Build a robot that appears vaguely as a potted plant and whose primary purpose is to point its solar panel “leaves” at the best light source in order to “stay alive”. The robot will be controlled by software running on a RaspberryPi, to include TensorFlow machine learning capability. The better the “plant” learns to find and track the best available light source, the more resilient it will be to being repositioned and unexpected low light periods. If this basic object can be achieved, then we now have all the capabilities of a bull-blown Linux box right there in an entertaining potted plant.