Influence of the interaction technology on the usability of a virtual training system and its effectiveness in teaching procedural assembly tasks

Résumé

Industrial maintenance and assembly tasks require memorizing processes and techniques for each machine that in some cases, it can involve more than 150 steps. The automation of these procedures is achieved through the repetition of the procedural task in the same environment (i.e. real or augmented), or similar (i.e. virtual). Virtual systems are suitable when the real-machine is not available, or is not convenient to practice with. Users interact with the virtual environment through interfaces which can be decomposed into interaction techniques and interaction technologies. Since there is a broad range of available interaction technologies, it is not straightforward which technology should be used in this context. Each interaction technology has some particular features which could, or not, affect the learning process in different ways: efficiency, effectiveness, user¿s satisfaction, learnability, easy of interaction, etc. This thesis presents a study of the existing interaction technologies in virtual assembly systems, and evaluates their impact on the learning process. The study is focused on transferring the cognitive skill, assuming the motor skill required is known or easy to learn (e.g. screwing, cutting, fixing, and gluing). A research methodology has been proposed in order to address this thesis work. The methodology establishes which measures and system requirements are needed, and it proposes two experiments. First experiment is focused on assessing one-handed interactions (i.e. mouse, desktop haptic, and Motion Capture system, also known as Mocap) for learning a procedural assembly task. Users are trained in the virtual environment, and later with the knowledge acquired they have to perform the real assembly task. Second experiment is focused on studying the usability of dual-handed interactions (i.e. haptic, Mocap, and a combination of both) for assembling a bimanual task. Based on the experiment results and evaluator observations, a design guideline has been developed for avoiding usability problems derived from a bad used or design of the interfaces tested. Additionally, a guide for choosing the most appropriate interaction technology for learning virtual assembly tasks is given.