Search papers, labs, and topics across Lattice.
The paper introduces MindDrive, an end-to-end autonomous driving framework that combines trajectory generation and selection through a structured reasoning paradigm. MindDrive uses a Future-aware Trajectory Generator (FaTG) based on a World Action Model (WaM) to simulate future scenes and generate trajectory candidates. A VLM-oriented Evaluator (VLoE) then leverages a vision-language model to perform multi-objective evaluations, improving safety, compliance, and generalization.
By fusing a world model with a vision-language model, MindDrive achieves state-of-the-art autonomous driving performance, suggesting a promising path toward interpretable and cognitively guided systems.
End-to-End autonomous driving (E2E-AD) has emerged as a new paradigm, where trajectory planning plays a crucial role. Existing studies mainly follow two directions: trajectory generation oriented, which focuses on producing high-quality trajectories with simple decision mechanisms, and trajectory selection oriented, which performs multi-dimensional evaluation to select the best trajectory yet lacks sufficient generative capability. In this work, we propose MindDrive, a harmonized framework that integrates high-quality trajectory generation with comprehensive decision reasoning. It establishes a structured reasoning paradigm of"context simulation - candidate generation - multi-objective trade-off". In particular, the proposed Future-aware Trajectory Generator (FaTG), based on a World Action Model (WaM), performs ego-conditioned"what-if"simulations to predict potential future scenes and generate foresighted trajectory candidates. Building upon this, the VLM-oriented Evaluator (VLoE) leverages the reasoning capability of a large vision-language model to conduct multi-objective evaluations across safety, comfort, and efficiency dimensions, leading to reasoned and human-aligned decision making. Extensive experiments on the NAVSIM-v1 and NAVSIM-v2 benchmarks demonstrate that MindDrive achieves state-of-the-art performance across multi-dimensional driving metrics, significantly enhancing safety, compliance, and generalization. This work provides a promising path toward interpretable and cognitively guided autonomous driving.
