Artificial Intelligence in Robotics
This week I spoke at the Idea to IPO meetup about AI in Robotics. Artificial Intelligence and Robotics have become hot topics that attract a variety of crowds, from engineering heavy-weights to hobbyists to the young teenagers looking to the future of technology.
I began my presentation by taking the audience back in time to share what robots used to be and look like. I remember growing up with robots in movies. Anyone recognize these characters? Johnny 5, WALL-E, C-3PO, R2-D2. They looked intelligent on screen, but since these robots were in movies, Hollywood took some creative initiative to make robots smarter than they really were on set.
Spoiler Alert: there might have been some trickery done by an actor or two, or a stage prop behind the scenes that moved these robots and their expressions and sounds. And even those amazing sound effects might have been added in post production.
Today’s robots are much more autonomous than than their ancestors were, thanks to the help of artificial intelligence systems like IBM Watson. And if I may say so myself, they also look a little more polished design-wise. Maybe even trying too hard to imitate humans? Robot designers and engineers have been inspired by generations of robots seen in sci-fi movies, and have prompted lots of research in user experience and design, sometimes even pushing the limits to innovate the next generation of robots to help with elder care. Did you know there’s even a robot on the International Space Station?
Another thing that’s interesting to explore is how robots have been trained over the last couple of decades. Many have been procedural robots. For example, a robot in a factory setting might be trained to follow a set of procedures. Move arm over to the conveyor belt, open and close claws, lift up a box, rotate around, and stack onto a palette. This intelligence has been preprogrammed. There are also safety checks for abnormal conditions, but again those are usually procedural checks. If all else fails, the robots are designed to stop when it’s safe and not try to overcome an obstacle it hasn’t been trained to deal with.
It’s true that today’s robots in customer service are also in part powered by procedural rules. But with artificial intelligence capabilities, we’re now able to enable the robots to learn as time goes on. Just like a human learns how to do things with repetition, robots can also learn from an ever-growing set of data points using machine learning and data science. There’s plenty of resources on the internet if you’re fascinated with self-learning robots.
If you’re interested in getting started with programming robotics, you don’t need to spend hundreds or thousands of dollars for a robot. Using hardware from the maker community, like a Raspberry Pi or a Particle Electron, you can control LEDs, servos, cameras, microphones and speakers to enable your robot to interact with the physical world.
Last summer I went on a blogging and coding spree with the TJBot. TJBot is a research project that was created at IBM Research back in November 2016. It makes prototyping a robotic functionality pretty easy. And there’s a number of languages that are supported: Node.js, Swift, and Node-RED. You can find example recipes to get started building your smart robot.
And the most important thing a robot can do for you is obviously to find a place to have lunch. Err, or having tweets read to me. I don’t know, maybe my favorite is teaching me how to spell words. There are so many things a robot can help with. Why don’t we have personal assistants roaming our homes and offices?!?
In an effort to make the TJBot open-source project even more accessible, a couple of months ago I developed and released a browser-based TJBot simulator, You can try out an accompanying lab that shows how you can enable speech (using Watson Speech to Text and Watson Text to Speech) and search for news headlines (using Watson Discovery News).
This is an exciting time to get into robotics and to watch how artificial intelligence, machine learning, and other technology drive robots to new levels of innovation. So, what’s your dream robot and what would it do? Share in the comments.
