Abstract
Existing latent world models are typically optimized for future predictability, yet the resulting representations are not necessarily useful for planning in autonomous driving. Predictions are commonly used for pretraining or auxiliary supervision rather than as direct conditioning signals for trajectory generation. We propose ForeDrive, which learns a planning-relevant latent representation and couples it asymmetrically to a Diffusion Transformer (DiT) planner. The planner consumes multi-horizon latent future representations learned with a JEPA-style world model; planning gradients update the shared online encoder, while stop-gradient routing trains the latent predictor with forecasting losses only. Because predicted futures have varying reliability across horizons and BEV trajectories are misaligned with image tokens, we use gated visual fusion, future-status injection, and Trajectory-Adaptive Bias (TAB) to inject future latents as guidance without overriding the current observation. Trained with pure imitation learning and using only the current front-view image as visual input at inference, ForeDrive attains 89.9 PDMS on NAVSIM v1 and 90.0 one-stage EPDMS on NAVSIM v2, without reinforcement learning or an external trajectory scorer.
Keywords
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Wang, S., Gu, Z., Huang, Y., Wen, W., Yang, X., Zhang, Y., Zhang, X., Che, N., Ling, J., Yu, Q., Liu, W., Xu, J., & Wang, X. (2026). ForeDrive: Foresight-Guided End-to-End Autonomous Driving with a Planning-Relevant Latent World Model. https://omanscience.com/en/articles/foredrive-foresight-guided-end-to-end-autonomous-driving-with-a-planning-relevant-latent-world-model
MLA 9
Wang, Sinuo, et al. "ForeDrive: Foresight-Guided End-to-End Autonomous Driving with a Planning-Relevant Latent World Model." https://omanscience.com/en/articles/foredrive-foresight-guided-end-to-end-autonomous-driving-with-a-planning-relevant-latent-world-model.
Chicago (author–date)
Wang, Sinuo, Zichong Gu, Yuhan Huang, Wenxin Wen, Xun Yang, Yiqing Zhang, Xingyu Zhang, Ningyu Che, Jie Ling, Qiankun Yu, Wei Liu, Jing Xu, and Xinggang Wang. 2026. "ForeDrive: Foresight-Guided End-to-End Autonomous Driving with a Planning-Relevant Latent World Model." https://omanscience.com/en/articles/foredrive-foresight-guided-end-to-end-autonomous-driving-with-a-planning-relevant-latent-world-model.
Harvard
Wang, S., Gu, Z., Huang, Y., Wen, W., Yang, X., Zhang, Y., Zhang, X., Che, N., Ling, J., Yu, Q., Liu, W., Xu, J. and Wang, X. (2026) 'ForeDrive: Foresight-Guided End-to-End Autonomous Driving with a Planning-Relevant Latent World Model', Available at: https://omanscience.com/en/articles/foredrive-foresight-guided-end-to-end-autonomous-driving-with-a-planning-relevant-latent-world-model.
Vancouver
Wang S, Gu Z, Huang Y, Wen W, Yang X, Zhang Y, et al. ForeDrive: Foresight-Guided End-to-End Autonomous Driving with a Planning-Relevant Latent World Model. https://omanscience.com/en/articles/foredrive-foresight-guided-end-to-end-autonomous-driving-with-a-planning-relevant-latent-world-model
IEEE
S. Wang, Z. Gu, Y. Huang, W. Wen, X. Yang, Y. Zhang, X. Zhang, N. Che, J. Ling, Q. Yu, W. Liu, J. Xu, and X. Wang, "ForeDrive: Foresight-Guided End-to-End Autonomous Driving with a Planning-Relevant Latent World Model," https://omanscience.com/en/articles/foredrive-foresight-guided-end-to-end-autonomous-driving-with-a-planning-relevant-latent-world-model.