Controlling Hand Movements Relying on Tactile Illusions: A Model Predictive Control Framework

In recent studies, we demonstrated that when a blindfolded participant slides her/his finger-pad on a ridged plate to reach a target, tactile feedback induces the illusory perception of bending towards the direction perpendicular to the ridges. The contribution of tactile motion estimate is hence optimally integrated in a Bayesian way, with the estimate of muscular-skeletal proprioception and motor command. As a consequence, a systematic deviation of hand trajectories in the opposite direction with respect to the one estimated by touch can be observed. The goal of this paper is to exploit the aforementioned tactile illusion to guide the user's finger, moving on a ridged plate, towards an arbitrary desired point A, while she/he is instructed to move towards another target B. To this aim, we designed a Model Predictive Control strategy in a simulated environment to estimate at each time instant the optimal ridge orientation that minimizes a proper cost function. We simulated fifty trials for different positions of points A and B, also varying the level of the noise associated with the motor command and to tactile cue. Results show that the final positions of the simulated trajectories are in a range of +/- 1.5 degrees with respect to the position of the desired final goal for similar to 80% of the cases. These results open promising applications in the framework of haptic retargeting where only one real object is used and the other items of the scene are virtual.