Defence researcher Canicious Abeynayake said a recent trial between the US and Australia gave DST Group the opportunity to test the performance of vehicle-mounted GPR equipment on different soil types.
"Defence’s areas of operation change from time to time, so we need to have algorithms that are robust enough to work well with different soil types," Abeynayake said.
"Countries using the GPRs have recognised the need to optimise and enhance automatic target recognition algorithms to match local conditions and that’s why we wanted to test the equipment in different geographical locations."
The Australian Defence Force has used GPR equipment for years for detecting improvised explosive devices (IED).
The quality of the signal processing algorithms plays a very important role in the ultimate effectiveness of a GPR system by transforming signals into meaningful information for the operators.
But Abeynayake said some current approaches to target identification in GPR lack a solid theoretical basis in the underlying physics, which is fundamental to understanding the responses from GPR.
"Only when target and clutter characteristics are both well understood can signal processing be applied effectively," Abeynayake explained.
"As targets of interest and operational environments keep changing we need ongoing research into advanced sensor configurations and algorithms."
DST Group said its hope is to have a fused approach to IED detection, with the combined strengths of cable detection, metal detection and radar.
"If we can combine all three we can improve detection probability and reduce false alarms. That’s our aim," said Abeynayake.
Another challenge with IED detection using GPR technology is caused by the localised variation in terrain and climate, and the variety of IEDs encountered.
Predicting GPR performance in diverse operational scenarios requires understanding the capabilities of GPR in relation to the physical properties of soils.
Over six weeks during the trial, the teams drove the US Polaris and ADF Husky GPR vehicles up and down lanes of Australian soil at sites across Australia that had been seeded with objects representing IEDs. The accumulated raw data will be used to assess improvements to the processing algorithms.
Abeynayake said the DST Group team will now be investigating the algorithms at the lowest level to see which components are underperforming in different soils and against different targets.