Improving the skills of forest harvester operators

Abstract

Forestry suffers from a shortage of trained machine operators, which jeopardises efficient and productive operations. Extensive training is required to skilfully master the complex tasks of operators of forest harvesters and forest forwarders. Therefore, the digitisation of the industry envisages training and support systems on machines that provide real-time support to operators, both on-site and remotely. The aim of this thesis was to improve training methods and pave the way for the development of future operator support systems, therefore a detailed analysis of harvester operators' work tasks, focussing on motor control skills and cognitive (work)load, was conducted. The work was guided by the following two general research questions, which were systematically answered throughout the studies presented in this thesis. (1) How can training methods for robotic arm operators be improved by analysing performance limiting factors in the bimanual control of the robotic cranes and (2) How can the machine operators be effectively supported with different sensorimotor support systems to ensure high level performance? To this end, a multi-pronged approach using qualitative and quantitative methods was adopted and five scientific studies were carried out. For three quantitative laboratory studies, a multi-joint robotic manipulator was designed and programmed as a simulation environment, which in its basic layout resembles the crane of real forestry machines. To identify the challenges in learning the motor control of such robotic cranes, this work focussed on the joystick control of the individual joints (joint control) or the movement of the tip (end-effector) of the robotic crane. Two experimental studies on the acquisition of operating skills with the two different control schemes, showed that in spite of a gain in mental workload reduction with end-effector control, movement accuracy remains difficult with both control schemes. This refers with joint control to the challenging use of the joints involved in the fine control of the robotic crane and with end-effector control to a general lack of accuracy. In a third study, visual and auditory (sonification) support systems were implemented in the simulation environment and compared for increasing accuracy. Auditory support systems showed higher effectiveness, which depends on initial operator performance level. In summary, this thesis has shown that behavioural analysis at the level of joystick movements and the analysis of crane movements can be very fruitful for studying the development of human control skills and deriving new performance indicators that can be used in operator training and the design of different operator support systems. The development of machines with increasing technical operator support will potentially lead to new challenges in real-world operation, where the management of cognitive workload and the detrimental effects, specifically of cognitive underload conditions, will require a rethinking and design of the operators’ work.