Deep Learned Path Planning via Randomized Reward-Linked-Goals and Potential Space Applications