Light Intensity-Driven Bidirectional Photoresponse Vision Sensor for Autonomous Obstacle Avoidance System

Bio-inspired vision sensors emulating neural-pathway processing hold significant promise for next-generation robotics and artificial intelligence. However, achieving biomimetic threat-distance adaptation, where escape initiation dynamically calibrates to looming object proximity within a single device as in insect neural circuits, remains challenging for bionic vision systems implementations. Herein, we present a vision sensor based on a 2D PVK/h-BN/MoS₂/h-BN/2D PVK heterostructure that achieves full dynamic emulation of insect phototactic/scototactic behaviors. The core innovation is symmetrical gate-field co-regulation, opposing gate biases on the top and bottom 2D PVK photosensitive layers trigger antagonistic field-effect modulation in response to light gradients. Low light intensity activates the top layer, inducing persistent positive photocurrent (PPC) via hole/cation accumulation, while high light intensity activates the bottom layer, generating negative photoconductivity (NPC) via electron/anion accumulation, mimicking adjacent ommatidial excitation/inhibition. Hopping-like ion transport enables ultra-long PPC/NPC persistence post-illumination. An autonomous obstacle avoidance system built with this sensor directly maps light-gradient signals to motor commands enabling voltage-tunable braking distance control via symmetric gate differential modulation, co-processing real-time intensity, historical accumulation, and rate-of-change for efficient collision avoidance in dynamic environments. This work provides a valuable reference scheme for bionic vision systems.