This AI robot can cook, play piano, solve Rubik's cube
What's the story
Genesis AI, a robotics company, has unveiled its revolutionary robotic system called GENE-26.5. The innovative system can perform complex human-like tasks such as cooking meals, solving Rubik's cubes, lab pipetting and wiring cables. It can even play the piano in real time! The company's CEO Zhou Xian said their team of around 60 people taught the robot to play a new song on the piano in just an hour.
Strategy
Genesis AI's 'full-stack' approach to robotics
Genesis AI believes that human-level robot hands are becoming a reality sooner than expected. The company advocates for a "full-stack" approach to robotics, emphasizing that advanced AI models alone aren't enough. They argue developers need human-like robotic hands, better training data, precise motor control, realistic simulations and systems capable of seamless integration and efficiency.
Innovation
The robots are trained using human demonstrations
The capabilities of the GENE-26.5 system were demonstrated with a robotic hand called Genesis Hand 1.0. It closely mimics a human hand with 20 degrees of freedom and soft-contact surfaces. The robots are trained using human demonstrations, simulation environments, and AI learning systems to perform complex tasks like grasping multiple objects at once or wire harnessing efficiently.
Training tools
Special gloves to teach the robot
Genesis AI has also developed special gloves equipped with sensors to capture human hand movements, grips, and pressure application during tasks. This data is used to teach the robot how to imitate human actions. The company says a 30-second "complex skill," like those seen in their cooking demo, requires a few hours of human data combined with less than half an hour of data from the robot performing the task.
Progress report
Limitations and challenges
The demonstrations by Genesis AI highlight the progress made in robotic hand coordination and dexterity compared to traditional industrial robots. However, these robots are not fully "general intelligence" yet. They still need training for specific tasks and some actions still fail frequently, especially delicate ones like cracking eggs. Despite this, Xian believes these are probably the most complex tasks ever performed by a robot in a human-like way at similar efficiency, speed and performance as humans.