Pixels to Proofs: Probabilistically-Safe Latent World Model Control via Parallel Conformal Robust MPC
arXiv:2606.15594v1 Announce Type: cross Abstract: We present SLS^2, a framework for safe feedback motion planning from pixels using robust model predictive control (MPC) in learned latent world models. Our approach trains an action-conditioned joint-embedding world model with compact Markovian latent states, enabling efficient gradient-based trajectory optimization through learned latent dynamics. To enforce safety for the true system despite imperfect latent predictions, we inform a GPU-accelerated system level synthesis (SLS) robust MPC scheme with conformal prediction to obtain calibrated l
This research addresses a critical limitation in AI-driven control systems, safety, which is essential for real-world deployment, especially as AI models become more complex and autonomous.
This paper offers a framework for safe feedback motion planning from pixels, a significant step towards developing robust and reliable AI agents capable of operating safely in complex physical environments.
The ability to enforce probabilistic safety guarantees within latent world model control, even with imperfect predictions, significantly lowers the barrier for deploying AI in safety-critical applications.
- · AI developers
- · Robotics industry
- · Automotive industry
- · Logistics sector
- · Companies relying on traditional and less reliable control systems in hazardous
Further development and adoption of AI-controlled systems in hazardous or complex environments will accelerate.
Reduced need for direct human supervision in certain industrial or robotic applications, increasing efficiency and automation.
The development of truly general-purpose AI agents capable of robustly and safely performing a wide range of tasks will become more feasible.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI