Text this: Continuous Multi-Target Approaching Control of Hyper-Redundant Manipulators Based on Reinforcement Learning