This Prosthetic Learns Your Habits and Gets Better the More You Use It

Esper Bionics’ AI-powered robotic hand prosthesis uses predictive technology to enhance customization.

Human augmentation has long been a fascination of science fiction, with many companies looking to expand human potential by making the technology a reality. Companies like Neuralink have made a splash in the media with their ongoing research to interface technology directly with the brain. However, if we consider the broadest sense of the term, human augmentation has been around for decades, including pacemakers and continuous glucose monitors.

A startup called Esper Bionics is looking to use technology to expand human capabilities at scale. In its quest to develop new devices, the company is starting with a self-learning robotic hand for people with limb differences.

Currently, there are an estimated two million people with limb loss in the U.S. alone, and this number is expected to double by 2050. Unfortunately, current prosthetic devices fall short in terms of their utility and aesthetics. Esper Bionics is developing its robotic hand to improve the lives of those with limb differences and accelerate the development of human-technology interfaces.

The Esper Hand gripping a fork. (Image courtesy of Esper Bionics.)

The Esper Hand gripping a fork. (Image courtesy of Esper Bionics.)

Esper Bionics Wants to Expand Human Potential

Esper Bionics, which was founded in 2019 by Dr. Dima Gazda, Anna Believantseva, and Ihor Ilchenko, is currently based in New York City, with research and manufacturing offices in Germany and Ukraine. The company is working to expand the full technology stack for electronic implants—developing the devices themselves, low-power electronics, and AI and advanced data analytics.

In conversation with Dr. Dima Gazda, cofounder of Esper Bionics, engineering.com learned more about the company’s history and ongoing R&D. As both an electrical engineer and medical doctor, Gazda has the unique education required to develop effective devices for human augmentation.

Originally, Gazda and his cofounders started the company to develop what they thought was the most important technology stack for the future of humanity: electronic implants. To start its R&D journey, Esper Bionics focused on the prosthetic industry, which is currently low-tech. Most industry-standard prosthetics are purely aesthetic and do not restore a limb’s functionality. Other companies are working to improve the technical capabilities of prosthetics, including Psyonic. However, Esper hopes to stand out with the speed and utility of its device, which actually learns user habits and customizes the functionality to each patient.

The Esper Hand as a Self-Learning Prosthetic

The goal of the Esper Hand is simple: design a prosthetic that can be controlled just like biological human hands.

Consider a hobby like knitting.

Typically, you would start slowly and inefficiently with the knitting and placement of the needles. However, over time, you would learn the mechanics of knitting until it became smooth, easy and effortless.

The Esper team developed its robotic hand with this in mind, focusing on creating a device that can learn from its user and become increasingly customized with use.

A series of digital signal processors, specifically electromyography sensors, currently control the device. The remaining muscles in the user’s limb control the movement of individual fingers, use different grips, and perform almost any task. Therefore, the device is not the same in every individual as it depends on the remaining active muscles for its control.

Gazda highlighted the mechanical precision of the device: “The [Esper Hand] is up to 10 times more precise in detecting muscle movement compared to most prosthetic devices.”

He mentioned that the device has faster activation and hand control, moving the bar of prosthetics closer to the reaction time of biological hands. To improve the device’s utility, it also includes mechanical protection from water and dust.

On the software front, Gazda discussed the company’s proprietary Esper Platform, which encompasses both a server and AI-powered applications. The software uses data inputs from the hand to learn the user’s habits and improve the device’s performance. For example, the hand can detect muscle activity to recognize certain situations and accurately predict the grip that would best fit a specific context, such as picking up a heavy mug or a delicate blueberry. Plus, the company’s proprietary machine learning algorithms can correct for common issues experienced by prosthetic users, such as sweat and differences in their range of motion.

The Esper Hand holding a pomegranate seed. (Image courtesy of Esper Bionics.)

The Esper Hand holding a pomegranate seed. (Image courtesy of Esper Bionics.)

“The server collects data from the hand and updates the control algorithms to fit the user’s everyday routine,” said Gazda.

The device also improves its ability to detect muscle activity over time, improving the activation, reaction time, and overall hand control. Interestingly, users can remotely adjust the features of their devices, and Esper can send automatic setting suggestions to help the user to improve their functionality.

Beyond the hardware and software, Gazda highlighted the industrial design that went into the production of the Esper Hand. The current design notably considered the aesthetics of the final device, incorporating feedback from individuals with limb differences who were looking for something that they would be excited to wear. Gazda added that at 380 g, the Esper Hand is currently among the lightest prosthetics available on the market.

As part of its industrial design, Esper Bionics is looking to develop alternative materials for a model that can be priced for developing countries. Other organizations are also working on prosthetics for developing regions, including the Victoria Hand Project.

FDA approval of the prosthetic is currently in progress, and the company has 10 users in the New York area, with 10 more users expected by the end of 2022. Gazda considers the company to be in beta testing right now and hopes to see the device expand beyond the U.S. before long.

Nika, an Esper Hand user, playing a video game. (Image courtesy of Esper Bionics.)

Nika, an Esper Hand user, playing a video game. (Image courtesy of Esper Bionics.)

What’s Next in Electronic Implants?

Gazda emphasized that his focus is on the future of wearable technology and human augmentation. Expanding from its robotic hand, Esper Bionics is working to develop prosthetics that can assist people with limb losses below the elbow, as well as help those with lower limb losses. As such, the company was chosen to assist with efforts in Ukraine to innovate prosthetics for veterans.

But Gazda wants to look well beyond prosthetics when considering the future of Esper Bionics. The goal is to develop electronic implants to improve human health and well-being. Instead of the Neuralink approach of integrating directly with the central nervous system, Esper is focused on integrating with the peripheral nervous system to improve the utility and accessibility of implants.

“Humanity as we know it is 150,000 years old. We have made major advancements in infrastructure in terms of transportation, buildings, and more. But this is the first time we can advance humans directly with technology,” said Gazda. “When we look back in 10,000 years, there will be a clear divide in the evolution of humans and a shift in our thinking about technology.”

Gazda added that in his opinion, electronic implants in humans will have a bigger impact on humanity than the automotive or space industries.

Although Esper Bionics is still at least five years away from implanted devices, the company is actively developing Esper Control, a wearable brain-computer interface device. All the devices in development will utilize the Esper Platform to help the products customize to each user’s individual habits.

It will be exciting to see how the company further develops the robotic hand and the other devices in its R&D pipeline over the next few years.