Near Real-Time From Above
Written by Sheldon Piepenburg
Over-simplified, UAV-based image products fall into one of two categories—full motion imagery (FMV) systems or more traditional image “frame”-based. FMV products are typically available in real time or NRT, but usually do not contain high accuracy metadata. Conversely, high accuracy image frame-based products usually require rather extensive production times, sometimes requiring months of preparation before the data becomes available to the warfighter.
In response to that dilemma, two UAV programs, one for the Army and one for the Navy, are currently under way with the assistance of ERDAS. Both systems share common requirements—the initial capture for and the production of NRT geospatial products.
A critical problem in theater is delivering the imagery from the receiving command post to the actual warfighter—a problem commonly referred to as “the last mile.” With planned improvements to the onboard processing hardware/software systems, the warfighter is closer than ever to solving this problem by having an Internet address in the sky.
One persistent mapping application undergoing evaluation and testing is evolving the UAV into a continuous corridor mapping system. Products developed on board the UAV offer the warfighter a tactical tool to accurately monitor strategic corridors. Imagery is continuously collected as the UAV flies back and forth over a target corridor, providing a virtual mosaic over time. Resultant orthophoto imagery provides change detection to automatically monitor the corridor for potential hostile activity, without the need of a frame-by-frame manual review.
The low-cost systems for the Army and Navy can support applications like those described above, as well as others. Each system consists of three basic components: the flight system or platform, the data acquisition system or payload, and image exploitation capability.
Flight Systems. Due to the hostile environment, UAVs experience a fairly high in-theater mortality rate. A key design consideration for both systems was the development of a tactical airborne mapping capability that is relatively low cost, easy to replace and maintain in theater, and simple to operate. Each service is evaluating the tactical airborne mapping system with two relatively low-cost platforms: the Army is using a Cobra, while the Navy employs a Tiger Shark.
Key UAV platform characteristics include:
- Large payload volume and weight capacity
- Modular design for easy payload integration
- Extremely stable and predictable flight characteristics
- Onboard generator and power distribution unit
- Shielding to prevent electromagnetic interference
- Autonomous Flight Management System
- Autonomous GPS guidance
- Affordable system acquisition and operation
- Tactical Common Data Link
Payload or Data Acquisition System. In developing the payload, attention was given to assembling a low-cost payload that could quickly and easily be fitted into a variety of UAV platforms, as well as transition to other fixed wing and rotary platforms. This requirement—to move the sensor payload between platforms—is significant.
Many UAVs have highly specialized payloads that are engineered to fit a single platform and are typically engineered for a specific CONOPS. The availability of a specific system at any given time is not guaranteed. With a low-cost sensor payload designed to be platform agnostic, the warfighter can take advantage of a variety of available UAVs regardless of their current payload configuration. This greatly improves mission success and the ability to respond to the changing threat.
The interchangeable payload was achieved by developing sleds that provide specific locations for the payload components for each UAV. These sleds make it easy and straightforward to mount and balance the payload, and when combined with the airborne processor, make for a highly autonomous system, requiring minimal connectivity to the UAV’s flight components.
Basic imaging components of the system are the sensor (camera), an Inertial Navigation System (INS) comprising a GPS for position data and an Inertial Motion Unit (IMU) for capturing aircraft attitude, and a controller system to coordinate the various instruments and airborne processes.
The imaging system incorporates a ruggedized metric sensor. Metric sensors have known properties that support rigorous mathematic models of the sensor, including lens characteristics such as optical distortions. This results in the ability to produce highly accurate NRT image products.
Depending on the lens focal length, UAV’s flying altitude and INS quality, accuracies within a few feet of actual ground position have been achieved with equally small ground sample distances (GSDs). At 300 meters, for example, both systems have achieved a GSD of better than 0.05 meters.
Positional accuracy is directly proportional to the quality of the INS/IMU, and the overall system cost. The use of quality INS/IMU GPS systems matched to the required accuracy of the mission helps manage overall payload cost.
Onboard Processing and Data Reduction. A key system requirement is to produce a usable image product and deliver the product to the warfighter in the field as quickly and efficiently as possible. The airborne processor is preloaded with mission data and sensor calibration. The airborne processor synchronizes the GPS and IMU data acquisition with the camera, providing highly acurate positional data for the image, rather than the platform.
Current products produced include single image georeferenced NITF 2.1 with RPCs and NITF 2.1 orthophotos. Having the imagery in NITF 2.1 format supports exploitation in both NRT and through DCGS. Using the Tiger Shark and a TCDL during Empire Challenge 2008, the ERDAS team delivered these image products over an OGC-compliant network within one minute of the initial image capture.
In 2008, the Navy Tiger Shark was tested over a four-week period at China Lake Naval Weapons Center, Calif. The Army Cobra was tested in the desert near Fort Huachuca, Ariz. In both cases, the geo-referenced image and ortho products were generated on board within a matter of seconds of acquisition, with positional accuracies of less than 2 meters and a GSD of 0.1 meter. Fast is never fast enough, and accuracy is never good enough.
The team’s ongoing efforts include staff at ERDAS, Geospatial Systems Inc. and the Penn State Electro-Optics Center, all working to improve performance. Images have already been captured and rectified in less than 15 seconds. Migrating the process to an airborne chip set will provide additional speed improvements. Additionally, higher accuracy GPS data (WAAS and differential) will be employed in future tests.
During planned operations over the next months, the system will undergo rigid accuracy testing and speed testing. In addition, a server-side URL application will be deployed in the UAV, making it possible through appropriate communication units to query and download any onboard image from the ERDAS Apollo Server.
Sheldon Piepenburg is director of the Americas and National Programs Services group at ERDAS. Debra S. Schwartz, a senior photogrammetric engineer at ERDAS, also contributed to this article. ♦