UST030

82 R eliable, safe autonomous vehicles demand high- performance sensors, and developers of the next generation of Lidars for sense & avoid applications are working on increasing their range while reducing their size and cost, mostly for driverless car applications. However, the size and weight reductions are also becoming more appealing to unmanned aircraft manufacturers. There are three main approaches to developing a Lidar sensor, although some hybrid versions combine different techniques. The best-established technique uses a mirror to direct a number of laser beams. The light reflected from an object is then captured by a photodiode or avalanche diode, and the time of flight (ToF) of the light from the laser to the sensor gives the distance to the object. The design of the optics of both the transmitter and receiver are essential to achieve high performance. The mirror reflects all the light from the laser and so provides the longest range, currently up to 200 m. The output from the sensor is a point cloud that represents all the reflections detected at any point in time. The second technique is to use an optical waveguide to steer a laser beam without using moving parts. Changing the phase and frequency of the laser creates areas of constructive and destructive interference that form ‘beams’ at a particular plane in the distance. This optical beam-steering approach has no moving parts and the optics can be smaller, allowing a smaller sensor. The third technique is a ‘flash’ sensor, where the light is generated by a large array of small laser diodes to illuminate an area. This requires less sophisticated optics for the transmitter but more complex optics and processing at the receiver, as there is more information available from the array of light. However, flash sensors have challenges with range, accuracy, size, weight and cost, making them potentially less appealing than other methods. Solid-state Lidars have also been widely publicised as being available and ready at a small fraction of the cost of electromechanical systems. The systems that do exist though are generally limited in range, can suffer severely in accuracy (more than 5 cm), may be large and power-hungry and currently not ready to compete with the mechanical systems. Sensors based around micro- electromechanical system (MEMS) micro-mirrors generally sacrifice range and field of view, but have potential for use in the near future. As a result, mechanical Lidar systems still dominate the market. They have limits though on how small they can be, their manufacture can only be automated to a Improvements in Lidar sensors are making them more attractive for sense & avoid in the air as well as on the ground. Nick Flaherty explains the latest advances February/March 2020 | Unmanned Systems Technology No trouble ahead The latest 128-channel biaxial rotating mirror sensor can pick up details up to 200 m ahead of a vehicle (Courtesy of Velodyne)