Reproduce “Unsupervised Anomaly Detection in Multi-Agent Trajectory Prediction via Transformer-Based Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Qing Lyu, Zhe Fu, Alexandre Bayen
The authors present an unsupervised method to find unusual multi-vehicle interactions. A Transformer forecasts vehicle motion, and an Isolation Forest scores scenes where motion differs from predictions; the authors check ranking stability and safety-proxy correlations. On NGSIM US101, it identifies 388 anomalies unique to the method, but these detections are not confirmed hazards or evidence of generalization.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.