Technical Paper
A Mapless Trajectory Prediction Model with Enhanced Temporal Modeling
2024-04-09
2024-01-2874
The prediction of agents' future trajectory is a essential task in supporting advanced driver-assistance systems (ADAS). It plays an even more crucial role in making safe decisions for autonomous driving (AD). The current prevalent trajectory prediction methods heavily take high definition maps (HD maps) as the prior knowledge. Although the surrounding environment information provided by HD maps improves the accuracy of trajectory prediction, the high cost and legal restrictions of HD maps limit their widespread use. Moreover, due to object occlusion, limited field of view, and other reasons, the historical trajectory of the target agent is often incomplete. This limitation reduces the accuracy of trajectory prediction. Therefore, this paper proposes ETSA-Pred, a mapless trajectory prediction model with enhanced temporal modeling and spatial self attention.