Portuguese 
English 
Spanish 
3 min of reading 
0 comments 
AI System Created at the University of Surrey Reduces Location Errors to 22 Meters and Promises Accurate Navigation Even Without GPS.
Driving without GPS seems unthinkable, but researchers at the University of Surrey believe this could change. They have developed an artificial intelligence system that locates devices even in areas where satellite signals fail. The innovation, called Pose-Enhanced Geo-Localisation (PEnG), represents a significant advance in smart navigation.
A New Way to Locate
PEnG impressed in tests. The system reduced location errors from 734 meters to just 22 meters. Unlike GPS, it does not rely solely on satellites.
The technology combines satellite and street-level images to identify the position and direction of a device.
-
Scientists find a methane-free ecosystem in the Peru-Chile Trench, sustained by unique microorganisms that live off sulfur and challenge theories about life in extreme environments.
-
Two wheelchair batteries, Arduino, and Bluetooth transformed garage junk into an iPhone-controlled electric “mini-tractor” capable of pulling a twin-engine aircraft weighing over 2.2 tons without relying on a professional tug.
-
Scientists have just discovered a new family of magnetism, and this fundamental physics discovery could pave the way for faster memories, advanced sensors, and more energy-efficient electronics.
-
Advance against cancer: USP scientists develop revolutionary biosensor that identifies pancreatic cancer in 10 minutes at low cost and can save thousands of lives in Brazil
This process occurs in two stages. First, the system narrows the location at street level. Then, it refines the calculation through relative pose estimation, a technique that assesses the camera’s orientation.
According to the researchers, even a common monocular camera, present in vehicles, is sufficient for operation.
Where GPS Fails
The developers highlight that GPS has known shortcomings. Tunnels, cities with skyscrapers like New York, and areas with low connectivity hinder coverage. In these situations, PEnG can fill the gap. For this reason, the team advocates that the tool is practical and accessible for future use.
“Many navigation systems depend on GPS, but coverage is not always guaranteed,” said Tav Shore, a graduate researcher in AI and computer vision. He explained that the goal was to create a reliable solution based solely on visual information. This strategy enables a level of accuracy considered unachievable without GPS.
Potential for Autonomous Vehicles
The Surrey team believes that the technology’s flexibility can transform the navigation of autonomous vehicles. “One of the most interesting aspects of this system is how it transforms a simple monocular camera into a powerful navigation tool,” said Professor Simon Hadfield, an expert in robotic vision and autonomous systems.
The researcher emphasized that PEnG is designed for unpredictable and fast-moving scenarios. This makes it ideal for the next generation of autonomous vehicles and robots operating in challenging environments. Therefore, the technology emerges as a solid alternative for systems that cannot rely solely on satellites.
Global Dependence on GPS
Currently, the dependence on GPS is enormous. Sectors such as logistics, aviation, defense, and public transport require precise navigation. However, GPS signals can be easily interrupted. Tall buildings, weather conditions, or even jamming equipment cause failures. This vulnerability opens the door for more robust alternatives.
With PEnG, the researchers advocate that navigation can become more resilient. In addition to convenience in daily life, the system can enhance safety in critical areas. For them, this is a change that affects both ordinary users and strategic sectors.
People-Centered Approach
Professor Adrian Hilton, director of the Surrey Institute for People-Centred AI, highlighted the human aspect of the project. “The work of our team demonstrates the people-centered approach we advocate, developing a system that addresses the challenges behind navigation technology, something that we all rely on,” he stated.
He added that locating accurately without relying on GPS is a fundamental step towards smarter and more resilient autonomous systems. This capability paves the way for operations in remote locations and unpredictable situations.
Next Steps
The project received support from the University of Surrey’s Doctoral Research Foundation Award. This funding supports promising early-stage technologies. Now, the researchers are working on creating a functional prototype for testing in real-world conditions.
Additionally, the team opted to release their research as open-source. This decision invites developers and engineers from around the world to collaborate in advancing the technology. The findings have already been published in the scientific journal IEEE Robotics and Automation Letters, further increasing their visibility.
With this, PEnG moves beyond being just a laboratory concept and approaches practical application.
-
-
-
-
-
-
8 people reacted to this.