Impedance Collaborative Dual Arms Service Robots with Self-Learning Strategies

This project entitled "Impedance collaborative dual arms service robots with self-learning strategies" arms to develop some service robots which can learn and work in a warehouse and cooperate with other robots or people. In order to deal with huge and strongly heterogeneous items, the robots must have robust visual recognition systems and feasible dual arms. For achieving that, we will progressively carry out the needed abilities of the robots within three years. First year: Recognizing, grasping, and transporting items are the basic abilities for a warehouse robot, however, these tasks become difficult when the items are huge and strongly heterogeneous. Therefore, in this year, we are going to improve these abilities and construct four subjects, including object recognition and classification, object 3D modeling, grasping pose learning, and impedance control for single arm. The object recognition and classification system will use the Broad Learning system to classify all objects for simplifying the computation complexity and speeding up the recognition time. The 3D object modeling is constructed for learning a suitable grasping pose no matter how the object is placed. Finally, the impedance control arm is designed for improving the safety and the operability of the arm. Second year: In this year, we will force on the dual arm collaborative manipulation abilities for a single robot, and construct four subjects: 3D object recognition system, dual arm collaborative motion planning, dual arm 3D path planning, and warehouse tasks implementation. The 3D object recognition will calculate the rotation angle by comparing the model in the database built in the first year to speed up the grasping pose generation. A dual arm coordinate structure will be constructed first and used to implement dual arm impedance control system. A suitable pair grasping poses for dual arms is then learned based on this system. The motions and the paths of each arm are then generated to allow the robot to execute some warehouse tasks. Third year: For achieving an automated logistics warehouse, there are many robots and do several kinds of works. Therefore, in the third year, we are going to construct an IoT system to connect all robots in the warehouse. The system will construct a global map in which each robot is located, and a task is arranged by its importance, emergency, and other factors, and allocated to the robots according to the distance between the task and the robot, and what the robots are doing. Based on this system, all robots will be monitored to ensure the safety of the warehouse. Besides, the robot will cooperate with human worker and actively provide assistance.