This Single-Lens Camera Can See Depth Like Two Eyes

Researchers at Spanish-Basque technology center IKERLAN are developing an unusual neuromorphic vision system that can detect 3D depth and track motion using only a single lens. The project is called BEGI, and its architecture is inspired by the compound eyes of insects. Instead of using two conventional cameras for stereoscopic vision or actively sending out laser pulses like LiDAR, BEGI is a passive event-based vision system. It detects changes in incoming light and processes only the visual information that is actually changing. The current prototype combines a microlens array with a Sony–Prophesee IMX636 dynamic vision sensor. Each pixel operates independently and generates an event only when it detects a change in brightness. The microlens array also provides information about the direction of incoming light, allowing BEGI to calculate depth, 3D structure and optical flow without requiring a second camera. The IMX636 has a resolution of 1280 × 720 pixels, latency below 100 microseconds at 1,000 lux and a dynamic range exceeding 120 dB under specified conditions. Prophesee lists maximum power consumption of approximately 205 mW for the sensor. That makes event-based vision particularly interesting for Physical AI applications such as robots, drones and autonomous machines, where reaction speed and energy consumption are critical. But the current BEGI prototype has an important limitation: resolution. To calculate depth, multiple pixels need to observe the same point from slightly different angles. With the current commercial sensor, the practical prototype is limited to less than one meter of depth range and approximately a 20–40 degree field of view. That makes the first version most suitable for short-range applications. IKERLAN is exploring industrial robotics, moving machinery and spacecraft docking, where the final approach happens at very short distances and extremely fast detection of small movements can be valuable. The next step is much more ambitious. IKERLAN plans to move beyond the current FPGA-based prototype toward dedicated silicon and a larger, higher-resolution sensor. The broader SiliconBurmuin initiative is developing BEGI as neuromorphic IP that can eventually be implemented in optimized ASIC hardware. According to the development team, a full-frame high-resolution version could theoretically extend depth perception to around 16 meters while increasing the field of view to approximately 120 degrees. That could open the technology to mobile robots, drones, autonomous navigation and other Physical AI systems. Instead of giving machines more cameras and more data, BEGI takes the opposite approach: capture only the information that matters. Could insect-inspired vision become the next major step for robotics? Subscribe to Xenon Lab News for more AI, robotics and future technology breakthroughs. #BEGI #IKERLAN #NeuromorphicVision #EventCamera #PhysicalAI #Robotics #RobotVision #ComputerVision #SonyIMX636 #Prophesee #3DVision #DepthSensing #EdgeAI #Drones #AutonomousSystems #SpaceTech #FutureTechnology #TechNews #XenonLabNews