Tools


Siemens Worldwide

Pictures of the Future

Contact

Contact

sts.components.contact.mr.placeholder Sebastian Webel
Mr. Sebastian Webel

Editor-in-Chief

Tel: +49 (89) 636-32221

Fax: +49 89 636-35292

Werner-von-Siemens-Straße 1
80333 Munich

Pictures of the Future
The Magazine for Research and Innovation
 

Autonomous Systems

Tapping the Power of Artificial Intelligence

Among the many interesting threads of thought investigated by researchers at the Siemens Artificial Intelligence (AI) hackathon was the question of how a new class of machine learning known as deep learning can contribute to aspects of perception, cognition, and decision making in intelligent systems – including computer vision, audio analysis, language understanding, and virtual reality technology.

In the course of three days and two sleepless nights, 13 groups of Siemens hackers developed artificial intelligence (AI) demonstrations that might have taken months to develop in a more structured environment.

Artificial intelligence (AI) is a rapidly developing technology area with application domains ranging from robotics and self-driving vehicles to health care and financial support. “Significant breakthroughs in AI, especially in object and speech recognition, reasoning and natural language processing have mainly been driven by Deep Learning technologies, posing new challenges and opportunities for industrial digitization strategy,” says Dr. Ulli Waltinger, organizer and head of the Machine Intelligence Group at Corporate Technology, Siemens’ global research organization.

As Anthony Wing Kosner has pointed out in Forbes, these developments have profound implications. “The word is spreading in all corners of the tech industry that the biggest part of big data, the unstructured part, possesses learnable patterns that we now have the computing power and algorithmic leverage to discern – and in short order. The effects of this technology will change the economics of virtually every industry.”

A team that focused on distributed deep learning on small devices developed a physical model representing buildings on which to test its concepts.

Vast Market

According to Research & Markets, a major market research company, the artificial intelligence market is expected to grow from $419 million in 2014 to over $5 billion by 2020, at a compound annual growth rate (CAGR) of over 53 percent.

With a view to establishing an interdisciplinary innovation environment in which AI prototypes can prove their potential and thus generate new ideas, Siemens recently held a hackathon – a marathon programming competition – at its Global Leadership Center in Feldafing near the shores of Lake Starnberg in southern Bavaria. Over the course of three days – essentially a maximum of 48 hours of work counting sleepless nights – thirteen teams of hackers developed a range of AI-focused demonstrations.

Technology Fusion

The winning demonstration was the “Automatic Model Train.” This entry represented, in the view of the hackathon jury, an excellent demonstration of how different technologies – AI, computer vision, sensor systems and sensor fusion – can come together in a mobility environment. Using a model train as an example, the team outfitted it with a computer vision system that successfully detected obstacles, estimated their distance, and stopped the train in response – even in simulated fog.      

One of two runner-up teams focused on “Distributed Deep Learning on Small (Embedded) Devices." The team’s demo illustrated how deep neural network architectures can support distributed learning by clusters of small devices. The hackathon’s other runner-up Team focused on “Artificial Intelligence in Virtual Reality Simulated Physical Systems.” The team’s demonstration was designed to simulate physical systems in virtual reality in order to train smart agents such as drones to learn how to anticipate problems in complex installations such as wind parks. The winning team received a check for €10,000, while each of the runners-up received a check for €7,500.

Amazing Speed

Commenting on the wide range of the Hackathon’s entries, Michael May, head of Business Analytics and Monitoring at Siemens Corporate Technology, said, “I was impressed with the originality and depth of the entries, and am confident that many of them will go on to become projects, prototypes and perhaps even products. It is truly amazing to see how quickly these entries were developed. Comparable projects on the European level can take months to achieve similar results. This really confirms the value of our hackathons.”

Hackathon organizer Ulli Waltinger with all of the event’s participants. Many of the demonstrations developed at the hackathon are expected to become projects, prototypes, or even products.

Added Norbert Gaus, Head of Research in Digitalization and Automation at Siemens Corporate Technology, “Our hackathons are not only a very inexpensive way of getting started with new technologies, but a very efficient and fun way of doing so with people from across the Siemens spectrum.”

Given the quality and originality of the demos, Gaus and May announced that they would sponsor a follow-up event during fiscal year 2017 and that the current group of entries would appear in a showroom context. “There is no better way of demonstrating the power of artificial intelligence than through our hackathons,” said May.

Arthur F. Pease