We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Robotics

Title: Anomaly Detection in Multi-Agent Trajectories for Automated Driving

Abstract: Human drivers can recognise fast abnormal driving situations to avoid accidents. Similar to humans, automated vehicles are supposed to perform anomaly detection. In this work, we propose the spatio-temporal graph auto-encoder for learning normal driving behaviours. Our innovation is the ability to jointly learn multiple trajectories of a dynamic number of agents. To perform anomaly detection, we first estimate a density function of the learned trajectory feature representation and then detect anomalies in low-density regions. Due to the lack of multi-agent trajectory datasets for anomaly detection in automated driving, we introduce our dataset using a driving simulator for normal and abnormal manoeuvres. Our evaluations show that our approach learns the relation between different agents and delivers promising results compared to the related works. The code, simulation and the dataset are publicly available on this https URL
Comments: 15 pages incl. supplementary material, 8 figures, 4 tables (accepted by CoRL 2021)
Subjects: Robotics (cs.RO); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
Cite as: arXiv:2110.07922 [cs.RO]
  (or arXiv:2110.07922v2 [cs.RO] for this version)

Submission history

From: Julian Wiederer [view email]
[v1] Fri, 15 Oct 2021 08:07:31 GMT (1223kb,D)
[v2] Thu, 28 Oct 2021 07:23:14 GMT (1222kb,D)

Link back to: arXiv, form interface, contact.