Humans increasingly share their environments with robots. They can be found in factories, hospitals, centers for the elderly, and even homes. The long-term goal is for such robots to assist their owners in everyday tasks. For that, they need to adjust to their potentially-changing environment and user's preferences, act safely for themselves and for people around them, and provide human-understandable explanations of their actions and intentions. The aim of this project is to design a learning architecture allowing humans (be them designers or non-expert
users) to easily force an adaptable robot to respect safety guidelines, as well as deflect its behavior before it engages in undesirable actions. Moreover, in the perspective of explainability and user- friendliness, our learning architecture will be able to provide intuitive explanations when requested, and to take human input into account while learning, so that users can customize their robot's behavior. We plan on testing our learning architecture on a motorized wheelchair, which was recently given to us specifically for this project by Invacare, a robotic healthcare company.
Effective start/end date1/01/1931/12/20

    Research areas

  • robots

    Flemish discipline codes

  • Other computer engineering, information technology and mathematical engineering not elsewhere classified

ID: 43825913