Eyes All Around: Design and Analysis of 360-Degree LiDAR Perception Using Equivariant Feature Learning in Unstructured Traffic
Researchers have developed a 360-degree LiDAR perception system designed for autonomous driving in complex, unstructured urban environments. The system utilizes panoramic processing and rotation-equivariant sparse convolutions to handle the challenges of dense traffic and varied road users. Evaluations on a custom dataset from Indian urban conditions demonstrated strong performance for vehicles, though detection of smaller entities like pedestrians and cyclists remained more difficult. AI
IMPACT This research could improve the reliability of autonomous vehicles in complex urban settings.