Benchmark Radar
AI BENCHMARK PROFILE

ExPhy

General AIKnowledge & ReasoningBeijing University of Posts and Telecommunications

24,000 simulated scenes for multi-object trajectory forecasting with explicit labels for mass, friction, and restitution, including ID and OOD splits over physical parameters and initial states.

Released
2026-08-20
Readiness
Runnable
Primary field
General AI

Why it matters

Enables joint evaluation of trajectory forecasting and physical property estimation, revealing that accurate forecasts do not guarantee correct physical understanding.

Motivation

Understanding object dynamics requires not only predicting future trajectories but also examining whether a model captures the physical properties that govern motion.

Primary resources

Benchmark Radar records only publicly supported details and links back to primary sources for verification.