Modelling and Data-Driven Hybrid Method for Intelligent Robotics

Service robots have developed considerably in recent years to assist human’s daily life. Model Predictive Control (MPC) appears to be one of the popular process control methods and has demonstrated its effectiveness for industrial applications. However, an accurate model is difficult to obtain when complex dexterous manipulation is required in some of the tasks in unstructured environments. Deep imitation learning is an end-to-end learning approach. However, it has inherent black-box effects. To this end, we aim to combine knowledge-based modeling and data-driven approaches to develop intelligent service robots to assist human’s daily life.