At the 2015 IdeaFestival, Google engineer Dr. James McLurkin shared his research on swarm robotics and how he believes they could be the future of space exploration, disaster rescue, and more.
The robot revolution has already begun. Robots are available for purchase on Amazon, they are in our homes, and they are helping soldiers stay safe in warzones.
But, what is the future of practical robotics? Dr. James McLurkin, a Google engineer and roboticist, believes that swarm robotics, or groups of robotics that work together to accomplish goals, is the research that will push the field forward.
The presentation, titled "The future of robotics is swam," opened the 2015 IdeaFestival in Louisville, Kentucky on September 29. It was part of the Thrivals 8.0 program in which the audience is primarily high school math and science students on the first day of the event.
McLurkin opened by explaining how the world's current views on robots typically fall into one of three plots that are perpetuated by the media:
- Frankenstein plot—society's view
- Tin Man plot—the quest to be more human
- Terminator plot—the killer robot world takeover
However, these views misunderstand how robots can help humans accomplish more, McLurkin stressed. But, how do we make sure that robots continue to work well with humans in the future? McLurkin noted the three laws of robotics as quoted from Isaac Asimov in his book I, Robot:
- A robot may not injure a human being or, through inaction, allow a human being to come to harm.
- A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
- A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.
"The problems with these rules is that robots can't read, at least not very well," McLurkin said.
To understand robotics, we must first understand intelligence. McLurkin said that there are philosophical questions of intelligence we must seek to understand as we study robots.
The first philosophical question is simply what is intelligence. The best thing we have in terms of robots, he said, is The Turing test developed by Alan Turing.
The second question to ask is whether or not intelligence emerges from the interactions of unintelligent components. We are all built from atoms, so atoms must either be intelligent in and of themselves, or their interactions must produce intelligence.
The final question to ask is does an intellect need a body, or does the type of intelligence depend on the physical form—his answer is yes.
The tasks that robots are well-suited for are known as three Ds: dangerous, dirty, and dull. However, there is a fourth D that McLurkin and his research focuses on: distribution. By distributing tasks across multiple robots, they can perform tasks ill-suited to a single robot.
By using this model, he said, we could send 20 robots to look for hot spots in a forest fire area, or we could send 200 robots in disaster areas to search for survivors.
To show what he has been working on, McLurkin demoed his swarm robots on the IdeaFestival stage to show how they communicate with each other. He successfully had them cluster together, form a line, and showed that they can move around as a group and rely on each other.
The robots have a limited range of communication, but they use both local communications with closest neighbors and the global network of their communications to keep themselves connected to one another. The robots can triangulate to be used as network points so a few robots can explore with extended range, or they can use a "breadcrumbs" technique to follow a few individuals to get home.
The robots are coded in C, and take about 45 minutes to build. They rely on a custom operating system to perform their tasks and use different colored lights to show what groups they are in.
McLurkin's robots were also able to organize themselves by their assigned, numerical ID. He commanded them to line up from lowest to highest numerical ID and showed that, as he moved them around, they would react and reorganize themselves. The lowest and highest numbers are known and the in-between numbers looked for the numbers directly above and below them.
Swarm robots could change the future for everything from home-building to disaster recovery. But, McLurkin mentioned another specific vision for their use—the exploration of Mars.
So, this begs the question of how, exactly, do you program 20, 200, or even 2000 robots?
"You look to nature," McLurkin said.
Ants, wasps, and honeybees, for instance, have been operating in swarm behavior for as long as they have existed. Engineers look to these bugs as examples of engineering that they can glean information from on how their robots can act.
For example, the honeybee's pattern for nectar collection is an algorithm that can be replicated in robots. The engineers and biologists work together to see what they might be missing in the pattern or the algorithm.
They also look at bugs such as cockroaches. For example, a cockroach can detect movements in air pressure and move out of harm's way extremely fast, so engineers like McLurkin are trying to replicate that in robots, too.
Robots will change the future, but it won't be without the study of millions of years of phenomena from the natural world to learn powerful patterns.
- Why AI could destroy more jobs than it creates, and how to save them
- Google AI beats humans at more classic arcade games than ever before
- Why it's time to prepare for a world where machines can do your job
- The first 10 jobs that will be automated by AI and robots