News


AI on a chip

26 June 2019 News Editor's Choice

There is still no solution which makes it possible to execute and train neural networks on embedded systems, such as microcontrollers. Nevertheless, it can be useful to conduct the training directly in the embedded system, for example, when an implanted sensor is to calibrate itself.

There is still no solution which makes it possible to execute and train neural networks on embedded systems, such as microcontrollers. Nevertheless, it can be useful to conduct the training directly in the embedded system, for example, when an implanted sensor is to calibrate itself.

The vision is sensor-related AI that can be directly integrated in a sensor system. A team of researchers at Fraunhofer IMS has made this vision a reality in the form of AIfES (Artificial Intelligence for Embedded Systems), a machine learning library programmed in C that can run on microcontrollers, but also on other platforms such as PCs, Raspberry PI and Android.

AIfES demonstrator for handwriting recognition. Numbers written by hand on the PS/2 touchpad are identified and output by the microcontroller.
AIfES demonstrator for handwriting recognition. Numbers written by hand on the PS/2 touchpad are identified and output by the microcontroller.

The library currently contains a completely configurable artificial neural network (ANN), which can also generate deep networks for deep learning when necessary. An ANN is an attempt to mathematically simulate the ­human brain using algorithms in order to make functional contexts learnable for the algorithms. AIfES has been optimised specifically for embedded systems.

“We’ve reduced the source code to a minimum, which means the ANN can be trained directly on the microcontroller or the sensor, i.e. the embedded system. In addition the source code is universally valid and can be compiled for almost any platform. Because the same algorithms are always used, an ANN generated for example on a PC, can easily be ported to a microcontroller. Until now this has been impossible in this form with commercially available software solutions,” says Dr. Pierre Gembaczka, research associate at Fraunhofer IMS.

Protection of privacy

Until now, artificial intelligence and neural networks have been used primarily for image processing and speech recognition, sometimes with the data leaving the local systems. For example, voice profiles are processed in the cloud on external servers, since the ­computing power of the local system is not always adequate.

AIfES demonstrator for handwriting recognition. All functions have been integrated on the Arduino UNO, which reads the sensor values of the touchpad, performs number recognition and outputs the result to the display.
AIfES demonstrator for handwriting recognition. All functions have been integrated on the Arduino UNO, which reads the sensor values of the touchpad, performs number recognition and outputs the result to the display.

“It’s difficult to protect privacy in this process, and enormous amounts of data are transmitted. That’s why we’ve chosen a different approach and are turning away from machine learning processes in the cloud in favour of ­machine learning directly in the embedded system. Since no sensitive data leave the system, data protection can be guaranteed and the amounts of data to be transferred are significantly reduced,” says Burkhard Heidemann, embedded systems group manager at Fraunhofer IMS.

“Of course it’s not possible to implement giant deep learning models on an embedded system, so we’re increasing our efforts toward making an elegant feature extraction to reduce input signals.” By embedding the AI directly in the microcontroller, the researchers make it possible to equip a device with additional functions without the need for expensive hardware modifications.

Reducing data

AIfES doesn’t focus on processing large amounts of data, instead, transferring only the data needed to build very small neural networks. “We’re not following the trend toward processing big data; we’re sticking with the absolutely necessary data and are creating a kind of micro-intelligence in the embedded system that can resolve the task in question. We develop new feature extractions and new data pre-­processing strategies for each problem so that we can realise the smallest possible ANN. This enables subsequent learning on the controller itself,” Gembaczka explains.

The approach has already been put into practice in the form of several demonstrators. For example, the research team implemented the recognition of handwritten numbers on an inexpensive 8-bit microcontroller (Arduino Uno). This was made technically possible by developing an innovative feature extraction method.

Another demonstrator is capable of recognising complex gestures made in the air. Here the IMS scientists have developed a system consisting of a microcontroller and an absolute orientation sensor that recognises numbers written in the air. “One possible application here would be operation of a wearable,” the ­researchers point out.

“In order for this type of communication to work, various persons write the numbers one through nine several times. The neural network receives this training data, learns from it and in the next step identifies the numbers independently. And almost any figure can be trained, not only numbers. “This eliminates the need to control the device using speech recognition: the wearable can be controlled with gestures and the user’s privacy remains protected.

There is a huge range of potential applications of AIfES. For example, a wristband with integrated gesture recognition could be used to control indoor lighting. And not only can AIfES recognise gestures, it can also monitor how well the gestures have been made. Exercises and movements in physical therapy and fitness can be evaluated without the need for a coach or therapist. Privacy is maintained since no camera or cloud is used. AIfES can be used in a variety of fields such as automotive, medicine, smart home and Industry 4.0.

Decentralised AI

The AIfES library makes it possible to decentralise computing power, for example by allowing small embedded systems to receive data before processing and pass on the results to a superordinate system. This dramatically reduces the amount of data to be transferred. In addition, it’s possible to implement a network of small learning-capable systems which distribute tasks among themselves.

Deep learning

AIfES currently contains a neural network with a feedforward structure that also supports deep neural networks. “We programmed our solution so that we can describe a complete network with one single function,” says Gembaczka. The integration of additional network forms and structures is currently in development. Furthermore the researcher and his colleagues are developing hardware components for neural networks in addition to other learning algorithms and demonstrators.

Fraunhofer IMS is currently working on a RISC-V microprocessor which will have a hardware accelerator specifically for neural networks. A special version of AIfES is being optimised for this hardware in order to optimally exploit the resource.

For more information visit www.fraunhofer.de





Share this article:
Share via emailShare via LinkedInPrint this page

Further reading:

Locked down but not knocked out
25 March 2020, ExecuKit , News
As a business, we are facing extremely hard times with the current worldwide pandemic and isolation. We are all maintaining social distancing and must learn how to do business with an enforced lockdown ...

Read more...
From the editor’s desk: A new era – after mourning comes rebuilding
29 April 2020, Technews Publishing , News
When I sit down to write these columns once a month, even if the topic I decide to write about is not all sunshine and roses, I always try and look for a silver lining. I’m simply not able to do that ...

Read more...
Personality profile: Chris Viveiros
29 April 2020, Otto Wireless , News
The operations director of Otto Wireless Solutions shares his background, experience, advice and thoughts on the future.

Read more...
Locked down but not knocked out
29 April 2020, ExecuKit , News
We as a company salute the component suppliers who have supported us and are there to assist during this time even if the requests we pose are not always realistic.

Read more...
Electronics industry tackling the coronavirus
29 April 2020 , News
2020 has been a crazy year already. From Australian bushfires to climate emergencies to US trade tariffs, the start of the decade hasn’t been very positive. Not only this, within the last few months there ...

Read more...
SqwidNet becomes first Sigfox-accredited lab in southern hemisphere
29 April 2020, iSERT , News
iSERT (Safety, EMC, RF, Testing), a leader in the provision of turnkey testing solutions for EMC, RF and safety testing, has collaborated with SqwidNet, the only licenced Sigfox operator in South Africa, ...

Read more...
Clearing the Static: ESD cleaning and hygiene products
29 April 2020, Altico Static Control Solutions , News
In light of the current COVID-19 pandemic, Actum Group would like to remind electronics manufacturers of our range of antistatic cleaning agents, cleanroom materials and PPE.       Cleanroom garments and ...

Read more...
Jemstech continuing its growth path
29 April 2020, Jemstech , News
Since its inception in 2005 and especially during the past five years, Jemstech has grown from a medium-sized electronic manufacturing services (EMS) provider into one of the major players for electronics ...

Read more...
Company profile: Phoenix Contact
29 April 2020, Phoenix Contact , News
Phoenix Contact was founded in 1923 in Essen, Germany, by Hugo Knümann. It entered the South African market in 1969 through a local distributor, and was later established as a wholly owned subsidiary ...

Read more...
In memoriam: Vivienne Caroline Dorrington - 19 June 1958 – 21 April 2020
29 April 2020, Technews Publishing , News
It is with great sadness that we announce the death of our CEO, Vivienne Dorrington, after a long and brave battle with cancer. Like her father, Ray Beaumont, who founded Technews Publishing in 1985, ...

Read more...