Use of electromagnetic (EM) methods to discriminate frag and geology from unexploded ordnance (UXO) has shown steady improvement over time as demonstrated by test site results. Performance of off-the-shelf and customized EM sensors has increased both in terms of UXO probability of detection and false alarm rate. Parallel with the increase in capabilities of UXO detection and discrimination is the increasing use of digital geophysical techniques on live sites. However, a significant gap exists in capabilities for UXO detection and discrimination between prove-out/test sites and actual UXO-contaminated sites. In addition, the future deployment of more sophisticated sensors with discrimination capabilities is limited by the requirement for greater data fidelity for wide application on most sites. One issue of importance for improving data quality is the need for improved spatial representation of the geophysical signature of subsurface UXO. Presently, survey data are inadequately spatially quantified with simple XY sensor locations and an assumed constant Z elevation. To accurately represent the geophysical signature of subsurface UXO, more complete and accurate information is required, including sensor elevation information, orientation data (yaw, pitch, and roll), and sensor velocity and acceleration data.

To achieve the goal of providing UXO discrimination technologies beyond the prove-out test plot for routine use on actual survey grids, the overall objectives of this project were to (1) quantify the effects of sensor orientation that diminish UXO detection and discrimination capabilities in real-world field conditions and (2) exploit existing and emerging hardware and data modeling technologies to mitigate orientation effects for increased UXO detection rates and reduced false alarms. 

Technical Approach

The technical approach to achieve these objectives included the following: quantification of sensor orientation effects via laboratory test data, platform data, pseudo survey data, and test plot data; development of a gimbaled sensor configuration incorporating advanced positioning robotic total station (RTS) and data acquisition system (DAS) technologies; investigation of the use of inertial measurement units (IMU); tests of this system in a variety of controlled and live site scenarios; and modeling of the sensor orientation effects to determine the impact of improved positioning on the ability to make target parameter estimates.

The design and fabrication of a gimbaled sensor configuration provided a mechanical solution to mitigating sensor orientation by allowing the sensor to rotate in the pitch and roll axis. The platform is a lightweight, non-metallic pushcart consisting of a wheel/axle, frame, a yoke, and EM61 sensor assembly. All of the components were constructed from non-conductive materials to eliminate EM signature effects. The EM61 remains aligned with the gravity vector regardless of the orientation of the mounting frame and deployment platform. The frame allows for the mounting of GPS electronics and EM electronics on the handlebars.

In addition to the use of a gimbaled sensor configuration, this project demonstrated that yaw, pitch, and roll data can be effectively collected using IMUs to accurately measure sensor orientation during electromagnetic induction (EMI) data collection. However, one issue with commercial IMU instruments is that they can be corrupted by the primary magnetic field of the EM sensor and can also influence sensor measurements. Therefore, this project developed a prototype low metal IMU using commercially available integrated accelerometer circuits comprised of an integrated circuit board containing the Analog Device ADXL311 accelerometer and housed in a plastic case with no substantial metallic components. Testing of the prototype IMU demonstrated that while the IMU worked well under most benchtop conditions, it failed to provide reliable measurements during survey tests. Subsequent additional improvements of the prototype IMU device failed to reduce noise levels and increase reliability. Therefore, further development of a low metal IMU was not pursued by this project and subsequent data collection was conducted using a commercial Crossbow IMU.

Sensor orientation effects can be mitigated through modeling techniques advanced through the efforts associated with this project. Sensor orientation information was utilized during the forward and inverse modeling process to account for signature variations caused by non-horizontal sensor orientation during data collection. As part of this project, the EM modeling process for sensor orientation information was developed that included full simulation of the three-dimensional orientation of EMI sensors, multi-time gate EMI response, and a beta classification technique. An EMI dipole response model was tested to facilitate modeling of EM data in conjunction with concurrently collected orientation information and incorporated into the beta inversion model used for target parameter estimation.


Augmentation of EM data with sensor orientation measurements is required to compensate for both orientation effects and varying levels of magnetic flux that pass through EM sensor coils and alter target signatures. The increased understanding of the effects of sensor orientation in addition to ways to measure and mitigate sensor orientation and to model these effects advances the efforts of the Department of Defense to further develop UXO discrimination capabilities. (Project Completed - 2006)