Smart Robots

Robotics has always had great fascination for artificial intelligence researchers. After all, the ability to function convincingly in a real-world environment would go a long way toward demonstrating the viability of true artificial intelligence. Building a smart, more humanlike robot involves several interrelated challenges, all quite difficult. These include developing a system for seeing and interpreting the environment (computer vision) as well a way to represent the environment internally so as to be able to navigate around obstacles and perform tasks.

One of the earliest AI robots was “Shakey, ” built at the Stanford Research Institute (SRI) in 1969. Shakey could navigate only in a rather simplified environment. However, the “Stanford Cart, ” built by Hans Moravec in the late 1970s could navigate around the nearby campus without getting into too much trouble.

An innovative line of research began in the 1990s at MIT. Instead of a “top down” approach of programming robots with explicit logical rules, so-called behavior-based robotics works from the bottom up, coupling systems of sensors and actuators that each have their own simple rules, from which can emerge surprisingly complex behavior. The MIT “sociable robots” Cog and Kismet were able to explore the world and learn to interact with people in somewhat the way a human toddler might.

Swarm robotics is an approach to robotics that emphasizes many simple robots instead of a single complex robot. A robot swarm has much in common with an ant colony or swarm of bees. No individual in the group is very intelligent or complex, but combined, they can perform difficult tasks. Swarm robotics has been an experimental field, but many practical applications have been proposed.

A traditional robot often needs complex components and significant computer processing power to accomplish its assigned tasks. In swarm robotics, each robot is relatively simple and inexpensive. As a group, these simple machines cooperate to perform advanced tasks that otherwise would require a more powerful, more expensive robot.

Using many simple robots has other advantages as well. Robot swarms have high fault tolerance, meaning that they still will perform well if some of the individual units malfunction or are destroyed. Swarms also are scalable, so the size of the swarm can be increased or decreased as needed.

One use that researchers have demonstrated for swarm robotics is mapping. A single robot would constantly need to keep track of its location, remember where it had been and figure out how to avoid obstacles while still exploring the entire area. A swarm of robots could be programmed simply to avoid obstacles while keeping in contact with other members of the swarm. The data from all of the robots in the swarm is then combined into a single map.

Swarm robotics has been an emerging field, and it has presented unique challenges to researchers. Programming a swarm of robots is unlike other types of programming. The model of distributed computing — using many computers to work on a single large task — is somewhat similar. Unlike distributed computing, however, each individual in swarm-style robotics deals with unique stimuli. Each robot, for example, is in a different location at any given time.

Some approaches to swarm robotics use a control unit that coordinates other robots. Other approaches use techniques borrowed from nature to give the swarm itself a type of collective intelligence. Much of the current research in the field focuses on finding the most efficient way to use a swarm.

Swarm robotics is a concept that's buzzed around since the 1980s, but now the technology is starting to fly. The environmental applications being explored range from coral restoration and oil spill clean-ups to precision farming – even the creation of artificial bees to pollinate crops.

Dr Roderich Gross, senior lecturer in robotics and computational intelligence, explains the concept: " In a swarm system there is no single point of failure – if a unit fails, the whole system keeps on going. Wherever you have a very heavy load that a human cannot manipulate, using a swarm of robots to do the job would be very sensible. That could be in a factory, transporting boxes. Or it could be a search-and-rescue scenario – maybe a collapsed building and you need to remove a very heavy part, or working in contaminated environments."

Scientists and designers at Heriot-Watt University have been looking at using a swarm of " coral bots" to restore ocean habitats. Dr Lea-Anne Henry of the university's school of life sciences believes that swarm robotics can " revolutionise conservation". Agriculture is looking into the potential for using swarms too. Professor Simon Blackmore, head of engineering at Harper Adams University works on larger robots that can work in fleets, able to identify weeds and administer microdots of chemicals with the result of using 99.9% less herbicide than traditional methods. He believes that, though the technology may appear an expensive luxury, it may have a wider appeal than the latest generation of conventional farm machinery such as expensive tractors and harvesters.

Perhaps the most famous – and controversial – swarm project to date is Harvard University's " Robobees", aiming to find an artificial solution to pollination to address the current decline in the global bee population. Here the robotic swarm is attempting to replicate one of nature's greatest swarms. But even setting aside the ethics of attempting to replace nature's pollinators, the idea may remain impossible.

The problems of organizing a swarm haven’t kept people from imagining what swarm robotics could offer some day. Some scientists envision a swarm of very small microbots being used to explore other planets. Other proposed uses include search-and-rescue missions, mining and even firefighting. When used with nanobots — microscopic-size robots — swarm robotics could even be used in human medicine.