Latent-WAM: End-to-End Autonomous Driving via Spatial-Aware Latent World Models

Summary

Latent-WAM is an efficient end-to-end autonomous driving framework introduced in an arXiv paper (2603.24581) that achieves strong trajectory planning through spatially-aware and dynamics-informed latent world representations. According to the authors, existing world-model-based planners suffer from three main shortcomings: inadequately compressed representations, limited spatial understanding, and underutilized temporal dynamics. Latent-WAM addresses these issues by building a latent world representation that encodes spatial awareness and dynamic information, enabling robust trajectory planning. The work falls within the computer vision (CV) research area, with authors including Linbo Wang, Yupeng Zheng, Qiang Chen, Shiwei Li, and Yichen Zhang. The paper was listed on arXiv on March 25, 2026, and the forum post was automatically collected on March 27, 2026.