List your company
ResearchFormerly WizWand

Multi-agent motion simulation on INTERACTION dataset (test)

0RMSE (m)Ground Truth
-0.92685.329111.58517.8409Dec 5, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Ground TruthObservation Type=N/A, Training Dataset=None (recorded data)2025.1204712
OursObservation Type=instance-centric, Number of Parameters=429K, Training Dataset=DeepScenario2025.1216.6524963
Ours (small)Observation Type=instance-centric, Number of Parameters=59K, Training Dataset=DeepScenario2025.1217.84249115
GraphAIRLObservation Type=agent-centric, Number of Parameters=145K, Training Dataset=DeepScenario, Target Reward Training=true2025.1219.711,026209
BCObservation Type=instance-centric, Number of Parameters=429K, Training Dataset=DeepScenario2025.1220.191,3511,968
LateFusionMLPObservation Type=agent-centric, Number of Parameters=104K, Training Dataset=DeepScenario, Target Reward Training=true2025.1220.251,348899
GraphAIRLObservation Type=agent-centric, Number of Parameters=145K, Training Dataset=DeepScenario, Target Reward Training=false2025.1220.471,110537
CVObservation Type=N/A, Training Dataset=N/A (Learning-free)2025.1223.172,8232,774