STRIDE: Evaluating Spatiotemporal Reasoning in Driving Edge Cases

A new benchmark, STRIDE (SpatioTemporal Reasoning In Driving Edge Cases), finds that even the strongest of 12 proprietary, open-weight and driving-specialized vision-language models, GPT-6-Astra, reaches only 43.4% average multiple-choice accuracy on long-tail, safety-critical driving scenarios, against an approximately 20% random-choice baseline. STRIDE spans six task families — spatial perception, spatial understanding, temporal memory, temporal extrapolation, trajectory prediction and scene-context awareness — pairing multiple-choice questions with free-form visual question answering. The benchmark's authors report that strong scene description does not translate into accurate safety-critical judgments, and that models struggle more with reconstructing object motion than recalling past positions.

arXiv:2610.22795v1 Announce Type: cross Abstract: In autonomous driving, perceiving surrounding objects is not enough: failing to understand how they move and interact over time can lead to dangerous decisions. Evaluating vision-language models VLMs therefore requires testing whether they connect past spatial observations to future consequences. Yet existing benchmarks often assess edge cases through basic perception questions or test temporal understanding in simple scenarios, leaving complex spatiotemporal reasoning underexplored. We introduce STRIDE SpatioTemporal Reasoning In Driving Edge Cases , an object-centric benchmark for long-tail, safety-critical driving scenarios. STRIDE covers six task families: spatial perception, spatial understanding, temporal memory, temporal extrapolation, trajectory prediction, and scene-context awareness. It pairs multiple-choice questions probing individual spatiotemporal judgments with related free-form visual question answering VQA requiring detailed scene explanations and justifications of driving decisions. We prioritize scenes with unusual object behaviors and complex traffic interactions, generate reference answers from structured dataset annotations through automated procedures, and use limited language-model assistance and human verification to ensure annotation quality. Across 12 proprietary, open-weight, and driving-specialized models, even the best performer, GPT-6-Astra, achieves only 43.4% average multiple-choice accuracy, versus an approximately 20% random-choice baseline. Further analysis reveals that strong performance in describing driving scenes does not necessarily translate to accurate safety-critical judgments, and that models struggle more with reconstructing object motion than recalling past positions. STRIDE thus exposes spatiotemporal driving limitations that aggregate VQA scores can conceal.