Flexion Robotics, a Swiss startup founded by former Nvidia engineers, has created a method for training humanoid robots to perform complex tasks autonomously. Their approach involves teaching robots skills in simulation and using a master AI algorithm to apply these skills in real-world scenarios, such as retrieving parcels and navigating office environments.
The technology relies heavily on reinforcement learning, allowing robots to learn through trial and error. This innovation could enhance the commercial viability of humanoid robots, as Flexion collaborates with various robotics companies to expand its applications across different humanoid forms.
Flexion Robotics' training method could revolutionize humanoid applications in various sectors. Watch for partnerships with hardware manufacturers and advancements in AI that may enhance the functionality of humanoid robots, potentially reshaping labor dynamics.