Human adaptive mechatronics methods for mobile working machines

Despite the trend of increasing automation degree in control systems, human operators are still needed in applications such as aviation and surgery, or machines used in remote mining, forestry, construction, and agriculture, just to name a few. Although there are research results showing that the performance between the operators of working machines differ significantly, there are currently no means to improve the performance of the human-machine system automatically based on the skill and working differences of the operators. Traditionally the human-machine systems are designed so that the machine is "constant" for every operator. On the contrary, the Human Adaptive Mechatronics (HAM) approach focuses on individual design, taking into account the skill differences and preferences of the operators. This thesis proposes a new type of a HAM system for mobile working machines called Human Adaptive Mechatronics and Coaching (HAMC) system that is designed to account for the challenges regarding to the measurement capability and the work complexity in the real-life machines. Moreover, the related subproblems including intent recognition, skill evaluation, human operator modeling, intelligent coaching and skill adaptivity are described. The intent recognition is solved using a Hidden Markov model (HMM) based work cycle modeling method, which is a basis for the skill evaluation. The methods are implemented in three industrial applications. The human operator modeling problem is studied from the structural models' perspective. The structural models can be used to describe a continuum of human operator models with respect to the operating points of the controlled machine. Several extensions and new approaches which enable more efficient parameter estimation using the experimental data are described for the conventional Modified Optimal Control Model (MOCM) of human operator. The human operator modeling methods are implemented to model a human operator controlling a trolley crane simulator. Finally, the concept of human adaptive Human-Machine Interface (HMI) is described. The analytic and knowledge-based approaches for realizing the HMI adaptation are presented and implemented for trolley crane simulator control