Motion Planning of an Automatic Blood Sampling Manipulator Using DDPG Algorithm

In this paper, an automatic blood sampling mechanism is proposed to reduce the working pressure of clinicians and the high risk of cross-infection caused by doctor-patient contact. The simulation environment for this mechanism is built via Gazebo in the Robot Operating System (ROS). Using Deep Deterministic Policy Gradient (DDPG) algorithm, the mechanism is trained to accomplish the blood sampling task. The reward function and the neural network is adjusted to optimize the training result. Simulations prove that the trajectory of automatic blood sampling mechanism using DDPG algorithm has excellent performance in rapidity and stability, thus the effectiveness and usefulness of the designed system.