News

The last two decades have led to a massive growth in the design space of interaction techniques, e.g., by smartphones and virtual reality headsets. Moreover, user groups become more diverse as digital skill requirements are increasingly permeating all areas of life. However, the current design process for new interaction techniques still relies on extensive usability testing, a process that simply does not scale. We argue that it should be model-based, allowing for simulation and optimization, augmented by artificial intelligence/machine learning when needed.

In this lecture, we touch on how to develop, use, simulate, control, and optimize these models using Model Predictive Control, Path Following techniques, and Reinforcement Learning, all while keeping the domain-specific difficulties – we are trying to model human behavior after all – in mind.