Autonomous security robots could work as a team
Cornell University engineers are developing a system which would allow teams of autonomous robots to act as a single entity with many roving eyes. These teams could be used in surveillance and information gathering, and in dangerous jobs such as cleaning up after a nuclear meltdown.
Previously, researchers have created security robots which use a single camera to collect footage; much like a single security guard. However, this new system in development at Cornell University aims to combine information collected from a team of up to ten mobile robots with fixed camera footage and information from outside sources.
These robots could be drones, autonomous ground vehicles or even humanoid robots walking through a crowd. As they roam, they share information. Their central control unit matches images – which could be from different angles, scales, focuses and illuminations – and combines them to create a more detailed picture of what is happening in the area.
The system may also have access to fixed camera footage and information about the context of the scene to identify what behaviours would be considered unusual. A person running in a park, for instance, would not be a cause for concern, while the same behaviour in a secure facility would be suspicious.
The engineers plan for the system to have cloud computing and internet access, which may be used to pull in extra information about what it observes, such as to recognise streets or models of car.
The system would use these sources of information to identify people, objects and behaviours, and track suspicious activity from place to place.
“Once you have robots that cooperate, you can do all sorts of things,” said Professor Kilian Weinberger, a computer scientist based at Cornell University who is working on the project.
The system is based on convolutional neural networks – artificial neural networks inspired by the brain’s visual cortex. Convolution operations mimic neurons responding to visual stimuli.
Early tests will be carried out on the Cornell University campus, using research robots along with webcams to survey crowded areas. If these tests are successful, these teams of security robots could be incorporated into campus security in the future.
The research, which is supported by the US Office of Naval Research, is primarily for military applications. These rapid, autonomous systems could be used to aid US Navy and Marine Corps operations, such as in intelligence gathering, surveillance, reconnaissance, security monitoring and situation awareness.
The robots could also be used to relieve people of dangerous or unpleasant jobs, such as in surveying wreckage after natural disasters or cleaning up radioactive waste.
Sign up to the E&T News e-mail to get great stories like this delivered direct to your inbox every day.