Unlike humans, robots do not come with the ability to think for themselves. This means that what a robot can do is pretty much what its creator has programmed it to do. While this is useful in factories on the assembly line where a robot might only have a few things to do, it becomes less useful when interacting with the real world.
Robots caretakers is expected to be a thing of the future, and to help that development along, Toyota Research Institute has adopted a unique approach to the problem by using virtual reality (VR) to teach robots skills it would need to be useful around a home. According to TRI’s senior manager Jeremy Ma, the reason for using VR is that because all homes are unique and it would be impossible to create a single set of instructions that would fit all households.
So by using VR, it allows researchers to put themselves into the virtual shoes of the robot and see what they see and see what data the robots are picking up. They can then create a set of parameters for the robot to follow, like being able to identify objects, what objects in the scene are important, or even further, what parts of that object are important, and from that, robots will learn how to handle it.
TRI also notes that its technique can also be applied to robots in other fields, but for now, it is clearly far from being ready even though it does show a lot of promise.