Exploiting map information for self-supervised learning in motion forecasting

Abstract

Inspired by recent developments regarding the application of self-supervised learning (SSL), we devise an auxiliary task for trajectory prediction that takes advantage of map-only information such as graph connectivity with the intent of improving map comprehension and generalization. We apply this auxiliary task through two frameworks - multitasking and pretraining. In either framework we observe significant improvement of our baseline in metrics such as minFDE_6 (as much as 20.3%) and MissRate_6 (as much as 33.3%), as well as a richer comprehension of map features demonstrated by different training configurations. The results obtained were consistent in all three data sets used for experiments: Argoverse, Interaction and NuScenes. We also submit our new pretrained model's results to the Interaction challenge and achieve 1st place with respect to minFDE_6 and minADE_6.

Tasks

Reproductions