Open Access

Katrin-Misel Ponomarjova, Thomas Wendt, Anke Fischer-Janzen, Sinan Süme, Bastian Kayser

• Applying methods in artificial intelligence to the field of assistive robotics has the potential to increase the independence of people with disabilities. The usage of AI to realize a shared control in this context is controversial, due to the high complexity of everyday tasks and the needed safety requirements. This paper presents the development of a user interface for AI-driven assistiveApplying methods in artificial intelligence to the field of assistive robotics has the potential to increase the independence of people with disabilities. The usage of AI to realize a shared control in this context is controversial, due to the high complexity of everyday tasks and the needed safety requirements. This paper presents the development of a user interface for AI-driven assistive robotic arms (ARA) that aims to assist people with physical disabilities in performing daily activities. This interface allows the user to select object manipulation tasks based on the objects recognized in a live video stream. Further, we compare several state-of-the-art, real-time object detection models to facilitate automatic robotic control. The results demonstrate the feasibility of the model and its potential integration into the overall robotic system.…