Conversation with RoboSense's Yang Xiansheng: Robot general intelligence starts with a pair of 'honest' eyes

▤ Summary

This article focuses on RoboSense's view of robot perception and the role of higher-quality 3D sensing in physical AI. It argues that robots still lack reliable vision for precise manipulation because current sensors struggle with depth accuracy, stability, and RGB alignment. The discussion highlights a new RGB-D sensing architecture that combines depth and color at the physical layer, reducing post-processing and lowering compute requirements. It also explains why better sensor data can improve training, simulation, real-time perception, and end-effector control for embodied AI systems.

▦ Linked Companies