New query-driven digital twin cuts AV planning errors by 24%

A team led by Nuocheng Yang at Beijing University of Posts and Telecommunications, in collaboration with the Singapore University of Technology and Design, has introduced a novel digital twin design for autonomous driving that flips the traditional data-sharing model. Conventional digital twins rely on constant, full-state synchronization from vehicles, which leads to high computational and communication loads from redundant data. The proposed query-driven architecture empowers the digital twin to proactively request only the environment data it deems necessary based on its simulation results, effectively filtering out irrelevant updates.

The researchers also formulated an optimization problem balancing planning position error with digital twin fidelity and communication constraints. To further enhance efficiency, they designed a cross-time-step progressive query mechanism that gradually refines data requests over successive simulation steps. Simulated evaluations demonstrate that the method achieves a 24% reduction in planning position error while simultaneously cutting communication overhead by 40% compared to traditional full-sync approaches. The work, published on arXiv (2606.28384), highlights a practical path toward high-fidelity, communication-efficient digital twins for autonomous driving and low-altitude economy scenarios.