Mistral AI Launches Robostral Navigate for Natural Language Robot Navigation
By replacing expensive LiDAR and sensor arrays with a single camera and language understanding, Robostral Navigate could lower the cost and complexity of autonomous robot navigation.
Reporting from 1 source: GIGAZINE.
French AI company Mistral AI has released Robostral Navigate, an 8-billion parameter navigation model that lets robots follow natural language commands using only a camera. The system can handle obstacles, adapt to different robot sizes, and supports wheeled, walking, and flying locomotion. Mistral claims higher navigation success rates than competing models.
A robot equipped with Robostral Navigate can be told 'Turn right before the locker, then turn left and go straight to the orange sofa' and plot a path using only its camera. The 8-billion parameter model processes RGB images alongside text instructions, automatically avoiding obstacles. Mistral AI says the system maintains accuracy even when the robot's size changes, and supports wheeled, walking, and flying locomotion. The company plans further research to enable autonomous navigation in offices and outdoor environments.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.