Well, getting some experience in machine learning sounds like a good idea. You are still young and getting as much experience as possible before "settling down" into a PhD lab is always something I recommend to undergrads. Switching labs at 1.5 years in does mean that you will miss out on some productive time and possible publication with your current lab. How is your current lab? Do you find it interesting? Does it look like you could publish something this year? A publication would look great on your MSTP application.
Another option is to talk to the PI and see if you could start a side project incorporating machine learning into the motor control research.