A 10-year-old Irish technology company called Movidius has just struck a lucrative deal with Google to bring human vision processing and a lot more to future smartphones, drones and robots. And it all began in Dublin.
On a summer morning in 1999, tech investor and entrepreneur Brian Long, the then-CEO of a company in Dublin called Parthus Technologies, invited me into the company’s office on Harcourt Street. Arrayed on a table surrounded by a team of engineers were concepts and mock-ups of exotic-looking devices that could feature far away in the future.
There was a fibre glass mockup of a future mobile phone with video and email on the screen – what we know today as a smartphone – and a curious looking device that Long said could potentially hold thousands of songs.
Remember this was three years before the iPod was launched, MP3s were still in their infancy and portable music in those days came via the discman.
Among the assembled engineers were David Maloney and Sean Mitchell, founders of a company called Movidius that today is enabling Google to create the next generation of machines with cognitive reasoning.
‘Many people see the machine age as the next revolution in technology, similar, in respects, to the rise of the internet and the explosion of the mobile industry. This is a whole new paradigm’
– SEAN MITCHELL, MOVIDIUS
Movidius, which is creating 100 artificial intelligence jobs in Dublin, has just struck a major deal with Google to ship its MA2450 chip in upcoming personal devices.
Recalling our long-ago meeting, Mitchell, now chief operations officer of Movidius, tells me that the concept I saw around the music player he had designed and demoed that sunny morning went on to form the guts of Apple’s iPod Shuffle, which emerged in 2005.
In 2002, Parthus merged with Israeli technology company CEVA and today it is a $60m-a-year electronics player and the world’s No 1 licensor of digital signal processor technology.
Adding up opportunities
After Parthus, Maloney went back to university to study a PhD focused on high-performance computing problems.
“We could see new applications coming on the consumer and mobile side that were going to be a challenge for standard architecture to keep up with. David created an accelerator to bring caching data structures from almost zero to 80pc.
“That was the germ of the original concept and we looked at game physics, video editing and human vision processing and we realised there would be a huge class of apps that would require linear algebra and algorithms.”
Mitchell explains that, in the world today, Google is leading the charge when it comes to creating computers with deep learning capabilities and machine intelligence.
“They are adopting our Myriad 2 family of vision processors to bring machine intelligence to personal devices like smartphones.”
‘Some people don’t like the term ‘internet of things’ but the reality is it is smart everything – cities, cars, buildings – and it is going to bring challenges but also enhance lives’
– SEAN MITCHELL, MOVIDIUS
Movidius’ chip technology has been at the centre of developments by Google to create Project Tango smartphones that can sense their immediate environment.
For example, a Project Tango smartphone with a Movidius chip will be able to use its sensors to calculate the entire dimensions of a room, thereby enabling the device to see its surroundings.
“The key thing to understand is that our chips enable devices to get this intelligence without having to connect to the cloud or the internet. This means machines are moving to a much more devolved intelligence with a low latency ability to recognise objects, for example.”
Mitchell said that the deal with Google will soon be followed by similar deals with device makers in the areas of robotics, drones and virtual reality, some of whom are household names.
“We have announcements in the pipeline for new products and categories, including augmented reality and virtual reality devices.”
Mitchell said that machine intelligence will transform a range of markets where visual intelligence is key, including security.
“Our goal is to be the de facto platform to bring visual intelligence across a range of markets.”
Sweating Moore’s Law
Mitchell said that areas like machines that could think or internet of things weren’t on the horizon when he and Maloney started Movidius.
“We were seeing the growing number of applications becoming extremely demanding in terms of the scale of the data processing that was going to be required, devices with a number of cameras with high frame rates and masses of data.
“Our original idea was to come up with a low-cost and power-efficient means of processing and managing these data flows.”
Mitchell points out that what was needed was a reevaluation of chip architecture, particularly memory data, to process vast amounts of data in a power-efficient way.”
One of the key features of Movidius’ ecosystem is that it provides the software and the hardware to create machine vision capabilities, but leaves enough room for partners to programme the chips to their own specifications.
Unable to comment on Google’s plans for machine vision devices, Mitchell said that it is a case of the right technology at the right time for the Irish company.
“When we set out with Movidius we knew we were going to go against popular wisdom. But we knew architectures were evolving, so it was precisely the right thing to do.”
Mitchell and Maloney strongly believed that chipmakers were doing something wrong. “Moore’s Law was slowing. It was shrinking on the technology side and we felt that more effort was needed on the software side of things. Processors were demanding more and more memory and software was becoming increasingly bloated.
“We were actually shocked by how little software on processors had evolved compared to the hardware.”
Recently, Movidius raised €38m in a move that will enable it to generate 100 new jobs in Dublin.
With impending announcements from major tech brands integrating its vision processor for the machine age, Movidius is on the cusp of great things.
“I think that this is a really exciting time. Many people see the machine age as the next revolution in technology, similar in respects to the rise of the internet and the explosion of the mobile industry. This is a whole new paradigm.
“Some people don’t like the term ‘internet of things’ but the reality is it is smart everything – cities, cars, buildings – and it is going to bring challenges but also enhance lives.”
On the 100 jobs Movidius is creating, Mitchell concluded: “We are scaling as aggressively as we can, constrained only by having to keep the bar high in terms of the quality of people we hire.
“The company is effectively doubling in size and we are scaling up to support a much broader range of customers and applications that require machines with vision.”