Coapt Engineering Gen2 Pattern Recognition

Coapt Gen2 Feature Image

Gen2, from Coapt Engineering, is a complete, robust, and proven myoelectric pattern recognition system. Based on decades of research and years of product experience, Gen2 significantly enhances user control over upper-limb bionic devices.

A Quick Look at the Coapt Gen2

The following video provides a brief overview of the Gen2 pattern recognition system:

Before reading further, a little background information may be helpful. To obtain this, please see our master article Myoelectric Pattern Recognition for Bionic Arms & Hands, which includes a basic description of how pattern recognition systems work.

As that article indicates, there are a few key areas crucial to the success of any myoelectric pattern recognition system. These are:

  • basic technology;
  • evaluation support;
  • setup & calibration;
  • training;
  • recalibration by the end-user.

We will examine Gen2 in each of these areas.

Basic Technology

There is a spectrum of pattern recognition technologies:

Pattern Recognition Spectrum

A simple system may use some form of pattern recognition or artificial intelligence to clean up myoelectric signals but its focus is still on basic bionic hand actions like open and close. We like to think of this mainly as an enhancement to direct control.

Universal systems offer a more complete system of pattern recognition, one that can map patterns of muscle movements to various bionic commands, including the movement of multiple bionic joints at the same time. These systems are called “universal” because they are capable of interfacing with different types of bionic components from different manufacturers.

Coapt’s Gen2 is a universal system. It is meant to be used with eight sensors and can be combined with bionic components from different manufacturers to form an overall solution.

Because Gen2 must support the functionality common to these components, it does not attempt to match the advanced capabilities of some of the more experimental systems on the horizon, which may allow spontaneous control over individual bionic finger movements.

Instead, Gen2 supports the movements allowed by whatever components it is attempting to control.

How does Gen2 work? Let’s walk you through the typical chronology of its evaluation, configuration, and subsequent use.

Evaluation Support

The first step for a prospective user is to evaluate whether he can benefit from Gen2.

Coapt supports this by supplying the Coapt Evaluation Kit. This kit is loaned but can be purchased if desired. It includes the necessary hardware and computer software for prosthetists and users to operate a virtual arm/hand (sensors, an EMG interface cable, and an evaluation cuff, if necessary, vary by solution and must be purchased separately).

Coapt Gen2 Evaluation Kit

Not only can this kit definitively answer the question of whether a user is a good candidate for Gen2. It can also be used for training purposes before the user takes delivery of a physical bionic arm/hand.

Setup & Calibration

This section assumes that the sensors have been properly placed on the residual limb, that they are connected to the main unit of the evaluation kit, and that the kit is connected to a running instance of Coapt’s Complete ControlRoom Gen2 software.

What is calibration? It is the process of mapping patterns of muscle movements to actions of the bionic arm/hand (whether real or virtual) and works like this:

  1. The Complete ControlRoom Gen2 software displays a prompt for the user to perform an action with a bionic arm. In this case, the application is asking the user to flex his arm up (i.e. bend the elbow) and hold:
Coapt Gen2 Calibration Screen
  1. The user performs the requested action. Note, in a virtual environment, the user attempts to move his phantom arm/hand, which triggers a pattern of muscle movements in the residual limb. The resulting set of myoelectric signals detected by the sensors might appear like this:
Coapt Gen2 Pattern Recognition Signals 02
  1. This pattern of signals is then mapped to the requested action and remembered as such. When the user repeats the pattern (or one similar to it), the Gen2 control system will issue the necessary commands to make the virtual or real bionic device perform the mapped action.

The calibration process is augmented by artificial intelligence algorithms that do the following:

  • Rate the quality of your calibrated motions. A 5-star rating means that you are getting good, reliable control for the action in question. Lower ratings mean that you may need to improve your control. Fortunately, the Control Coach software provides feedback on how to do this.
  • The AI algorithms learn over time. The more calibrations you perform, the more the system refines its mapping between your movements and your desired actions.

TRAINING

Control Coach can be used for user training/practice.

Coapt Gen2 Control Practice

The recommendations that it provides can help end-users improve the quality and consistency of their patterns of muscle movement. This is the end goal of all pattern recognition training processes.

That having been said, Coapt does not appear to have put as much emphasis on training as its main competitor, Infinite Biomedical Technologies, does with its Sense pattern recognition system. In particular, the game-playing approach used in IBT’s Myotrain software seems likely to encourage more practice.

However, we do emphasize the word “seems” here. Until we get sufficient end-user feedback on actual outcomes, we can’t say which training system is superior.

RECALIBRATION

One of the challenges of myoelectric control systems, especially those dependent on skin-surface sensors, is that the state of a user’s residual limb is not constant. It can change with things like temperature (swells when hot, shrinks when cold), humidity (skin sweat), and muscle fatigue.

When this happens, users need the ability to quickly recalibrate their pattern recognition system. Coapt takes a unique approach to this issue, allowing users to perform recalibration without the need for a software interface, as demonstrated in this video:

We think this is a really smart way to perform recalibration.

Clinical Insights

Add

The following comments are from clinicians with extensive experience with Coapt’s Gen2 Pattern Recognition System:

…Coming Soon

User Feedback Survey & Results

Are you or have you previously been a Coapt Gen2 user? If so, please share your insights with others looking at Gen2 as a possible pattern recognition system.

Survey

Results

We do not yet have a sufficient number of survey participants to publish fair and accurate results for the Coapt Gen2 system.

As soon as we do, we’ll update this section.

Considerations Before Buying Coapt’s Gen2 System

Without sufficient feedback from independent users, we’re not yet able to form a complete opinion about Coapt Gen2.

However, Coapt does seem to have checked all the boxes when it comes to delivering a complete myoelectric pattern recognition system.

The real question for prospective users is whether to opt for proven pattern recognition systems like Coapt’s Gen2, IBT’s Sense, or Ottobock’s Myo Plus now, or to wait until new devices from Atom Limbs, BrainRobotics, and Esper Bionics come to market with promises of even more advanced pattern recognition systems.

As it stands now, it will likely be a few more years before all those new devices are on the market and another year or two after that before we get sufficient user feedback to know if they have delivered on their promises.

In other words, a lot of this will come down to each prospective user’s purchase window and risk tolerance.

Related Information

If you are considering a dual-site, direct-control myoelectric system, do not miss our article on Finding the Right Myoelectric Control System.

For a comprehensive description of all current upper-limb technologies, devices, and research, see A Complete Guide to Bionic Arms & Hands.

Click here for more information on Coapt Engineering.

Follow Us