The hot industry topic of lidar in autonomous vehicles was high on the agenda on Saturday, with James Jung from AEye giving an invited talk on optical MEMS devices in high-performance systems. Jung, director of MEMS technology at the venture-backed San Francisco startup, avoided revealing specific details of AEye’s approach, instead discussing the advantages of MEMS-based lidar. The “point-scan” approach MEMS supports offers significantly better range than alternative approaches like flash or line-scan lidar, although that can lead to enormous data demands to operate with the kind of low computing latency necessary for safe driving at speed. More obvious advantages incude the small form factor of MEMS and the ability to scale volumes for cheap production, although packaging still presents some challenges. AEye, which raised $40 million in a series B venture round in late 2018, is tackling that with its proprietary “iDAR” technology. It is based on scanning more intelligently, rather than simply using more photons and data points to produce more detailed point-clouds. Jung said that the approach has proved successful in a number of hazardous “edge case” driving scenarios where the agility of lidar is critical – for example identifying road debris, or spotting “hidden” pedestrians where they would be invisible to cameras or radar. He added that biaxial architectures, where the transmit and receive optical paths are independent and can be developed and designed in a modular fashion, suits the MEMS approach best. Based around 1550nm laser sources, which take advantage of more relaxed eye safety regulations than at 905nm to allow deployment of more photons, AEye’s lidar systems are said to achieve ranges in excess of 300 meters. Jung said that the company had focused on redefining the three “Rs” of lidar – rate, resolution, and range – to produce a technology capable of long-distance object classification combined with the low latency and resolution requirements needed for safe autonomous driving.