Multi-LiDAR placement, calibration, co{registration and processing on a Subaru Forester for off-road autonomous vehicle operation

摘要

For autonomous vehicles 3D, rotating LiDAR sensors are often critically important towards the vehicle's abilityto sense its environment. Generally, these sensors scan their environment, using multiple laser beams to gatherinformation about the range and the intensity of the reection from an object. LiDAR capabilities have evolvedsuch that some autonomous systems employ multiple rotating LiDARs to gather greater amounts of data regardingthe vehicle's surroundings. For these multi{LiDAR systems, the placement of the sensors determine thedensity of the combined point cloud. We perform preliminary research regarding the optimal LiDAR placementstrategy on an o {road, autonomous vehicle known as the Halo project. We use the Mississippi State UniversityAutonomous Vehicle Simulator (MAVS) to generate large amounts of labeled LiDAR data that can be usedto train and evaluate a neural network used to process LiDAR data in the vehicle. The trained networks areevaluated and their performance metrics are then used to generalize the performance of the sensor pose. Datageneration, training, and evaluation, was performed iteratively to perform a parametric analysis of the e ectivenessof various LiDAR poses in the Multi{LiDAR system. We also, describe and evaluate intrinsic and extrinsiccalibration methods that are applied in the multi{LiDAR system. In conclusion we found that our simulationsare an e ective way to evaluate the e cacy of various LiDAR placements based on the performance of the neuralnetwork used to process that data and the density of the point cloud in areas of interest.