AI-Powered Humanoid Robot Startup Emerges from Stealth
Mentee Robotics’ robot uses large-language models, machine learning and AI to give it locomotion, navigation and reasoning capabilities
AI robotics company Mentee Robotic is emerging from stealth with its contribution to the humanoid robotics race.
The company’s MenteeBot is designed with dexterity and reasoning capabilities for a “wide spectrum of activities” in domestic and industrial contexts. Its capabilities include autonomous navigation, locomotion, object detection and natural language understanding.
Established in 2022, Mentee’s founders include Amnon Shashua, co-founder of autonomous vehicle (AV) company Mobileye, director of Facebook AI research Lior Wold and machine learning expert Shai Shalev-Shwartz.
Their experience enables the company’s robot to leverage machine learning, 3D mapping and AI capabilities derived from AVs and advanced driver-assistance systems.
MenteeBot’s locomotion is based on simulation-to-reality machine learning technology. This platform uses reinforcement learning on a simulated version of the robot, collecting data on its response to situations for real-world training.
Neural network-based technologies are used for environment mapping, while transformer-based Large Language Models are used for interpreting commands and “thinking through” the required steps for completing a task.
The robot will also feature an array of cameras and electric motors, which the company said support “unprecedented dexterity and fully integrated AI.”
A production-ready prototype is expected to be ready for deployment by the first quarter of next year.
"We are on the cusp of a convergence of computer vision, natural language understanding, strong and detailed simulators, and methodologies on and for transferring from simulation to the real world,” said Shashua.
“At Mentee Robotics we see this convergence as the starting point for designing the future general-purpose bi-pedal robot that can move everywhere (as a human) with the brains to perform household tasks and learn through imitation tasks it was not previously trained for.”
About the Author
You May Also Like