AU Lecture - Learning Force-dominant Skills from Human Demonstration & Cross-modal Understanding and Prediction for Cognitive Robots
 
time: 2018-03-10

    On March 9, the academic lecture of AU organized by the School of Automation Science and Engineering was held on the sixth-floor conference room of Building 3. The topic of this lecture was Learning Force-dominant Skills from Human Demonstration & Cross-modal Understanding and Prediction for Cognitive Robots. Dr. Li Miao from School of Power and Mechanical Engineering of Wuhan University and Dr. Zhong Junpei, a researcher from the Institute of Artificial Intelligence under the AIST were invited as the major guests. The lecture was chaired by Prof. Yang Chenguang.

    Before the lecture, Prof. Yang gave a brief introduction to Dr. Li and Dr. Zhong. At the very beginning, Dr. Li explained the topic of using advanced machine learning algorithms in the control and design of complex mechanical systems. He introduced the theory of stable grab planning for the establishment of complex shaped objects, and proposed the dynamic grabbing adaptive method based on sensor data and the process of developing a closed loop grabbing system for robots in an unstructured environment. Secondly, Dr. Zhong introduced the process of inference and perception by the human brain, and proposed a hierarchical prediction cognitive model for robots. Then he introduced how to apply this cognitive model to cross-modal understanding and prediction of robots and present them on site.

   The audiences showed great interest and we believed that they could benefit a lot through this lecture and stimulate enthusiasm for scientific research.