The Project
AI-XRobo allows the student to become the teacher by giving trainees control over end-to-end training processes for ML-driven industrial robots. Parameters, interactions, and models are presented as intuitive, tangible elements, while an AI tutor is on standby to answer in-depth questions.
Technology
The VR application is organized into four distinct stations.
The VR application is organized into four distinct stations. At the first station, the user selects from several predefined parameters for the upcoming robot-training process, such as the activation function, the loss function, and the learning rate. These choices are captured as interactive elements the user carries forward to the next station. There, they take control of a robot arm and record a sequence of movements that will serve as the starting point for the training process. This demonstration is saved as a portable 3D asset. At the training stage, the trainee assembles the selected inputs on a console and starts the training process.
A set of robot arms situated behind the console visualizes the training progression in accelerated time, providing a direct view of iterations as they pass. Once the process is done, the console outputs a finalized model trained with the provided parameters and recorded movement sequence.
This model is then brought to the final production stage, which consists of multiple robot arms along a conveyor belt. Each robot can be loaded with the model to influence how it behaves.
AI-XRobo therefore provides a gamified environment that makes the effects of individual parameters on the ML training process visible, fostering hands-on understanding and measurable learning outcomes.
