CHICAGO — Coapt LLC yesterday announced the second generation of its patented technology for controlling bionic upper limbs. Complete Control Gen2 uses machine learning and artificial intelligence to provide an interface that adapts to human users, according to the company.
The initial Complete Control product was the first myoelectric pattern-recognition system to achieve Class II clearance from the U.S. Food and Drug Administration, stated Coapt. Gen2 includes the AI-enabled Control Coach, which provides users automated feedback and instruction for more intuitive control and more natural movement.
“Everything we have engineered into Gen2 is aimed to improve our users’ quality of life,” stated Blair Lock, co-founder and CEO of Coapt. “Vast experience with users has driven the development of the Control Coach AI and improved adaptive algorithms.”
“After my accident, I lived for more than 25 years without a prosthetic device before finding Coapt,” said Coapt user Jodie O’Connell-Ponkos. “Their new Complete Control Gen2 makes it even easier for me to do whatever I need to do — train horses, cook dinner, and now, study occupational therapy to help my fellow amputees in their journeys. Plus, if I need to reset my prosthesis control on the go, I can recalibrate in just one minute, and keep moving on with my life.”
Coapt said its human-machine interface products are available now for upper-limb amputees in the U.S., Canada, Australia, New Zealand, and much of Europe. Lock answered the following questions from The Robot Report about Complete Control Gen2:
Coapt and engineering
How long did it take to re-engineer Complete Control System Gen2 in comparison with the time to develop Gen1?
Lock: Re-engineering for Gen2 has been a development that started a few years ago. The Gen1 system was built via a much different process and came to life through many years of academic research, technology development, and clinical evaluations.
What sorts of features to improve control does the system include? What sort of feedback did you receive that prompted them?
Lock: The improved features span a broad spectrum, from electronics hardware and wireless connectivity to algorithmic features for improved user function.
Similarly, feedback came in across that spectrum: Clinicians wanted more power-efficient hardware, easier connectivity and improved UI features; wearers sought more tools for addressing day-to-day control challenges and insight into their performance.
Can you give a brief explanation of how Control Coach works? How is it more intuitive than earlier technology?
Lock: The AI-enabled Control Coach is a new feature that is in addition to the intuitive real-time control system, running somewhat in parallel. The intuitive real-time control system has been improved from its Gen1 counterpart for efficiency and performance.
Control Coach monitors users’ calibration data and analyzes it for known, potentially problem-causing scenarios. It then provides suggestive and corrective feedback to users. The Control Coach AI engine was programmed with information and scenarios built from our knowledge base of hundreds of users and their co-learning trajectories.
How has Coapt incorporate recent advances in sensors, actuators, and machine learning?
Lock: The Gen2 system bases its function on high-fidelity surface muscle signal (EMG) information. To do this, the system has been engineered with the latest in small-signal differential amplification technology and recording electrode connections.
The machine learning core of the platform is kept up to date with in-house enhancements for computational efficiency and reliability.
What’s the power supply and duration?
Lock: Coapt does not produce any actuators. Power supply for the Coapt system is always the battery that is already part of a user’s prosthetic arm. Duration is most affected by how “active” each wearer is, but these prosthetic batteries are targeted to provide a full day of wear time.
How important is “plug-and-play” capability?
Lock: This is vastly important for the systems we provide it to work with. Coapt does not design or manufacture the actuated prosthetic components – there are established and upcoming firms doing a great job with that area of the wearable technology. We focus solely on the control system, and therefore need to make it widely compatible to connect to the available components on the market.
The Robot Report is launching the Healthcare Robotics Engineering Forum, which will be on Dec. 9-10 in Santa Clara, Calif. The conference and expo will focus on improving the design, development, and manufacture of next-generation healthcare robots. Learn more about the Healthcare Robotics Engineering Forum, and registration is now open.
Gen2 and users
What limitations did previous prosthetics have that no longer exist with Complete Control Gen2?
Lock: Previous prosthetic arm control systems rely heavily on switching and gross user movements that aren’t 100% intuitive, nor [are they] related to the resulting prosthetic motion. In addition, the conventional systems are “set” in place and provide no adaptation to the user as their vast control needs change.
Who can benefit from this technology?
Lock: This Coapt system is currently packaged and optimized for upper-limb prosthesis wearers. It is theoretically applicable to other applications requiring intuitive human-machine interface, such as lower limb prosthetics, power-actuated orthotics, and virtual/augmented rehabilitation tools.
Are you a Coapt Pattern Recognition user ~ Hop over to our Private Coapt community on Facebook and see our weekly Functional Friday posts and get your questions answered by your fellow Coapt users from calibration to “I wonder how to” #coaptcompletecontrol #amputee #prosthetics pic.twitter.com/ar16syLHAi
— Jodie O’Connell-Ponkos (@JodiePonkos) June 7, 2019
O’Connell-Ponkos described recalibration. How does that work?
Lock: Recalibration is a short process where the machine learning core gets new example data labeled with the types of intuitive user movement.
To acquire this data, the controller is set off on a 30-second sequence of producing its own prosthetic movements. During this time, the wearer mimics those motions — imagine the “Simon Says” game — as their EMG data is recorded and analyzed. The computation training algorithm incorporates this new data to morph the existing user model.
How does the updated product compare with other products on the market? How much does it cost?
Lock: Thus far, there have been no true comparative products on the market. In the near future, we may see a few other entrants into this exciting space. However, none of which offer the advanced features and advanced hardware platform that Coapt has achieved in its Gen2 system.
Cost varies slightly, based on the complexity of the prosthetic device it is being connected to. Typically, the Coapt system adds a 20% to 45% premium over the standalone prosthetic arm.