tech
Robostral Navigate: Single-Camera AI Navigation
Robostral Navigate is an 8B model that enables robots to autonomously navigate complex environments using only a single RGB camera, achieving 76.6% success on unseen R2R-CE benchmarks—outperforming multi-sensor approaches while being more efficient. Built entirely in-house with simulated data and token-efficient techniques, it generalizes across robot types and adapts to real-world obstacles unseen during training. The model combines pointing-based navigation with reinforcement learning for continuous improvement, paving the way for unified embodied AI in robotics.
TL;DR
- Achieves 76.6% success rate on unseen R2R-CE benchmarks, outperforming multi-sensor systems.
- Operates using only a single RGB camera, without LiDAR or depth sensors.
- An 8B model, trained entirely in simulation using token-efficient techniques.
- Generalizes across wheeled, legged, and flying robots and various sizes.
- Combines pointing-based navigation with reinforcement learning for continuous improvement.
- Built in-house, not relying on existing open-source VLMs.
- Trained on approximately 400,000 trajectories across 6,000 simulated scenes.
- Utilizes a token-efficient training algorithm based on prefix-caching, reducing training time significantly.
- Further improved by online reinforcement learning (CISPO) to handle failures and adapt to changing environments.