INTERNATIONAL CONFERENCE ON MINE CLEARANCE TECHNOLOGY 2-4 July 1996 Copenhagen, DENMARK TECHNOLOGY FOR MINE CLEARANCE OPERATIONS APPLICATION OF SENSOR SYSTEMS TO MINE SURVEY Lawrence J. Nee Introduction 1. The first step in any successful demining operation is planning the effort, one of the key elements of which is to determine the magnitude of the mine problem and the locations of the minefields. Information gathering techniques, such as interviewing former combatants and mine victims can provide significant insights but do not necessarily reveal the entire extent of the problem to be tackled. What is needed is a rapid, low risk and cost effective means to survey a region for mines and produce maps for use as planning tools. These maps will also facilitate the process of marking minefields to minimize future civilian casualties. Continuing advances in sensor technology and digital signal processing promise to revolutionize the process of minefield mapping and survey. 2. The current state of the art sensor for mine detection is the metal detector. There exist a number of different technical approaches to the problem of metal detection, but the results are similar. A number of countries are making the investment in alternative technologies to detect mines from airborne and ground vehicle platforms, as well as to improve hand-held detectors. If we are to advance the state of the art of minefield mapping and survey, it is the airborne systems that will yield the most rapid advances due to their ability to survey large areas without risk to the operator. CURRENT STATE OF SENSOR TECHNOLOGY 3. The most mature development of an airborne detection system today is the system in development by the United States Army. The system is known as the Airborne Standoff Minefield Detection System or ASTAMIDS. While it might not be the most technically sophisticated system, it is the closest system to being fielded for service use. ASTAMIDS is being developed as a mission payload to be flown on an Unmanned Aerial Vehicle (UAV) platform for use in day/night combat operations. Consideration is currently being given to fielding a modified version of the system on a UH-60 Blackhawk Helicopter for use by United States Forces deployed to Bosnia. The UAV deployment requirement for the military mission imposes several technical challenges. First, due to weight, volume and power constraints for the UAV, the sensor system is highly engineered to fit in the platform. Second, since it is to be flown in an unmanned platform, a great deal of emphasis is placed on data compression and the use of a digital downlink to transmit the data to a ground station, where the actual processing of the imagery for minefield target features is accomplished. Currently, two competing sensor approaches are being pursued for the ASTAMIDS system. One approach is the use of Second Generation Thermal Infra-Red sensor that permits totally passive operation based on the thermal or heat signature of a mine. The other approach involves the use of a laser along with an IR sensor to add an additional channel of polarization data. 4. As with most high resolution imaging systems, a digital signal processing system is used to handle the large amounts of digital image data from the sensor. State of the art parallel processors are used to process and enhance each individual pixel before processing of the imagery is performed. The signal processing algorithms examine the imagery for the presence of clusters of anomalies that resemble minefields in their characteristics. First the individual anomalies are examined for the proper size and shape that might be a mine-like target. Then clusters of likely anomalies are examined to see if they fit the pattern of deployment for a buried, surface or a scatterable minefield. Through multiple passes over a likely mined area, the boundaries of the mined area are determined and recorded using input from an on-board Global Positioning System (GPS). 5. Prior to the current phase of development, the United States Army conducted technology demonstrations which showed the viability of Electro-Optic/Infra-Red technology for the detection of both surface and buried landmines. Both active and passive approaches were demonstrated to have potential to be successful in a mine detection role. The detection of buried mines presents the most difficult challenge. Some lessons learned from that demonstration are that we need a sensor swath width of sufficient size to allow automated signal processing algorithms to function effectively in identifying a sufficient quantity of mine-like targets. To ensure the detection of scatterable minefields, a sensor resolution of 25 millimetres was found to be necessary to see these mines, which are typically the smallest of antitank mines with diameters down to 100 millimetres. Finally, given the data acquisition rates and the potential for high amounts of clutter, computer-aided target feature extraction is a necessity to assist the operator in his detection task. DEVELOPMENTS IN THE FIELD OF SENSOR TECHNOLOGY 6. While EO/IR systems are most likely to be the first airborne detection systems to be fielded, they are not the only viable sensor technologies for mine detection. When the military requirements of day/night operations are removed, visual systems are viable for mine detection from an airborne platform. Separated aperture radar, millimetre wave radar and ground penetrating radar all have been examined for this role. Each has its strengths and limitations. Resolution of mine targets from any considerable height is likely to be a problem for ground- penetrating radar given the size and depth of burial of land mines. Sensor schemes which rely on change detection may prove to be viable in a military role but are not likely to be useful for demining operations where mines have been emplanted for some period of time. 7. Hyperspectral systems using both visual and IR spectroscopy have been studied and show some viability to detect signs of disturbed soil where a mine has been emplanted. These signs of disturbed soil will fade with time and the utility of these sensors is limited by the length of time the target has been buried. APPLICABILITY TO DEMINING 8. In many respects, the application of airborne detection systems to the demining problem is simplified by removal of many of the constraints faced in the development of systems to meet the tactical needs of the military. Certainly it is still necessary to use sensors that are stabilized to compensate for the pitch, roll and yaw of the airborne platform. The imagery or sensor information must still be co-registered with position information from a GPS or INS system to facilitate generation of maps or map overlays of the mined areas. Digital signal processing systems must still be used to handle the high data rates from airborne sensors. What is simplified is the platform constraints, that is, we can use a manned aircraft instead of a UAV, the real time processing constraints that drive the military application to a digital data downlink are eliminated and we have more flexibility to deal with the time and rate at which the data is collected. 9. Taking the last point first, in a demining environment, day/night operations are not a must. Consideration of diurnal cycle limitations can lead to optimization of data collection flights to times when the target signature is at a maximum. Multiple passes can be used to verify data. Since real time processing is not a hard requirement, digital data can be recorded and processed on the ground without any need for data compression. Another approach is to carry the processors in the airframe and do the processing on board. This would facilitate the operator directing the pilot to pass over likely mined areas for a closer look. Finally, the cost of sensors can be reduced to looser size, weight and power constraints. 10. Most, if not all, of the software written for the military application of airborne detection is directly applicable to the demining problem. With some modification, it is also conceivable to use many of the same processing schemes with the non-imaging systems since they are still anomaly detectors which will still rely on patterns to indicate the presence of minefields. The parallel processors described above are currently available in the commercial marketplace and are not limited to military applications. Much of the computer processing capability being used in the current phase of development is based on commercial off-the-shelf systems that are readily available. LIMITATIONS 11. As with any sensor system, there are limitations driven by the choice of the sensor. For the EO/IR systems, visual access to the mined areas from the airborne platforms is required. Solar loading is necessary for optimum performance to get temperature differentials that are detectable by the sensors. There will also be two distinct periods during the day when there is no temperature differential for the sensors to detect. These diurnal crossover points vary in time of day from one day to the next based on variations in weather conditions. Some sensors have a temporal limitation in that the longer a mine is buried, the less likely it is that you will be able to detect it. COST IMPLICATIONS 12. In spite of the differences in requirements between a system for military applications and for demining, an airborne detection system is likely to be extremely costly until the cost of the sensors come down. Current sensors are expected to cost several million dollars. Skilled technicians would be required to maintain the system. These considerations would likely make the investment in such a system cost prohibitive for an individual country striving to set up a demining programme. A more practical approach would be acquisition of the system by some organization which would provide for use of the service on a pay-as-you-go basis. CONCLUSIONS 13. The use of airborne sensors to survey a region for mines in support of the initiation of a mine clearance programme is certainly achievable within the next five years by leveraging the investment by the military in an airborne minefield detection system. Little in the way of new development is required. The major hurdles would be selection and integration of the sensors to be used, integration of the sensors with airframe and selection of the data-processing equipment to suit the sensors selected. The largest hurdle of all would most likely be the acquisition cost of the system.