AI in action: from science fiction to our living rooms

AI is everywhere. Fraud detection, online customer service, Siri, Alexa, Call of Duty, smart cars… the list is endless, and it’s having a growing impact on the way we live our lives. It’s having an even bigger impact on robotic platforms.

At PAL Robotics, we’ve spent the last 13 years designing and building collaborative service robots that we believe will improve our quality of life, and AI is an essential part of ensuring that our robots can work successfully alongside humans. It’s often the ‘simplest’ tasks robots find most difficult, and learning by demonstration and imitation are key methods they use to pick-up these skills.

Since our very first biped, REEM-A, built to beat a human at chess, we’ve incorporated AI. As a result of pioneering technology, the robot learned to map and recognise its environment so that it could navigate safely around humans and obstacles. Over the last decade, we’ve used AI for a wide range of tasks – even teaching robots to recognise door handles, so they can open doors by themselves!

The quickest and most efficient way to generate change is to collaborate, which is why we design robots that allow universities and research labs to adopt, amend and build on our platforms. All our software is 100% ROS compatible, so any programmer or engineer can incorporate our code if they’re using ROS – the Robot Operating System. Simulations and tutorials for our robots are available for free, online, so that anyone – anywhere in the world – can test new software applications on a real robot.

Have a play on our mobile manipulator TIAGo.

Furthermore, building modular, flexible robots makes it easier for our engineers to incorporate new software into existing frameworks, which is crucial if we want to make use of the latest developments in AI.

You’ve probably heard of the REEM robot being used by Dubai police, but did you know it was manufactured in Barcelona? So that REEM can recognise objects, places and faces, we decided to incorporate IBM Watson’s chatbot tool, Conversation. We made the decision to publish the code we used to make it ROS compatible on GitHub, so that anyone looking to incorporate AI software into the ROS framework can do it for free.

Although probably the most well-known, IBM Watson isn’t the only AI software we’ve used to build platforms that can interact with humans. We recently had an incredibly funny morning teaching TIAGo to recognise employee’s names using Snowboy’s hotword detection engine. The wide range of accents in the office confused the little robot at first, but it got there in the end!

AI’s importance to the future of robotics is now unquestioned: it has moved from the realms of fiction into our cars, offices and living rooms. But the need for continual collaboration and open exchange, both within the field of AI but also robotics more broadly, cannot be underestimated.

 

Author: Elisa Alston, PAL Robotics