Executive Summary : | Unmanned aerial vehicles or drones can be used for intrusion detection at borders, delivery of medicine/aid to remote areas, volumetric mapping of mines, crop-monitoring in agriculture etc. Unfortunately, low-cost drones require manual control making them unreliable for long-distance use (5km) without line-of-sight communication thereby necessitating on-board ranging devices for obstacle detection and self-navigation.
Among the available ranging devices, stereo cameras are inaccurate, need image contrast and power hungry image processing. Ultrasound ranging on the other hand, has limited range (3m-5m) due to significant attenuation in air (1.6dB/(cm.MHz)) and it is prone to ambient temperature variations and vibrations from high-speed drone propellers. Therefore, current fully autonomous drones use accurate yet expensive LiDARs (light detection and ranging) for collision avoidance.
LiDARs use light for ranging similar to radio signals by radars. The micro-meter wavelength of optical signals (~0.4um-1.5um) enables accurate detection of small objects within the field-of-view (FoV). Furthermore, unlike radars, use of lasers with narrow beam divergence (~0.1deg) helps distinguish close-by objects in the scene. For example, Velodyne's VLP-16 LiDAR (~0.1% accuracy at 100m) costs ~₹2.5 lacs each, prohibiting its ubiquitous use. Our goal is to develop an indigenous low-cost pulsed LiDAR for drones as explained in this proposal. |