Prof. Wen-Hua Chen, Department of Aeronautical and Automotive Engineering, Loughborough University, UK
Bio: Wen-Hua Chen holds professor in autonomous vehicles in the Department of Aeronautical and Automotive Engineering at Loughborough University, UK, where he is also heading the Controls and Reliability Research Group. Before joining Loughborough in 2000 as Lecturer in Flight Control Systems, Dr. Chen was a Research Fellow and then a Lecturer in Control Engineering in the Centre for Systems and Control at the University of Glasgow, Scotland. Dr Chen has a considerable experience in advanced control, signal processing and computational intelligence and their applications in aerospace and automotive engineering. In the last 15 years, he has been spending most of his effort in developing autonomous system technologies and their applications in agriculture, environment and defence. Prof Chen is a Chartered Engineer, and a Fellow of IEEE, the Institution of Engineering and Technology and the Institution of Mechanical Engineers, UK. He has published about 250 papers with about a total of 9,000 citations.
Title: Towards High Level Automation through Integrating Computational Intelligence and Control: A Case Study
Abstract: Automation is generally realised by automatic control systems with clearly specified references. To further increase the level of automation where only a high-level goal is specified, autonomous control with reasoning is required. This talk presents a case study of this type of new control systems – control a mobile sensor platform (e.g. a ground robot or an unmanned aerial vehicle) to approach unknown sources of airborne chemical and biological substance release. Hazard substance release in atmosphere is of major concerns in environment monitoring, anti-terrorist, and disaster and emergence management. The whole system consists of chemical sensors, mobile sensor platforms, reasoning and planning algorithms. By utilising the current and previous chemical sensor readings, reasoning algorithms developed in a Bayesian framework estimate key parameters associated with the release and environment conditions. Based on that, at each step, the decision for the next move of the sensor platform is optimised in order to maximise the chance of finding the source and reduce uncertainty in location estimation. Driven by the inference algorithm and informative based planning and control, the sensor platform is able to approach unknown sources under an unknown environment condition without a specified goal location and driving path. The Bayesian inference algorithms are implemented through the particle filtering technique. Experimental tests of the complete system were successfully conducted, which overcome the challenges of intermittent sensor readings due to air turbulent conditions, unknown release including location and release rate, unknown environment conditions (e.g. wind direction and speed) and a high level of noise in chemical sensors. The developed autonomous search systems could be widely used for environment protection and monitoring, oil and gas industry, and disaster or emergency management and keep the first responders out of harm.