Limb position effect in myoelectric control: strategies for optimisation and standardisation

Objective.Myoelectric control uses electromyography (EMG) signals for muscle-machine interfacing with applications in prostheses, augmented/virtual reality, and consumer electronics. However, factors such as changes in the limb's position during activities of daily living reduce the controller's reliability. Therefore, there is a need to develop techniques that reduce this limb position effect to increase the widespread adoption of these technologies.Approach.We created an open-source device to standardise myoelectric control experiments. The device has sixteen locations for automatically positioning the participant's arms to perform hand gestures or grasp objects, with lights and sensors for guidance and timekeeping. We used this device to collect data from nineteen participants (one with a congenital upper-limb difference) across three modalities: performing hand gestures with a static limb, performing the same gestures with a dynamic limb, and moving grasped objects. We recorded forearm EMG and kinematics of the upper limb. A linear discriminant analysis model was trained, and classification accuracy was evaluated across arm positions and modalities.Main results.Classification accuracy decreased when tested on untrained positions, confirming the limb position effect. More training positions improved accuracy, with four providing an optimal balance between training burden and classifier accuracy. Furthermore, multi-modal classifiers that considered both EMG and kinematic data were found to have optimal performance when trained with a dynamic limb.Significance.The limb position effect can be countered by training with multiple positions and including kinematic data. Multi-modal classifiers should be trained with a dynamic limb to achieve high accuracy with a minimal training burden. For the first time, these findings are presented for congenital limb differences. Finally, our open-source, automated device will help standardise datasets between laboratories, aiding the further development of robust and widespread myoelectric control.