Abstract
In this paper, we present LiRaNet, a novel end-to-end trajectory predictionmethod which utilizes radar sensor information along with widely used lidar andhigh definition (HD) maps. Automotive radar provides rich, complementaryinformation, allowing for longer range vehicle detection as well asinstantaneous radial velocity measurements. However, there are factors thatmake the fusion of lidar and radar information challenging, such as therelatively low angular resolution of radar measurements, their sparsity and thelack of exact time synchronization with lidar. To overcome these challenges, wepropose an efficient spatio-temporal radar feature extraction scheme whichachieves state-of-the-art performance on multiple large-scale datasets.Further,by incorporating radar information, we show a 52% reduction in prediction errorfor objects with high acceleration and a 16% reduction in prediction error forobjects at longer range.