Encoding manual rotations on a motionless knob.

Despite the robustness and versatility of touchscreens affording haptic rotation, physical knobs remain widely adopted in the control layout of professional machines and appliances. Their low cost, established design, and efficiency in encoding rotations - even when an operator's attention is focused elsewhere - make them an optimal choice. However, physical knobs are often prone to electro-mechanical damage in settings such as food or cleaning service facilities. To overcome potential consequent safety and productivity issues, we have designed and prototyped a motionless cylindrical device capable of encoding manual rotation. The device tracks finger contact positions on its lateral surface through capacitive sensing, which are then processed by a neural network-based encoding algorithm designed to classify manual rotations in real-time on low-cost embedded hardware. A user test evaluating manual rotation confirmed accuracy in line with a previous experiment conducted on a motionless knob. In parallel, a decrease in precision was observed, possibly as a consequence of the sensing technology and encoding algorithm. Subjective questionnaires assessing specific aspects of the interaction quality with the prototype reinforced previous findings, suggesting that achieving natural and intuitive gestures on a motionless knob requires adaptation of a deeply embodied interaction primitive such as manual rotation.