Search papers, labs, and topics across Lattice.
The paper introduces Deep-Flow, an unsupervised anomaly detection framework for autonomous driving that uses Optimal Transport Conditional Flow Matching (OT-CFM) to model expert human driving behavior. By constraining the generative process to a low-rank spectral manifold via PCA and incorporating a lane-aware goal-conditioned Transformer, Deep-Flow achieves kinematic smoothness and stable log-likelihood estimation. Evaluated on the Waymo Open Motion Dataset, Deep-Flow demonstrates an AUC-ROC of 0.766 and identifies critical out-of-distribution behaviors missed by traditional safety filters.
Autonomous vehicles can now identify subtle, dangerous driving anomalies missed by traditional safety filters, thanks to a new method that models expert driving behavior in a low-dimensional spectral space.
Safety validation for Level 4 autonomous vehicles (AVs) is currently bottlenecked by the inability to scale the detection of rare, high-risk long-tail scenarios using traditional rule-based heuristics. We present Deep-Flow, an unsupervised framework for safety-critical anomaly detection that utilizes Optimal Transport Conditional Flow Matching (OT-CFM) to characterize the continuous probability density of expert human driving behavior. Unlike standard generative approaches that operate in unstable, high-dimensional coordinate spaces, Deep-Flow constrains the generative process to a low-rank spectral manifold via a Principal Component Analysis (PCA) bottleneck. This ensures kinematic smoothness by design and enables the computation of the exact Jacobian trace for numerically stable, deterministic log-likelihood estimation. To resolve multi-modal ambiguity at complex junctions, we utilize an Early Fusion Transformer encoder with lane-aware goal conditioning, featuring a direct skip-connection to the flow head to maintain intent-integrity throughout the network. We introduce a kinematic complexity weighting scheme that prioritizes high-energy maneuvers (quantified via path tortuosity and jerk) during the simulation-free training process. Evaluated on the Waymo Open Motion Dataset (WOMD), our framework achieves an AUC-ROC of 0.766 against a heuristic golden set of safety-critical events. More significantly, our analysis reveals a fundamental distinction between kinematic danger and semantic non-compliance. Deep-Flow identifies a critical predictability gap by surfacing out-of-distribution behaviors, such as lane-boundary violations and non-normative junction maneuvers, that traditional safety filters overlook. This work provides a mathematically rigorous foundation for defining statistical safety gates, enabling objective, data-driven validation for the safe deployment of autonomous fleets.
