My research topic is "Deep Reinforcement Learning for Natural Robotics Grasping". Most robots control solutions focus on applying IK for motion planning. Such methods can get accurate results, and at the same time, they need heavy computation for every step's IK. Some times the movement trajectory looks wired because the system chose an energy-consuming plan. In this project, we aim at developing a new system based on RL to let robots realize a human-like(intelligent, smooth) grasping task.
|Last modified:||28 February 2019 12.04 p.m.|