The QuDa-KI project pursues two scientific goals. The qubit-based representation of (robot-related) data streams, in particular sensors and actuators, is being developed in order to be able to use them in quantum-enhanced machine learning algorithms. The focus here is on qubit-based minimal representations of essential features in order to be able to implement use cases with the few qubits currently available in the NISQ era. Furthermore, classical machine learning methods, especially in supervised learning, e.g. classification, are investigated with respect to new hybrid quantum extensions. These extensions are to be optimized especially for the smallest possible data sets in order to enable the interaction with qubit-based minimal representations.