USU researcher looks to the way groups in nature interact to improve autonomous robotic technology
by Kinsey Love
Have you ever watched a flock of birds swoop and dive in the sky in perfect union? All the members of the flock seem to move as one body, quickly darting back and forth, each member picking up on the subtle cues of its neighbor to determine its next move. Without any practice, each bird performs difficult maneuvers in perfect form because of the communication that is constantly networked through the flock.
The beauty of a flock and networked communication is that if one or two birds are removed, the whole body will stay intact and continue to function as it did before.
Although control systems and robotics research isn’t as simple as bird watching, it has taught engineers a thing or two. Wei Ren, USU electrical and computer engineering researcher, is studying the theory and applications behind distributed coordination and control of multiple autonomous vehicles. Just as a flock of birds uses cues and signals to communicate within the flock, Ren designs and analyzes mathematical algorithms that are used by robotic vehicles to network and communicate so they too can work together, moving as a single body, to accomplish certain objectives.
Ren and his team of graduate students who work in the COVEN Lab, or COoperative VEhicle Networks Laboratory, use distributed consensus algorithms to network and control teams of robotic vehicles. The basic idea behind information consensus is that each vehicle updates its information based on the information of its nearest neighbors so that the final information of each vehicle converges to a common value. These algorithms allow the robots to rendezvous, control formation, flock, and perform other maneuvers.
“In the past, there was a central station that sent commands to every robot in the team,” said Ren. “This was not efficient because if the central station failed, then the entire system or the entire operation would fail.” He further explained that, single vehicles performing solo missions will yield some benefits, but cooperation between teams of vehicles has the potential to yield even greater benefits. The theory is that many small, inexpensive vehicles acting together can achieve more than one larger vehicle.
Researching coordination and control of multiple autonomous vehicles is fairly new, but Ren hopes that over the next few years, his research with multi-vehicle systems will make exponential progress. It is his goal to be able to control teams of robots that communicate and network with their local neighbors quickly and efficiently in a distributed manner.
“Research such as this is newly emerging and has many opportunities for exploration,” said Ren. “It is different from traditional control systems and robotics research. It is so much more than just controlling a single robot—it is creating real-time interaction between intelligent machines, and in turn, generating global behaviors from the local interaction, and that is what really gets me excited.” There are numerous real-world applications for cooperatively controlled groups of robots. They have the potential to be used in the military for reconnaissance, battle damage assessment, or surveillance. They also could be used in civilian sectors such as mining, environmental monitoring, or agriculture.
Currently, the COVEN research laboratory has a few groups of ground robots: AmigoBots and Pioneer 3-DXs, iRobots, and unmanned aerial vehicles (UAVs), which are aerial robots. AmigoBots and Pioneer 3-DXs have onboard computers, high precision wheel encoders, and six forward and two rear sonars that allow them to communicate and sense the position of other robots. These components enable them to create shapes and other formations. The iRobots are similar to the Roomba vacuum cleaning robots. They are round and flat and are loaded with internal motion, wheel, infrared ground, and front bump sensors. The UAVs are fixed wing platforms that can fly on autopilot and are equipped with GPS units, temperature sensors, and even cameras.
So far, Ren and his team of graduate students have successfully designed algorithms to control these robots in simple tasks. They are able to create moving geometric shapes with the ground robots and perform a type of “follow the leader” task with the UAVs. They have even written an intercept algorithm, through which a team of robots will find and intercept a previously identified enemy target.
One of Ren’s latest projects is called a quadrotor, which is a helicopter-like robot with four separate horizontally spinning blades. “This robot shows a lot of promise because of its versatile movement abilities,” said Ren. “It can hover, fly at low altitudes, and it is very small and agile.” He hopes to fully automate this helicopter and eventually design control algorithms that will enable cooperation between the quadrotor and other ground robots.
Cooperative control systems may sound like fantastic sci-fi inventions from the latest video game, but there are still numerous challenges to the research. These robots are still limited by current technology: the availability of bandwidth and connectivity is still limited, as are computational resources. These limitations, though, are not hindering the COVEN lab too much, and they have high hopes for the coming years.
Ren’s research has been funded by USU’s Utah Water Research Laboratory and by a USU Community/University Research Initiative grant. His new research is supported by a National Science Foundation CAREER grant. In the future, Ren expects his research to progress, and for cooperatively controlled vehicles like COVEN’s to be a common part of life and society. “There are endless military possibilities where a team of robots can perform tasks and maneuvers that are too dangerous for humans, such as battle damage assessment or reconnaissance,” said Ren. “The possible civilian and household roles are increasing all the time.”