
天轶2.0
Using a robotic arm and gripper, and leveraging the VLA end-to-end large model, the robot can automatically recognize elevators and complete autonomous elevator riding through operations such as button pressing and card swiping, without requiring any equipment modifications. With the chassis movement, the robot can autonomously perform common daily tasks such as carrying objects, pressing buttons, and opening/closing cabinet doors.
For patrol and inspection, the wheeled-arm robot integrates multimodal perception and embodied manipulation capabilities, suitable for scenarios such as office premises and factory sites. It features visible-light and thermal-infrared recognition, and can customize identification of arbitrary items on the fly through a large model. For guided tours and conversations, it provides reception leading and guided explanation services tailored to customer needs, with low-latency voice interaction. It supports generalized Q&A based on the customer's knowledge base, with second‑level response, interruptibility, and follow‑up queries. An emotional large model recognizes user actions and expressions, proactively initiates interaction, and coordinates whole‑body motion responses. One robot, multiple uses — the humanoid form offers higher versatility. Based on the same hardware, different software modules can be selected to achieve various functions, improving overall utilization. For example, "patrol at night, reception during the day." For remote operation, the robot chassis and arms can be controlled via VR devices or a teleoperation cockpit to perform various fine manipulations. A single operator can take over multiple robots, intervene in decision‑making during abnormal events during patrols, or replace human labor in high‑risk environments in special industries, ensuring worker safety.
Get weekly updates on humanoid robot funding, new companies, and product launches.