Summary
Simulating how multiple vehicles interact over several seconds is hard because decisions happen at different speeds and things keep changing. The authors propose a system called HI-FLOOP that breaks down the problem into smaller steps happening at different times, like long-term goals, short previews, and immediate motions. Their system keeps multiple possible futures in mind but only commits to what actually happens every half second, helping it stay consistent and realistic. They also use a special method to fix mistakes quickly by switching between models handling shorter time frames. This approach improves how well simulations predict real traffic behavior over eight seconds.
multi-agent traffic simulationstate-feedback loopsmulti-timescale modelingclosed-loop generationhierarchical modelingtrajectory predictionbranch consistencyoracle-minADEmotion planninginteraction modeling
Abstract
Multi-agent traffic simulation seeks diverse, coordinated, and physically realistic futures from maps and observed history. Long-horizon closed-loop generation must reconcile multiple decision time scales while its context evolves with generated states. Existing methods often unfold long futures from the initial scene and resolve intent, interaction, and motion monolithically, weakening cross-scale consistency and adaptation. We present HI-FLOOP, a branch-consistent multi-timescale state-feedback framework. Eight scene-level Worlds represent joint hypotheses, and all agents share the selected World identity throughout an 8-second rollout. Within the branch, an 8-second Goal anchors intent, a 2-second Preview coordinates interactions, and 1-second Control produces physical motion. Every 0.5-second commit feeds back only its executed prefix as new facts, while unexecuted hypotheses never enter factual memory. Joint Preview Interaction (JPI) induces a sparse directed future graph from Preview and uses conflict probabilities and signed arrival-time differences to gate interaction refinement. For generated-state recovery, a prefix-frozen A-to-B cascade lets frozen Model A generate 0-1 seconds, then transfers typed physical state, admissible context, and the branch index, but no latent state, to an independent Model B for re-encoding and 1-2-second recovery. On the full H-D public-validation split of 955 scenarios, one complete S1 run yields an Overall score of 0.689987 with the official evaluator. Under agent-centric oracle evaluation, HI-FLOOP achieves oracle-minADE@8 of 1.196636 m over the 8-second horizon and 0.526 m over the 6-second horizon.