Search papers, labs, and topics across Lattice.
SteinsGateDrive decouples LLM planning from real-time control in autonomous driving by pre-computing counterfactual driving futures using a world model and LLM. The system generates three types of world-lines: nominal, interaction counterfactuals, and hazard-stress futures, which are then selected and validated at runtime. Experiments with GPT-5.4 mini on a highway driving protocol demonstrate a reduction in effective lag from +3.07s to -0.01s with a 4-second horizon, while maintaining collision safety.
Achieve near-zero effective latency in LLM-driven autonomous driving by decoupling planning and control, enabling real-time reactions to hazards using pre-computed, safety-validated futures.
Cloud-hosted LLM driver agents provide useful semantic judgments, but their inference latency exceeds stepwise vehicle-control windows. Learned world models predict futures, but they usually keep future generation and action selection inside large coupled loops. We present SteinsGateDrive, a latency-decoupled planner-runtime architecture in which the worldline metaphor from the eponymous story names one plausible consequence of an intervention: the LLM selects counterfactual driving futures before the final control instant, and a runtime reuses the selected forecast only while safety contracts remain valid. The generator builds three world-line roles: alpha nominal ego-conditioned futures, beta interaction counterfactuals around nearby vehicles, and gamma hazard-stress futures such as braking, cut-ins, or blocked corridors. The selected branch becomes a typed StrategicForecast with horizon, validity/abort conditions, fallback, and authority. On a within-subject, matched-seed normal-highway protocol with 10 seeds and 20 steps, GPT-5.4 mini reduces effective lag from +3.07 s at 1-second horizon to -0.01 s at 4-second horizon while preserving the measured no-collision safety boundary. The architecture's safety contribution comes from the atom-predicate runtime check, not from the drift score, which functions as a refresh-frequency knob.
