Sensory loss in patients with neurological injuries is associated with poor neurorehabilitation training outcomes. However, in robotic neurorehabilitation, sensory interventions receive systematically less attention than motor training. To boost the potential of robotic rehabilitation, we are evaluating the feasibility of employing a haptic robotic device for the assessment and training of touch sensibility. Here, we present preliminary results on the test-retest reliability from 20 healthy participants. Participants performed a sensory discrimination task, where different surfaces were rendered by the haptic device, while they were passively moved by the robot and when they actively explored the surfaces. Participants' performance from two consecutive days was compared and no significant differences were found, neither for active nor passive exploration. However, the active condition had poor reliability, while the passive condition had moderate reliability, suggesting that sensory assessment may benefit from robotic guidance to achieve a more repeatable tactile evaluation.