Machine learning for hand tools

Using TinyML to optimize work processes and ensure quality

Even in Industry 4.0, there are manual processes in the production chain. To incorporate these processes, experts from the domains of positioning, networking and machine learning (ML) have developed an embedded intelligent sensor module for hand tools, which can be integrated into existing production IT infrastructure. This use of ML to optimize and implement AI-based processing chains on embedded systems is known as tiny machine learning (TinyML).

The compact sensor module uses acceleration, rotation rate and magnetic field sensors and can be attached to hand tools. Data captured by the sensor forms the basis for an AI pipeline, which detects and identifies all work steps of the hand tool. As such, the system detects relevant actions such as the tightening or loosening of a screw, for example, or identifies the location or condition of the respective tool at the time of the action. An app provides notifications about the progress of the work or any deviation from the target process. The training process and the evaluation of new models run on a fully automated basis (AutoML) and require no expert knowledge. Activities to market and launch the technology are being undertaken with our partner, the High-Performance Center Electronic Systems (LZE).