Just like children are able to learn from their parents, robots are now able to do the same with our help through VR, but in this case, you only need to show them the task once. Sounds perfect, doesn’t it?
The new algorithm has been introduced to these robots and their performance is powered by two neural networks.
- One of the networks takes images and helps the robot see the position of the targeted object. What is interesting here is that the robot was taught through VR, which means that the robot never actually saw the real world and is able to interact with it.
- The second network keeps records of actions demonstrated in the VR, and the robot imitates the tasks given paying close attention to what task comes next.
This platform is still a prototype, and there is a lot of room left for advancements. Once this system has been completely developed there will be no limitation to what extent the robots could be used to aid humans.