Shared Control Of Robot Manipulators With Obstacle Avoidance - A Drl Approach