Beach Innovation: The free soul of electromyography gesture recognition

Compared to Kinect and Leap Motion for image sensing, MYO, as the world's first device to apply electromyography technology to the field of gesture recognition, is a clear stream in the somatosensory world and has gained a lot of popularity in 2013.

In fact, myoelectric technology had some research progress as early as the 1970s, but it is mainly used in the medical field and requires the insertion of probes on the skin. Until the development of the sEMG (Surface Electromyography) in the 1990s, monitoring with the skin-contacting electrodes prevented people from suffering from skin damage.

At the same time, researchers are also trying to find more applicable areas of sEMG. For example, MIT has successfully applied electromyographic techniques to prosthetic limbs; the field of aerospace is also a new area of â€‹â€‹trial: Since space is inconvenient to move, spacecraft Can a person use sEMG technology to perform various operations through gestures? The answer is of course "YES".

While working at NASA, Li Wei, founder and CEO of Beech Innovation, a graduate of the University of Magdeburg in electrical engineering, is responsible for the interactive development of the astronaut ground training system. The system mainly conducts ground training for astronauts through VR, and the interactive system uses the myoelectric recognition technology.

Li Duo was distressed at the fact that new technologies were limited to the research stage but could not be better promoted in Germany. He decided to resign from the work of the German Aerospace Administration and chose to go back to the country. After returning home, he had two undergraduate roommates Liu Hancheng and Li Zhenhua. This project will miniaturize surface muscle electrical signal recognition technology and â€œconcentrateâ€ it on a small bracelet.

The Beach Innovation was established in November 2015, but prior to this, the project team has been under the support of Xi'an High-tech Zone for more than a year and has obtained the investment and support from the Chinese Academy of Sciences. One of the partners, the companyâ€™s brand leader Zhang Xuanxuan told Lei Feng, because the interactive operating system was Li Yongâ€™s research and development in Germany, so there was no dispute on intellectual property rights.

"Everybody is getting together. On the one hand, we are optimistic about the project itself. On the other hand, we also hope that we can achieve some breakthroughs and innovations through our own technologies."

What Beache Innovation wants to do is, in simple terms, to free up hands and make interaction a more natural and free experience.

EMG's free soul

Hearing is a psychic beast under the Dizang Bodhisattva. It can identify all things in the world through listening, and is especially good at listening to people. The DTing developed by Beechcraft Innovations comes from the transliteration of Heyin.

"In myths and legends, Hearing can receive all information in the universe and at the same time pass this information to believers in order to distinguish right from wrong." Li Wei introduced the origin of DTing to Lei Fengwang (searching for "Lei Feng Net" public number attention) .

DTing can monitor the corresponding muscle groups of the traction fingers through sensors wrapped on the forearm of the human body, and rely on the core algorithm to translate and convert them into corresponding signal outputs, and can accurately perform joint motion judgments.

Although EMG technology has developed for such a long time, it has not ushered in explosive growth. Zhang Xuanxuan said that compared with technical constraints, he believes that the demand has not yet reached accumulation. Just as Apple has not yet introduced the mouse into computer operation, everyone is only satisfied with using the keyboard; before the iPhone was released, everyone did not feel that there was something wrong with the function machine. In the field of interaction, the demand that is closer to nature has not yet reached the point where it is necessary to instigate product replacement.

There are three more popular approaches to the interactive field of somatosensory: myoelectricity, image sensing, and mechanical structure pressure. For image-sensing gesture recognition, the inevitable problem is its mobility and range of recognition. In addition, it requires the CPU and operating memory of the connected device to be used, which will increase the burden on the connection level.

"Image sensing as a fixed space may be a very good solution, but it does have a lot of inconvenience for gesture recognition. Image sensing does not distinguish between right and left hands, right and left hands cross or block, and there is no way to identify it, but myoelectricity There is no such restriction. At the same time, EMG recognition is not affected by the intensity of external light and the proximity of operation."

For mechanical sensing, the hugeness of the device itself becomes a barrier to mobility, and it cannot even be used outdoors. Compared to the other two identification methods, EMG has a slightly lower power consumption and mobility.

Wearing a chop dance

Although EMG has a free soul, it also has many objective limitations. Like the dancer who needs to wear a pimple, how can it be optimized in the existing situation? Muscle electrical signal itself is relatively weak, it is relatively difficult to collect and eliminate interference, coupled with EMG gesture recognition is very sensitive to humidity, sweating will inevitably affect the experience.

And each person's muscle condition is different, the signal that produces will be different, and turn the muscle signal into the instruction, also need the complicated core operation. The DTing device integrates the three steps of acquisition, operation, and issuing of instructions. It can be operated without an external host. This improves the mobility.

"Transmission is sent directly to the already processed command signal. It does not need to link the computer or the host computer like image sensing. It needs to occupy the computing space and memory of the computer or the host computer."

At present, the prototype can recognize more than 20 kinds of established gestures, each of which is more than 90%, but the machine acquisition process will be slightly longer. As for the unavoidable humidity problem, Li Xin believes that in the known technology areas, there is room for improvement in EMG. At present, they are actively solving this problem.

In October this year, DTing plans to launch a crowdfunding platform at home and abroad, pricing it at the thousand-dollar level. The Beechcraft innovation tells Lei Feng Net that they will launch consumer and developer versions, and also provide a full set of interactive solutions for the docking industry vendors, hoping to bring innovation to the industry's interactive experience.

In the interaction field, especially in the AR/VR interaction, EMG has a high reduction advantage in force sensing. In the sports health field, DTMing can also analyze the muscle status during exercise through EMG monitoring. Can assist in the treatment of muscle rehabilitation and exercise training. In addition, myoelectric can control the prosthesis, and even help the deaf and mute person to achieve the role of sign language interpreter. I believe in the future, DTing can bring us more surprises through muscle electrical energy.