Inferring touch location on a handheld tool from mechanical vibrations

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
Manipulating a tool, such as a simple rod, enables us to gather information from our surroundings beyond the limits of our body: an example would be how we can accurately sense the location of touch on the rod. Being able to differentiate sensory information coming from a tool is theorized to be an important component of efficient tool use; its behavioral and neural mechanisms, however, have only recently begun to be characterized. Recent studies suggest high frequency mechanical vibrations resulting from hitting with the rod play a role in perceiving touch location by stimulating the mechanoreceptors of the hand. In the present thesis, both modeling and experimental approaches were employed to investigate the contribution of vibrations in tool-based sensing. We modeled the rod’s behavior based on the Euler-Bernoulli beam theory and defined how vibrational properties change based on the rod’s parameters and the location where the object was hit. We then used an accelerometer array to successfully measure cutaneous vibrations during a tool use experiment. Using the recorded vibrational patterns, we were able to model the response of three types of tactile afferents (SA1, RA and PC). Finally, we applied a support vector machine (SVM) based regression learning algorithm, which could differentiate the mechanoreceptor response array on a continuous parameter space with high accuracy. Our results suggest that the firing patterns of tactile afferents encode location related information which cortical areas are able to differentiate.
Faculteit der Sociale Wetenschappen