Conceptual Modelling of Tasks A Design Pattern Approach from E-mail to Robots

Abstract: HTML clipboardThe thesis deals with the dual problem of finding out and modelling the needs of users in various situations and constructing models of supportive software. Task analysis has not been a very hot topic for research in the last years, but in my work I have tried to apply task analysis to certain application areas. In this thesis the focus is therefore on two different kinds of applications, the first being a traditional software application, electronic mail systems. The second application area is programming of service robots for the home, and specifically robots that are intended as support for people with functional disorders, e.g., as a result of neural diseases or accidents with neurological consequences. The work in the thesis spans over a long period of research, but there is a thread going through the thesis based on a perspective that regards work tasks as consisting of sequences of patterns. In the e-mail application area, the pattern theory was not outspoken, but can be seen in the software prototyping tool, GRASP, that was developed in pure Prolog. The task pattern system, TAPAS, that was developed from the work in robots and the interviews with people with various degrees of disorders is clearer on the pattern approach. With the GRASP system, the idea was that users could be involved in the development through prototyping tools that would enable a more individual design of the application and the interface functionality. When this was taken into the area of design for inclusion, it was clear that the pattern approach in TAPAS would need to cater for an individual design of the robot’s tasks. Individualisation is a costly process but it can be eased through the use of a pattern approach to task modelling. The conclusion of the work in this thesis is that task analysis and conceptual modelling still have application areas within software development, especially where there is a large need for individual adaptation of the applications (such as robots as support for people with functional disorders). I show that a combined knowledge elicitation method and task modelling tool based on a pattern-based perspective can facilitate a more flexible and individual design of software. More precisely, in the robot context, I show that a robot that is programmed through a pattern approach receives a potential to become flexible enough to allow for end user reprogramming under certain circumstances. The conclusion of my work is that task analysis in combination with conceptual modelling of tasks as patterns is a promising method combination for the programming in task areas where there is a large need for individualization and flexibility.