A sensor-based physical tool adaptation framework for facilitating motor skills learning

Abstract

Many motor skills that people learn throughout their lives involve mastering a physical tool, such as riding a bike, writing with a pen, or playing basketball. When learning these skills, people often use physical learning aids to provide support. However, currently these learning aids only come in predefined levels. For instance, training wheels on a bike are either mounted or taken off. This jump from an easy task to a much harder one makes the transition difficult in learning the skill. In this thesis, we address this challenge by adapting the physical tool according to the learner's progress. For instance, while learning to ride a bike, we monitor learners' balancing skills and as they improve, we gradually lift the training wheels to reduce support and increase the difficulty. Thus, this approach enables a step-by-step transition from an easy to hard level that, like existing adaptive learning systems for math and language skills, is personalized for each individual learner. To illustrate this idea, we built an end-to-end system that allows designers to setup adaptable tools that physically change when a learner's skill level increases. This system uses sensors integrated with the tools to measure progress; parametric 3D modeling to adapt the tool; and either actuation or re-fabrication to deploy the physical change.