SoftHand: grasping intelligence for lower arm prostheses

So far, lower arm prostheses often only functioned as a cosmetic disguise to conceal the missing body part. While newer models help the wearer with grip patterns, every hand grip has to be readjusted and newly activated. There is still no prosthetic device that is easy to control and allows a flexible response to objects the wearer is grasping.

In this MEDICA-tradefair.com interview, Dr. Eike Jakubowitz talks about SoftHand and explains its control mechanism and the grasping intelligence that puts this type of flexible prosthetic option "within reach".

Dr. Jakubowitz, we are talking about a prosthetic hand you helped develop at your Institute. Could you briefly tell us what makes this hand so unique?

Dr. Eike Jakubowitz: Our SoftHand Pro is a development that is based on soft robotics and allows robots to grasp and manipulate various objects. We are taking this robotic hand and are working at transforming it into a prosthetic hand. What makes the SoftHand Pro different from earlier prostheses is that it exhibits synergistic hand motion patterns. Thanks to kinematics – i.e., the motion of all fingers combined – this already covers 80 percent of all hand grip patterns. The remaining 20 percent can be achieved thanks to soft robotics features. The fingers of the prosthetic device are flexible to where they adapt their grasp to the shape of the object. This means less effort is required from the patient when it comes to controlling the device. He or she only needs to mentally perform one of two movements – opening or closing the hand - via the forearm stump. In the case of earlier prosthetic hands that feature different grip patterns, patients must first select the right grip pattern to grasp the respective object. This step is eliminated with the SoftHand Pro, making grip motions easier, more natural and smoother.

How does this control mechanism work exactly?

Jakubowitz: We actually only use the two control signals of opening and closing the hand. When the patient makes a stretch movement in the wrist, the hand opens. If he bends the wrist, the hand closes. Patients only have to control the wrist extensor or flexor muscles. To do this, we conventionally access the action potential of the muscles on the skin surface of the existing arm stump and use these impulses for the control mechanism.

Why do you refer to this as "grasping intelligence"?

Jakubowitz: When we studied healthy test subjects, we detected that most grasping movements require similar movement patterns. These are so-called synergies. The movement of the finger joints and the wrist repeatedly perform the same movement sequences for different objects. The prosthetic device’s grasping intelligence comes from the mechanism only wanting to completely perform the dominant part of the motion. Imagine it like the pincer grasp where the thumb and index finger are touching each other versus a tripod grip, which requires contact of the thumb, index and middle finger. The mechanics of the fingers and the hand make these different movement sequences possible: In the case of the SoftHand Pro, a pulley that is attached to the motor snakes through all the fingers of the hand. If the prosthesis wearer now wants to perform the tripod grip, the fingers are slowed down by the object contact and the differential gear to where only the first three fingers make contact.

What did you study on the healthy test subjects?

Jakubowitz: We first asked them to perform different gestures like waving, pointing a finger or making the "OK" hand sign scuba divers use. We also made the test subjects perform object-related movements, that being movements made while using an object. This might be grasping a pen to write or opening a laptop for work. Thirdly, we asked them to perform so-called transitive movements, where an object is moved from one place to another for example.

Are there integrated sensors in the fingertips that measure the resistance?

Jakubowitz: The outlined function of the SoftHand Pro is not based on the processing of external measurement data and therefore generally does not require any sensor technology. Having said that, for the recently introduced second prototype, we tested additional devices to give patients feedback on the grip strength the hand applies to a grasped object and contact information in the form of vibration for example. We placed different feedback systems on the upper arm and other parts of the arm to give patients feedback on the strength with which they grasp, indicating whether they touch different surfaces or the softness of the object. Wearers of earlier prostheses miss out on this type of feedback, prompting them to crush a paper cup for example.

However, after several development cycles, the third prototype of our SoftHand Pro is already so fine-tuned to where these feedback systems are no longer needed for many tasks. The third prototype is honed to where the patient is able to detect whether the grip force is sufficient for a paper cup simply based on the positive locking of the soft robotics.

What are the next steps for the device?

Jakubowitz: We started working on the first prototype two years ago as part of the H2020 "SoftPro" EU Project. At the time, the SoftHand Pro had reached the American Technology Readiness Level (TRL) 5. We aim to have the hand ready for the market within the next two years and make it accessible to patients for use. This would correspond to a TRL 8 or 9.