title: "Robostral Navigate : navigation IA à caméra unique | Mistral AI"

Robostral Navigate is an 8-billion parameter model that enables robots to navigate complex environments using only a single RGB camera and natural language instructions. It achieves a 76.6% success rate on the R2R-CE benchmark, surpassing multi-sensor approaches without requiring LiDAR or depth sensors. The model was developed entirely in-house and trained on simulated data using efficient methods like prefix-caching and online reinforcement learning. This approach allows the model to generalize across various robot types and adapt to real-world obstacles not encountered during its training.