Automated Assessment of Infrastructure Preparedness for Autonomous Vehicles

Several highway design elements are presently determined based on constraints due to human drivers. Autonomous vehicles (AV) are expected to replace human drivers with an expectation of having differing driving characteristics. It is, therefore, imperative that current infrastructure is reassessed for AV scenarios. This paper proposes a raycasting approach to test autonomous vehicles (AV) in a virtual environment using voxelated LiDAR data. Sight distances and speed limits are mapped based on AV and road characteristics. An alternative algorithm was used in order to compare it with the proposed approach and was found to produce results agreeing within 1.94%. It was found that certain AV scenarios may lead to unsafe driving conditions for autonomous vehicles. Such information is crucial as it allows policymakers to design maps, guidelines, legislation, and countermeasures to prevent unsafe driverless vehicle scenarios.