Case Study: Improvised Explosive Devices
Improvised explosive devices (IED) are a huge problem for British soldiers serving in Afghanistan. These explosive devices have been a major cause of casualties. The hunt is on for new technologies that can tackle this threat.
Oxford Machine Intelligence is currently looking at ways of applying neural network and other complementary pattern recognition techniques to the problem of detecting improvised explosive devices from ground penetrating radar (GPR) mounted on an unmanned airborne vehicle (UAV).
Airborne vehicles will scan for improvised explosive devices using ground penetrating radar. The radar signals will be analysed by sophisticated pattern recognition software. Such unmanned airborne vehicles could provide instant warnings and a precise fix on the location of suspected explosive devices.
The problem of identifying the presence of landmines and improvised explosive devices from ground penetrating radar signals is notoriously difficult. The challenge is to reliably detect improvised explosive devices amidst changing soil types and ground conditions, the presence of rocks and other irrelevant objects, signals reflected from the ground, and various forms of signal noise.
Oxford Machine Intelligence is investigating a range of machine learning techniques for this problem, including supervised neural networks, image analysis, data fusion and statistical decision theory. In particular, we are applying our knowledge of neural networks for transform-invariant visual object recognition in natural cluttered environments.
This work is taking place within our research centre based within the Oxford University Department of Experimental Psychology. The centre is headed by Dr Simon Stringer, who has over 20 years experience in computational mathematics and artificial intelligence.
To contact us: please e-mail Dr Simon Stringer (CEO) at: