Robots are becoming much smarter than ever before.
Researchers at the University of Surrey are developing robots that can do something people cannot—they can “see” what objects are made of, build detailed 3D maps of unfamiliar places, and safely explore dangerous environments.
The new technology could help protect workers in industries such as nuclear power, rail transport, construction, defense, and emergency rescue.
The project is being developed with two British companies, Industrial 3D Robotics and IS-Instruments.
Together, they are combining advanced cameras, artificial intelligence (AI), and mapping technology to give robots a much better understanding of the world around them.
Most cameras, including those in smartphones, work much like human eyes. They capture images using red, green, and blue light.
But the new robots use special cameras called hyperspectral cameras.
These cameras collect information from many more types of light, including infrared and ultraviolet light, which people cannot see.
Different materials reflect light in different ways. This creates a unique pattern, often called a “spectral fingerprint.” By reading these fingerprints, the robot can identify what different objects are made of, even if they look almost identical to people.
For example, the robot can tell the difference between rust and dirt, detect disturbed soil that may indicate hidden objects, identify suspicious materials, or even distinguish between pills that look exactly the same.
This ability could be extremely useful in places where making the wrong decision could have serious consequences.
The researchers also combined the cameras with a technology called Simultaneous Localization and Mapping, or SLAM. This allows a robot to move through an unknown area while creating a live 3D map at the same time. An AI-powered system called FeatureSLAM makes these maps even more detailed by helping the robot better understand the objects and materials it encounters.
To test the system, the team mounted the equipment on a Spot robot dog made by Boston Dynamics.
As the robot walked through an area, it created a detailed 3D map while also recording information about the materials around it. Instead of simply recognizing walls, doors, or equipment, the robot could also understand what those objects were made from.
The researchers found that using material information improved the robot’s ability to determine its location by about 16% compared with existing methods. That increase in accuracy could make robots much more reliable when working in complex or hazardous environments.
The technology could eventually help inspect nuclear facilities, search damaged buildings after earthquakes, examine railway infrastructure, or support rescue teams in dangerous disaster zones and combat areas. Because the robots can gather information that humans cannot naturally see, they may reduce the need to send people into risky situations.
The research team has already applied for a patent and is now working with industry partners to turn the technology into practical commercial systems. As these robots continue to improve, they could become valuable partners in keeping people safer while performing some of the world’s most dangerous jobs.
