04/2017 - 06/2017
The project executed during this course builds upon the PhD project ‘Uitkijkpost’ of Xu Lin, within the Designed Intelligence group in the Industrial Design faculty of the TU/e. We built a system to help connect residents of an elderly home to the outside world. The system consists of an intelligent driving robot, which aims to provoke reactions through its embodied behavior within an outside world environment.
The robot’s aim is to increase interestingness of live video footage in the care home based on positive and negative feedback regarding its movement patterns. Using a physical controller, four parameters controlling its movement can be adjusted. Using a learning algorithm called Support Vector Machine, the physical controller predicts whether or not the combination of variables will be appreciated.
Although some of the information in the lectures overlapped with the lectures from my previous course, the project during this course was based on a different type of learning algorithm and completely new to me. Using a Support Vector Machine algorithm, I learned how make accurate predictions based on pre-defined and updateable training data. Since most of my responsibilities considered implementing the algorithm and handling the training data, this course enabled me to apply this type of machine learning in any future project.
Compared to the previous course, more attention was spent on the social aspects of the concept and how it could be implemented in its corresponding context. I learned enough about machine learning to understand its complexity, which leads me to believe that I as a designer would not always be the one implementing the actual algorithm in a professional context. Instead, knowing when machine learning could be applied in a product for a societal context, and how the resulting information could improve the user’s experience, in my eyes is a valuable and more realistic skill.