In the KI SIGS research project, SZENARIS together with the Universities of Hamburg and Bremen developed a movement training system that was equipped with artificial intelligence to correct or prevent errors during training.

For this purpose, a prototype for recognition of body posture and movements was developed based on self-organizing neural networks. During exercise, the user positions himself in front of a display with a tracking camera, on which a virtual representation of his body can be seen. The camera now recognizes the movement and compares it with the ideal execution. The dynamic feedback is supported by virtual training agents on the monitor and/or a robot. Both provide verbal and gestural feedback on the user’s performance.

The system is supplemented by a game design with the aim of promoting user motivation and creating adaptivity with regard to different fitness levels.