Our research in in-sensor computing for artificial robot vision lies at the forefront of integrating advanced sensor architectures with embedded computational capabilities to enhance robotic perception. For example, the development of an anti-distortion bioinspired camera featuring an inhomogeneous photo-pixel array (^(M)Nature Communications, 2024) introduces innovative strategies for mitigating image distortion in complex environments. In addition, a curved neuromorphic image sensor array based on a MoS₂-organic heterostructure (^(ⅰ)Nature Communications, 2020) demonstrates how bio-inspired design can optimize signal processing, reduce computational overhead, and improve the efficiency of semantic segmentation (^(ⅱ)Science Advances, 2025). Complementing these advances, we also integrated synaptic phototransistors with quantum dot light-emitting diodes (^(ⅲ)Science Advances, 2022), showcasing a system capable of simultaneous visualization and recognition of UV patterns. Beyond vision systems, robotic hands are another key element in humanoid robots. Collectively, these breakthroughs underscore our commitment to pioneering next-generation human-friendly robotic systems.