Reproduce “X-World 1.6: Building Controllable Virtual Cities with World Models” 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
XPeng, Boyang Wang, Boyi Sun, Chaoda Zheng, Dongbin Zhang, Guangfeng Jiang, Hanlong Chen, Hongbin Lin and 27 more
X-World 1.6 extends interactive driving simulation with reusable reference-anchored scenes and controllable scenarios. Separate reference-conditioned models use point clouds, 3D Gaussian Splatting scenes, or observations to constrain persistent surroundings. Connected Self Forcing trains generated histories with feedback to earlier computations. On evaluated cases, mean final displacement error fell 20.9% versus X-World 1.0; paired failure-reproduction reviews were improved or comparable in 96.4%, not an absolute success rate.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.