Double action Q-learning for obstacle avoidance in a dynamically changing environment | IEEE Conference Publication | IEEE Xplore