Bio-inspired artificial vision draws inspiration from the remarkable capabilities of natural vision systems. A key advantage of the bio-inspired artificial vision lies in its capability to address challenging conditions in image acquisition and processing—scenarios where conventional vision systems may struggle with. By leveraging the strategies employed by natural vision systems, novel imaging systems that can acquire image data more compatible for AI-based processing and target recognition can be developed. Specifically, we have developed a range of AI-native imaging systems inspired by various natural visions, including human (Nature Communications, 2017 ; Nature Nanotechnology, 2022), fish (Nature Electronics, 2020), cuttlefish (Science Robotics, 2023), fibber crab (Nature Electronics, 2022), bird (Science Robotics, 2024), and cat (Science Advances, 2024). In parallel, we engineered covalent heterostructures of ultrathin amorphous carbon nitride and silicon, which have enabled high-performance vertical photodiodes (Nature Synthesis, 2025) for ultrathin image sensing systems. These artificial vision systems can be integrated with mobile robots such as humanoid robots, submarines, unmanned vehicles, and drones to minimize computation load and time in edge-level image processing.