Is it feasible using today's technology to use an AI training algorithm to custom teach a robot to do common household cores?

Question

Like making a bed, washing dishes, taking out the garbage, etc., by training it on the video of specific individuals doing those cores in their own unique environments?

I have researched what machine learning is capable of doing at this point in time, and it seems this may be now feasible when done on a customer-specific basis and enable by an A.I. enhanced, full articulated, robot along the lines of an enhanced InMoov. https://en.wikipedia.org/wiki/InMoov

If it's feasible, what are the AI algorithms I should be considering to train my robot to do these tasks? Isn't deep learning the most promising of these selections: https://www.ubuntupit.com/machine-learning-algorithms-for-both-newbies-and-professionals/?

Answer 1

I would suggest using a neural network with back propagation. From what I know, they can be applied to many different circumstances and work well. For your more simpler and repetitive tasks like moving an object, you can just use simpler regression methods.