I'm a graduate student working with the NAO robots. Part of my experiment involves a type of reinforcement learning paradigm where I would like the robot to reach to a ball in space. The trouble is that I need this reach to be performed thousands of times and so, I would need some kind of simulated environment to model the problem. I first started using NAOSim because the software came with the purchase of the NAO robot. But the problem we're having with the NAOSim software is that we have no way of controlling specific objects using our python scripts. For example, at the start of every trial, I would like to reset the environment and get/set position of a ball that I would like the NAO to reach to. No having this functionality in the API makes the process of automation very difficult.
My question about SimSpark is two-fold. The first being, can I create a similar environment whereby I can just place a ball in space have the NAO robot reach to it? Would i be able to do this using a script? My second question is regarding the code associated with programming SimSpark. I read that SimSpark is programmed using Ruby. Does that mean that I would have to write scripts in Ruby to control the simulated robot? All my code right now is in Python and I can control the real NAO robots using that code. Is the SimSpark API python friendly?
Thanks for your help!