FeetBack Feedback of tactile sensations from high-tech prosthetic hands

When extremities have been amputated, the limbs can be replaced by artificial ones. There are three different types of prosthetic hands.

1. Cosmetic prostheses
2. Body-driven prostheses
3. Myoelectric prostheses

Cosmetic prostheses are rigid replicas designed to look as natural as possible, so that the wearer feels comfortable with them. Body-driven prostheses, on the other hand, can move their fingers or grippers by the wearer tightening or loosening a cable. This cable is not attached to the missing limb, but to another part of the body. This allows the wearer to control the grip force fairly well, as the pressure can be felt via the cable. The third type of prosthesis is the myoelectric type. Here the prostheses are controlled via electrical signals that are associated with muscle contractions and can be measured on the skin surface of the residual limb. This means that the control of the hand is mechanically decoupled from the user, which enables more complex grip patterns than those of a body-driven prosthesis. However, the wearer does not feel how firmly an object is being held in their hand and has to perform constant visual checks of the prosthesis. This is tiring and leads to many users rejecting their prosthetic devices.

Researchers worldwide are working on a solution to give artificial hands a sense of touch and feed this tactile information back to the wearer. Many studies focus on tracing sensor data from the fingertips to the residual limb within the prosthesis. In a previous project, BFH worked with EPFL and CSEM on similar solutions and achieved positive results. However, the test persons were not happy with the design because it looked too heavy and unattractive on their arm.

Course of action

In the FeetBack project, researchers from the Institute for Human Centered Engineering HuCE at BFH took a novel approach: instead of feeding back the sensor data via the shaft, it would instead be transmitted discreetly via the wearer’s shoe. Like the hand, the feet react sensitively to vibrations, and it is this fact that the researchers sought to exploit. The development team was particularly keen not to resort to invasive methods of any kind. The aim was to develop a device that could be attached to a commercial prosthesis and enhance it with the sense of touch. With this they hope to reduce the mental exertion required to control a myoelectric prosthesis and increase the owners’ acceptance of their devices.

Results

The system they developed is made up of two parts: a sensor glove for the prosthesis and the FeetBack insole incorporating vibration motors. The sensor glove measures the grip force of the hand with one sensor each on the thumb and index finger (the grip functions of the leading prosthesis manufacturers focus on these two fingers). The grip force is then transmitted by radio signals from the hand module to the foot module, where the force is then assigned to one of six levels. If no load is placed on the hand, the wearer receives no feedback. However, as the grip force increases, specific vibration motors in the insole are actuated. Motors inside the insole located close to the toes indicate weak grip forces. The further back towards the heel the motors vibrate, the higher the grip force. This means that the grip force is spatially coded in the shoe.

The suitability of the system for real-world use was researched in a study in conjunction with Balgrist University Hospital in Zurich. Four people took part in the study, all of whom had been using a myoelectric prosthesis for several years and were familiar with the same prosthesis model. This meant they could perform the experiments with their own personal device. The prosthesis was only modified by adding the sensor glove over the regular cover.

The experiments showed that it is possible to interpret information from the sensors of the prosthetic hand through vibrations on the foot. This confirms that the basic principle of sensory feedback from hand to foot is a success. However, it was revealed that the interaction between human, prosthesis and the prototype FeetBack system is too slow for tasks requiring instant responsiveness.

Outlook

The results with the FeetBack system are consistent with the results of similar studies involving tactile feedback on the arm or in the prosthetic socket. With this project, the researchers provide the basis for further research focusing on sensor feedback on the feet, not just the arm.