ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2210.04672
42
9

Exploiting map information for self-supervised learning in motion forecasting

10 October 2022
Caio Azevedo
Thomas Gilles
S. Sabatini
D. Tsishkou
    SSL
ArXivPDFHTML
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 minFDE6\mathrm{minFDE}_6minFDE6​ (as much as 20.3%) and MissRate6\mathrm{MissRate}_6MissRate6​ (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\textit{1st}1st place with respect to minFDE6\mathrm{minFDE}_6minFDE6​ and minADE6\mathrm{minADE}_6minADE6​.

View on arXiv
Comments on this paper