- Products
- Product Reviews
- Product How Tos
- New Product Releases
- Product Categories
Product Review
Microchip elevates HMI design with world's first e-field-based 3D gesture-recognition chip
Stephen Evanczuk11/13/2012 8:42 AM EST
Comment
Stephen.Evanczuk
It would be interesting to see application of this technology in those areas. I ...
Halibut
I see this new technology having a massive impact in a number of applications ...
Microchip Technology has unveiled its GestIC gesture-based non-contact technology with its announcement of the world's first electrical-field-based 3D gesture controller IC, the MGC3130, enabling a new class of non-contact user interfaces. Based on technology acquired with Microchip's purchase earlier this year of Germany-based Ident Technology, the MGC3130 offers always-on gesture recognition with minimal power requirements, permitting its use in battery-powered mobile and portable designs.
At its most basic level of operation, the MGC3130 is capable of detecting motion at 200 samples/s and 150 dpi resolution within a sensing area defined in the x-y plane by sensing electrodes and extending in the z plane from 0 to 15 cm above the sensing area. Beyond this more basic "3D touchpad"-like capability, however, the MGC3130 goes further - using its on-chip software library to analyze its acquired x/y/z motion data and deliver a real-time stream of recognized 3D gestures like hand flicks, circular motions, and the like to a host MCU.

Figure 1. The MGC3130 IC uses a simple planar sensor array to read disturbances in an e-field, finds x/y/z position, and streams recognized gestures to a host MCU.
Combined with external electrodes and a host MCU, the MGC3130 combines both hardware and software to offer a complete 3D gesture recognition solution with minimal additional external components (Figure 1). Delivered in a 5 x 5 x 0.9 mm 28-lead QFN package, the mixed-signal device integrates a signal processing unit, 32 KB flash memory, 12 KB RAM, one Tx channel, and five Rx channels (Figure 2).

Figure 2. The MGC3130 mixed-single IC combines a complete motion sensing analog signal chain with firmware for position detection and gesture recognition.
The device offers I2C and SPI interfaces for streaming sensor x/y/z data or post-analysis gesture information to a host MCU. The new device consumes only 9 μA in deep sleep mode, 45 μA in programmable self wakeup mode, and 30 mA in processing mode (all typical values at 3.3 V).
Tx signal generation uses frequency hopping between 70-130 kHz to find the optimal operating frequency in the presence of noisy environmental conditions. The five identical Rx channels each include a signal conditioning circuit that filters and amplifies the analog signal before driving it to a dedicated ADC.
The GestIC e-field approach detects disturbances in a transmitter-maintained e-field (top figure) due to the presence of a conductive object such as the human hand (bottom figure). The MGC3130 IC uses one transmitting electrode to generate the e-field, and five receiving electrodes positioned around the periphery of the sensing field.
Hand movement within the 3D sensing area results in a characteristic change in the e-field that can be measured through the receiving electrodes and used to identify the origin of the e-field distortion. With this information, the 3D gesture controller can calculate object position, track its movement, and identify pre-determined patterns of movement such as hand gestures.
GestIC's e-field technology operates in closer proximity than camera-based gesture-recognition methods such as the Microsoft Kinect. Unlike camera-based systems, however, GestIC's approach results in no blind spots close to the gesture stage. Furthermore, the use of e-field technology ensures that ambient light or sound have no effect on gesture recognition. GestIC technology avoids interference with other radio sources by frequency hopping within a radio range of 70-130 kHz - outside the range of typical consumer radio sources.
The result is a solution that is both highly sensitive to intended stimuli and highly immune to noise sources and not affected by clothing, surface texture, or reflectivity. At the same time, the use of low-cost electrode materials minimizes system cost. In fact, designers can use existing capacitive sensor structures such as a touch panel’s ITO coating.
At the heart of the MGC3130's implementation of the GestIC technology, Microchip's on-chip Colibri software performs sensor data analysis and gesture recognition. Stored in the MGC3130's internal 32 KB flash, the Colibri software library uses a Hidden Markov Model for recognizing gesture patterns similar to that used in voice recognition. The initial release of the Colibri library identifies predefined hand gestures including flick, circular and symbol gestures. Symbol gestures are multi-directional gestures with a defined start and end point - such as a “check mark” hand movement.
A combination of on-chip hardware- and software-based power management features enable "always-on" gesture detection while maintaining low-power operation for battery-powered applications. In this method, called "approach detection," Colibri software uses the device's built-in self-wakeup mode to alternate between sleep and scan operation. When an object in the e-field exceeds specified threshold values, the MGC3130 will switch to processing mode and wake up the host MCU as needed.
Within the approach detection sequence, the device performs an approach scan - typically requiring only one Rx channel to help minimize power consumption. Engineers can set the time between approach scans, typically in a range from 20ms to 150 ms for most applications. Although the device automatically handles temperature compensation, the MGC3130 can also perform a calibration scan - required for applications using basic x/y/z position data. Here, the device activates several Rx channels to calibrate sensor signals. To reduce power consumption, engineers can reduce the interval between calibration scans during periods with low user activity for example.
For MGC3130 evaluation and development, Microchip offers the Sabrewing Single-Zone Evaluation Board kit, which contains MGC3130 reference circuitry and two sets of selectable frame electrodes (5” and 7”). The kit also comes with Microchip's Windows-based AUREA control software, which includes MGC3130 real-time sensor data display, 2D and 3D visualization of position, visualization of recognized gestures, AFE parameterization, and GestIC library loader.
Samples of the Microchip MGC3130 are available now in a 5x5 mm 28-pin QFN package, and volume production is expected in April 2013 with pricing at $2.26 each in high volumes. The Microchip Sabrewing MGC3130 Single Zone Evaluation Kit is available now for $169. The AUREA graphical user interface is available now via a free download.
For additional information, visit Microchip.
Microchip has provided a video demo of GestIC technology and the MGC3130 in action:
If you liked this and would like to see a weekly collection of related products and features delivered directly to your inbox, click here to sign up for the EDN on Systems Design newsletter.
Most Popular
Datasheets.com Parts Search
185 million searchable parts
(please enter a part number or hit search to begin)
Our technical library houses over 4,000 high-quality sponsored white papers, application notes, reference guides, use cases—all organized by company.



mykl who
11/13/2012 10:53 AM EST
This first-of-its-kind technology is destined to blossom into applications that have been patiently waiting for it to arrive. The future is almost here!
Sign in to Reply
Stephen.Evanczuk
11/13/2012 11:26 AM EST
It really is quite intriguing. I'm guessing it won't be long before we see light switches that you wave on and off. What What sort of new applications can you imagine for this kind of technology?
Sign in to Reply
edsfam
11/13/2012 3:57 PM EST
Wave your hand down and the lights go off ... and the toaster goes down ... and the coffee maker starts. Wave your hand up and the windows open ... the microwave door opens ... and the TV switches to Satellite ... but the lights do go on. Hooray for technology.
This should be interesting.
Sign in to Reply
eafpres
11/13/2012 9:18 PM EST
This may have many applications in control systems UI, switch/slider replacement, knob replacement, etc. Possibly in aircraft, consoles (such as power plants, factory automation, building automation, music (mixing panels)) etc. Hard to guess the total price including the sensor, and the x-y size limitations (they mention 5" and 7"). Also hard to tell if it will really compete w/touch technology, but it might be much cheaper and the 3D aspect might be useful--however people may not remember complex 3D guestures, so how much complexity can be used is a question.
I think the comparison to Kinect is not relevant--these are really for totally different applications.
I note their diagrams are misleading--the stated z range is 15 cm (6 inches), but in both diagrams they show interaction with the arm out to at least 20 inches or more.
Anybody know what SAR limits are at 70-150kHz?
Sign in to Reply
Stephen.Evanczuk
11/14/2012 7:48 AM EST
The x-y sensing area of 5" and 7" mentioned here relates to their eval board, and my understanding is sensing areas at least twice that size are easily accommodated.
I agree about people remembering gestures, and I think Microchip (and Ident Tech before them) recognized that: The Colibri gesture-recognition library is purposely kept to maybe a dozen or so recognized gestures. Beyond that, I'm guessing early products using free-space gestures will find people touching the product. It's going to take a while to learn you don't have to touch it to operate it, I think.
As far as existing HMI devices, Microchip told me that don't necessarily see this replacing existing HMI such as touchpads, but rather complementing them. One of the scenarios they mentioned is using it with a touchpad where you place your index finger on a touchpad and move your (other) hand in the 3D sensing area. The center of mass of the hand is a second point, providing two points of reference in three dimensions for 6-axis control.
Sign in to Reply
Halibut
11/17/2012 5:30 AM EST
I see this new technology having a massive impact in a number of applications beyond the obvious ones already listed, especially medical / Lab use - being able to control and interact with equipment without any chance of contamination; Robotics - both control and sensory applications; industrial controls in harsh environments; security / access control - using a sequence of gestures; Control of various equipment by the elderly or infirm / disabled ..... to name but a few. I think we will see a whole new generation of HMI as a result of this development, and no doubt further enhancement will emerge from Microchip in the future. Bravo!
Sign in to Reply
Stephen.Evanczuk
11/19/2012 12:34 PM EST
It would be interesting to see application of this technology in those areas. I asked Fanie Duvenhage, the Director of Microchip's Human-Machine Interface Division, about its application for people with disabilities - specifically hand tremors - and they said: "It's hard to say. It could be problematic. Most of the gestures we've done tend to be larger movements rather than subtle movements, so I'm not sure."
Keep in mind that the gesture recognition system works using the Hidden Markov Model method, where gestures from several users are sampled to build a recognition set. There are always going to be some errors in recognition that are a combination of using an inappropriate sample set (by this, I mean a sample population that deviates significantly from the target user population), plus differences between an individual user and the sample set (e.g., let's say my hand is much smaller than those used in the sample set and I move my hand much more quickly). Add to this the "noise" of a hand tremor, and it's fair to say the software would need quite adaptive. Maybe adaptive recognition would be ok for speciality medical apps, but I think we've seen that most users aren't keen in waiting for their voice-recognition system (which probably uses the same HMM method) to understand them.
Question to anyone: Would you be willing to train a gesture-recognition system? Or would you expect it to understand your (basic) gestures out of the box?
Sign in to Reply