Tagged as “personal robotics”, Baillie’s vision of robots as watchdogs, nurses and cleaners, for example, is shared by most robotic developers today. But Baillie believes he has found the key to making it a reality. He has developed operating software that can be used in any robotics platform, in the way that MS DOS and Windows enabled personal computing 30 years ago.
Baillie’s story goes back to the Sony Computer Lab opened by the giant electronics computer some 15 years ago near to France’s elite Ecole Normale Supérieure, with its superb mathematical brains selected by the elite French Grandes Ecoles system. There, while working with Sony’s robotic dog Aibo, Baillie realised what artificial intelligence needed to progress: failure.
After all, to acquire intelligence, a human being has to go through trial and error. Why would it be different for an artificial intelligence? Of course, that means that a computer should be able to interact with its environment. But that is exactly what a computer embedded in a robot can do. “Robots,” explains Baillie, “create bridges between perceptions and motions on one hand and cognition with all its symbols manipulation processes on the other.”
Baillie also discovered that this kind of research can take years to develop in the relatively isolated environment of a lab. The processes could be vastly accelerated by commercialising key technologies to enable a community of robotic developers to share and improve their software discoveries, just like in an open source environment, and to create an even bigger community of robots learning by doing. So for Baillie the quickest step to personal robotics was to found a commercial company able to serve those communities.
“While working with Aibo I realised that the built-in software interface limited the implementation of new robotic capacities,” Baillie recalls. “Students have to go through a long education process before they start to be able to program their own new robotic functions.” Going back to the drawing board, Baillie designed a new operating system that would be open source, easy to use and powerful. Named Urbi, this software allows functions such as face and language recognition, location sensing. It was an instant hit among robotic developers.
To grow this emerging ecosystem Baillie made the first step from lab to market in 2006 with the incorporation of Gostai, a contraction of Ghost Artificial Intelligence – a robot in the cult manga series Ghost in the Shell – that Japan’s large community of robot fans would find easy to pronounce.
Still, the rules of business are not the same that the one of research. With no entrepreneurial education Baillie has to behave like one of his robots: learn by doing. Fortunately, beyond his passion for robotics – his newsletter, available on the Gostai website, is a must-read for robots fans – he found that business could be fun and economic analysis intellectually challenging.
He found backing as well. The research lab where he began his academic career, ENSTA (Ecole Nationale Supérieure des Techniques Avancées), provided space and computer capacity. Then his company moved to Agoranova, an incubator financed by the city of Paris, and he won his first external funding from Oseo, a state-sponsored institution that specialises in financing start-ups. “For someone willing to look, there are now tremendous support mechanisms,” says Baillie.
But what is true for support mechanisms is different when it comes to venture capital. There is no shortage off funds among French venture capitalists – they have recently benefited from tax breaks – but robotics is not a priority for them. It is still considered too risky. That’s a mistake, says the enthusiastic Baillie. He believes that personal robotics could become one of the largest industries in the world.
Happily for Baillie, business angels networks and in particular XMP, an angel organisation set up by alumni of the Ecole Polytechnique, the Ecole des Mines and the Ecole des Ponts, bought into his vision. All in all, Gostai has been able to secure close to a million euros to date. That money has been the key to the launch of Gostai’s main initiative to date: GostaiNet.
With Gostai a robot can learn to speak, to recognise objects or to read emails or documents on the Internet. But those features require typically large processing capabilities, something you don’t find in a low-cost robot. So Baillie modified its business plan to benefit from the emerging possibilities of cloud computing and software as a service. Like with the App Store of Apple’s iPhone, developers will be able to post new apps for robots on GostaiNet and customers will be able to buy them.
This new business model is having a deep impact on Gostai’s development, and GostaiNet has now gained now the ears of large telecommunications operators, which like its subscription model. With the pieces of personal robotics now falling into place, Gostai reckons it has secured a central role in this emerging market.