Abstract—Providing a somatosensory feedback in a myoelectric prosthesis is an important goal yet to be achieved. Electrical stimulation of the skin to activate the tactile sense is one possible method to restore the missing sensory information. In this scheme, the user of the prosthesis has to recognize the information delivered as a pulse train, and then to respond by generating appropriate control actions in a timely manner. We have developed a flexible real time setup to investigate this context. The task for the healthy subjects was to control the systems of different dynamics emulating the response of a prosthesis in different control modes (position and velocity). The subjects used a joystick to steer the system along a pseudo random multi step reference trajectory. The current tracking error was delivered to the subject in the form of electrotactile feedback. The same task was performed with visual feedback as a benchmark. The results demonstrated that it was feasible for the subjects to control all of the tested systems. The nature of the system (position, velocity) did not make an important difference, but the time lag showed to be a significant variable, with the long time constant deteriorating the performance. This is an initial work towards the full time and frequency domain characterization of a human operator in a sensory substation system, which is a very important step towards an optimal design of the tactile feedback. I. INTRODUCTION Myoelectric prostheses are in routine use already for many years. The command interface is based on capturing the electrical activity of the user’s muscles, decoding the intent and commanding the prosthesis accordingly [1]. Using this simple and reasonably robust feedforward communication channel, the users can accomplish a variety of daily activities, and therefore the motor functions lost due to an amputation can be restored to a certain degree. However, the most important drawback of the electric prostheses is the lack of direct somatosensory feedback. Since the feedback is instrumental for the control of movements, the amputees learn to rely on indirect cues. This partly compensates for the lack of feedback but also increases the cognitive efforts. Sensory substation is a method for the restoration of missing sensations. The basic idea is to capture the information that was originally received by the lost sensory This work is supported by the German Ministry for Education and Research (BMBF) via the Bernstein Focus Neurotechnology (BFNT) Göttingen under Grant No. 01GQ0810 and the Ministry of Education and Science of the Republic of Serbia (project no. III–41007). Strahinja Dosen and Dario Farina are with the Department of Neurorehabilitation Engineering, University Medical Center Göttingen, Germany (e-mail: sdosen@bccn.uni-goettingen.de; dfarina@bccn.uni- goettingen.de). Goran Krajoski, Damir Djozić and Nikola Jorgovanovic are with the Department for Systems, Signals and Control, Faculty of Technical Sciences, University of Novi Sad, Serbia (e-mail: krajoski@gmail.com; djozza@gmail.com; nikolaj@uns.ac.rs). organs (e.g., skin of the hand) and redirect it to the receptors which are still functional (e.g., skin of the residual limb). The tactile sense can be activated through a direct mechanical stimulation (e.g., vibrations) or electrically, by delivering low level current pulses through concentric electrodes [2]. This electrode configuration comprises an active inner field and an outer ground ring, which assures that the current travels superficially, activating the cutaneous afferents and avoiding the stimulation of deeper sensory-motor structures (e.g., muscles). Compared to vibrotactile devices, the electrical stimulation has faster response and lower power consumption since there are no moving mechanical parts. On the other hand, if the parameters are not properly adjusted, electrical pulses can elicit painful sensations. The information to be sent to the user can be coded by modulating the parameters of the stimulation profile (e.g., current amplitude, pulse width, and frequency) and/or by changing the active channel (i.e., spatial coding). The idea of using the electrotactile stimulation to close the loop in the control of prosthetic devices is not new. It was an active research topic in the period from 60’s to 80’s [3]. The first task in this early research was to characterize in detail the psychometric properties of the electrotactile information channel. Various parameters, such as, temporal and spatial discrimination, dynamic range, magnitude scaling, different coding schemes, recognition of discrete levels and continuous signals, different waveforms and body locations have all been thoroughly investigated in a number of studies [2], [4]. Importantly, in these experiments, the electrotactile channel was tested in an open loop manner. The stimulator was generating the signals and the subject was a passive recipient whose task was to correctly recognize the delivered stimulus. The subject would indicate the recognition by pressing a button (discrete stimuli) or moving a mechanical slider (continuous stimuli). These experiments gave an important insight into the nature of the electrotactile stimulation and the potential channel information capacity. However, in an actual implementation within the prosthetic control loop, the goal of tactile feedback is to govern the operation of the device. Therefore, it is not enough for the user to only recognize (receive) the information, but he/she has to be able to utilize this information to generate appropriate and timely control actions. There were only a few studies that tested the electrotactile feedback by simulating the actual control loop [5]. However, the tests were very limited and the system dynamics simple (e.g., unit gain). We have developed a flexible setup that allows us to test different control scenarios. The task for the subjects is to operate the systems of different dynamics so that a system tracks a predefined trajectory. Importantly, the feedback about Closed Loop Control of Dynamic Systems using Electrotactile Feedback Strahinja Dosen, Goran Krajoski, Damir Djozić, Dario Farina, Nikola Jorgovanović