To achieve this, our research operates in a dual-track parallel mode, with a simultaneous focus on AI chip design. Our work covers chip architectures optimized for demanding AI tasks and employs a strategy of software-hardware co-optimization. This integrated approach not only fundamentally enhances a robot system's intelligence and capabilities but also ensures we deliver high-performance, low-power AI solutions, thereby enabling comprehensive innovation from the foundational hardware to high-level applications.