Reproduce “SRA: Spatial Reasoning Adapter via Evolving Social Interaction Graphs for Trajectory Prediction” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Jaewoo Jeong, Seonkyu Song, Hyeonwoo Park, Jegyeong Cho, Yujin Bae, Giwon Lee, Daehee Park, Kuk-Jin Yoon

The authors introduce SRA, an adapter that updates interaction context during iterative trajectory prediction. At each denoising step, it builds a sparse graph from agents’ current predicted futures and uses uncertainty-weighted messages to correct the host predictor. Across three hosts and four datasets, SRA improves ADE in all 12 pairs; FDE improves in 11, with MID on SDD worsening.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.