Last week (November 14th), the Nao robot was able to show off its new skill: the little humanoid robot can now play Connect 4 with you!
This is the first truly interactive behavior to really play in the physical world. It was developed by HumaRobotics, the R&D/Consulting branch of Generation Robots, the most complete online shop for programmable robots in France.
Jerome Laplace, the CEO of Generation Robots, answered my questions on how to bring the Nao robot to life, and how to create the (social) robots of tomorrow.
Sebastien: What was the hardest part in making your Connect4 program?
Jerome: How, we start with a hard question. In fact, many things were hard, but I would say the most tricky part was to estimate the depth using just one single camera (the current version of NAO can use only one camera at a time and anyway, when trying to insert the token, the robot is too close to use the other camera). We also had to try many vision algorithms in order to have a quick and efficient algorithm.
Sebastien: Do you use all the sensors of Nao? What are the most useful sensors of the Nao robot in your opinion?
Jerome: We use the cameras, the bumpers, the inertial system and all the feedback from the motors to compute inverse kinematics to estimate the pose of the robot.
For me, at the moment, the most useful sensor is the camera.
Sebastien: I am in the Nao Developer Program too, and I was wondering if there were any companies that would invest in it. What I mean is that Nao is not ready to be shipped soon, but you still invest a lot of resources to create applications.
Sebastien: I don't know other companies involved in creating applications for NAO. The point is indeed that we don't have clear information on when NAO will be on the market and how big does AR think this market will be. So clearly, this is an investment for us and I would say an adventure.
Sebastien: What convinced you to create a team to develop applications for Nao?
Jerome: More and more companies are creating their robot but no real application has been seen. I mean, applications that are useful at home or at work and not application in the labo that are only seen on videos. I'm a dreamer who thinks that the social robot will be a reality sooner or later. But, to make it happen, there is a need for companies specialized in software. That's why I wanted to try Generation Robots to become one of these companies.
Sebastien: Also, I would like to know what you think of the Nao robot as a product. Do you think Nao will ever be launch on the market? What changes would you like to see on Nao to make it perfect? In particular, do you think the legs are necessary for a good user experience, or could it be change to wheels or something else?
Jerome: NAO is a really, really great product. I would like him to have a better hand and fingers because catching some object opens a very large set of opportunities in terms of behaviors and it justify the use of a robot instead of a computer or a tablet.
Before going to public, I think that AR needs to improve the motors.
Regarding legs against wheels, this is a great debate. Indeed, having wheels will make the robot easier to use but with wheels, the robot will lose a great deal of interest. As I told you, I do believe in the social robot. A robot, to become accepted and to generate more interest, needs to have a humanoid form. So I prefer NAO to have legs and I think this is very important in fact.
Sebastien: You say we need humanoid robots to make social robots. But what would make a social robot more than a very expensive toy?
Jerome: I'm sure that we are not able to imagine what the robots will be able to do in the future like people in the 50's were not able to imagine social network When they started building computers. This point is very important, many people think they know exactly what robots will become but I think this is not the case.
In a near future, I would say that there is a high demand for robots that can look after elders, children, ill or disabled people. Think of robots that can speak, help and play with a person that is alone, this is a great health factor (you have an example with elder who have pets and talks to them all the time...). The robot will be able to look after, be sure that drug has been taken, that the person is not asking for help...
Sebastien: Thank you very much for your answers.
It's true that no one can really predict what will be the first groundbreaking applications of robots in the future. Creating the robots of tomorrow is an exciting adventure.
I guess the best way to show people that social robots can be made, is to make it yourself. Let's hope Generation Robots can create a new generation of robots.