Flexion Robotics, a Swiss startup founded by former Nvidia robotics researchers, has developed an AI system that allows humanoid robots to perform complex tasks autonomously. The system uses simulation and reinforcement learning to teach robots to carry out chores such as retrieving parcels and organizing items. The company claims its approach is more efficient than traditional methods that rely on human teleoperation.
The AI system combines different AI models to enable robots to perform tasks in the real world. It uses videos of humans doing various activities to train the software, which then matches learned skills from simulation to real-world tasks. For instance, to reach a mail room, the model may learn to open doors and use elevators. The system also controls the robot’s motors to allow walking, limb movement, and balance maintenance.
According to Nikita Rudin, Flexion’s cofounder and CEO, the software’s “secret ingredient” is its use of reinforcement learning, which trains computers through trial and error. Each layer of the software, from the master AI model to the simulation and motor control, employs this method. Tech leaders like Elon Musk and Jensen Huang argue that humanoids could significantly impact the economy by replacing human labor. However, Flexion’s demonstration highlights the need for advances in AI to empower humanoids effectively.
Source: wired
