Reproduce “Trajeform: Embedding Crowd Data” 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

Marilena Lemonari, Panayiotis Charalambous, Yiorgos Chrysanthou

Trajeform evaluates crowd simulators by learning a shared representation from real pedestrian trajectories and projecting simulated tracks into it. It aligns trajectories to reduce sensitivity to starting position and orientation. The real-data distribution provides a reference for comparison. With a shared decoder, its mixture-of-Gaussians scores 0.008 MMD versus MPACT’s 0.029, but trails RVO and MPACT on Deviation From Goal.

Summary by alphaXiv.

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