End-to-end sensorimotor control problems of AUVs with deep reinforcement learning | IEEE Conference Publication | IEEE Xplore