Wang Xingxing: Physical AI’s Core Challenge Is Aligning Models With the Physical World

According to ChainCatcher, Unitree Robotics founder and Chairman Wang Xingxing said at the ninth Hongqiao International Economic Forum that physical AI’s main challenge is aligning models with the physical world. He said multimodal models already have relatively strong understanding capabilities, but errors remain between sensor measurements and real-world states at the input stage, and between action commands and actual movements at the output stage. Wang expects these issues could be resolved within one to two years.