Towards Metric-Agnostic Trajectory Forecasting

2026-07-01

AI Summary

This work addresses the challenge in trajectory prediction where existing methods often tailor training objectives to specific evaluation metrics, leading to optimization conflicts and suboptimal performance across multiple metrics. To overcome this limitation, the authors propose a metric-agnostic prediction paradigm: first, a probabilistic prediction model based on the DONUT architecture—termed DONUT-NLL—is trained using negative log-likelihood (NLL) to learn a unified trajectory distribution; then, during inference, a newly introduced TraDiE strategy adaptively maps this distribution to generate K optimal trajectories along with their confidence scores, tailored to any given evaluation metric. By decoupling training from evaluation, the method achieves state-of-the-art performance across all metrics on the Waymo Motion Prediction Benchmark, significantly enhancing both generalizability and practical utility.

Abstract

Accurate trajectory forecasting of surrounding traffic participants is a core capability for autonomous driving, enabling vehicles to anticipate behavior and plan safe maneuvers. We observe that current state-of-the-art forecasting models on Argoverse 2 and the Waymo Open Motion Dataset tailor their training objectives to the different benchmark metrics. Because these metrics encourage conflicting behavior, we propose a paradigm change for trajectory forecasting: training models with metric-agnostic probabilistic objectives and treating metric optimization as a downstream task applied to the predictive distribution. Concretely, we introduce Trajectory Distribution Evaluation (TraDiE) policies, metric-specific policies that map a predictive distribution to the set of $K$ trajectories and confidences required by trajectory forecasting metrics. We evaluate this framework by introducing DONUT-NLL, which adapts the training objective of the state-of-the-art trajectory forecasting model DONUT to directly optimize the predictive distribution. Using our policies, DONUT-NLL achieves state-of-the-art results on all metrics of the Waymo motion prediction benchmark.

Problem

• trajectory forecasting
• metric-agnostic
• autonomous driving
• predictive distribution
• benchmark metrics

Innovation

• metric-agnostic
• trajectory forecasting
• probabilistic modeling
• TraDiE
• predictive distribution