt to new situations.
The third direction involves developing robots with dexterous motion and fine manipulation capabilities for both home and industrial manufacturing scenarios. This includes advancing the commercialization of typical applications like industrial sorting, where robots need to pick and place objects of various shapes and sizes. The center will work on innovative gripper designs and control algorithms to achieve human-like dexterity.