Around the globe, aviation organizations, both public and private, are working to find ways to safely immerse Unmanned Aerial Systems (UAS) into our communities and the existing aviation airspace.
To this end, the Queensland University of Technology’s Australian Research Centre for Aerospace Automation (ARCAA) in Australia, is leading the way. During the last year, the group has conducted one of the industry’s first successful trials of an onboard Detect-And-Avoid (DAA) solution for small Unmanned Aerial Vehicles (UAV) under 20 kilograms, and also an unassisted landing capability for a UAV.
Here’s an inside look at the impressive advancements that the research and engineering team at the ARCAA have developed as part of the recently completed two-year, AU$7-million (US$5.3 million) Project ResQu and the role their advancements will play in emerging, commercially available unmanned systems.
Building on the Basics
Project ResQu is the follow-up program to Smart Skies, a three-year, AU$10 million (US$7.6 million) project initiated in 2008 by ARCAA and its partners, including CSIRO (the Commonwealth Scientific and Industrial Research Organisation, Australia’s national science agency), Boeing Research & Technology–Australia (BR&T-Australia), Insitu Pacific Ltd., U.S.-based Boeing Research & Technology (BR&T), and the Queensland University of Technology (QUT). It was co-funded by the Queensland State Government Smart State Funding scheme. The Smart Skies project team concentrated on advancing key technologies that would enable civilian UAS to be safely operated in non-segregated airspace, thus enabling operators to support the many potential applications for unmanned aircraft.
Dr. Luis Mejias, the Acting Director for ARCAA, says, “With Smart Skies, we focused more on air traffic control and management and developing an automatic program that separates manned and unmanned aircraft.”
In 2011, ARCAA initiated Project ResQu, a follow-up project to Smart Skies, with one of their goals being to evaluate prototype detect-and-avoid technologies on a commercial UAV platform. Due to its size, weight, structure, and proven performance, Insitu Pacific’s ScanEagle was selected as the UAV of choice for the project.
The ScanEagle is a small, low-cost, long-endurance UAS built by Insitu, a subsidiary of Boeing. It was designed for intelligence, surveillance, reconnaissance, disaster and emergency response, aerial survey, and photogrammetry applications.
Project ResQu was co-funded by the Queensland State Government Smart Futures Fund, BR&T-Australia, Insitu Pacific Ltd., CSIRO (the Commonwealth Scientific and Industrial Research Organisation, Australia’s national science agency), and the Queensland University of Technology (QUT).
Project ResQu, which concluded in 2014, comprised four project streams designed to bridge the gaps between research, industry implementation, and public uptake.
Two key areas of the project were the migration of the DAA system, first developed in Smart Skies for a Cessna-scale platform, to the 20 kilogram ScanEagle UAS, as well as the development of an Automated Emergency Landing System (AELS). The AELS could enable the safe operation of the UAV over populated areas in the event of a failure such as communication breakdown or engine failure.
Dr. Dmitry Bratanov, research engineer with ARCAA, joined the Project ResQu engineering team in 2013 to support the DAA advancement for UAS. The DAA migration effort included the selection, design, and integration of new hardware, algorithm, and software optimization, as well as the comprehensive testing of the integrated system.
Two other streams were also part of Project ResQu, although not part of this article. These streams included the development of an unmanned aircraft system for performing Miconia weed surveys (led by CSIRO) and a focus on appropriate regulations to provide assurances for safe UAS operations (lead by BR&T-Australia and QUT).
Downsized and Optimized
Clearly, a critical component of the safe use of UAS in controlled and uncontrolled, or unsegregated, airspace is reliable DAA technology.
The existing DAA technology used on ARCAA’s Cessna 172R (the ASL) relied on the IMU-FSAS, a tactical grade inertial measurement unit from iMAR GmbH. As part of the aircraft’s GNSS/INS positioning technology, the IMU-FSAS forms an extremely accurate, integrated positioning system. However, the IMU weighs more than two kilograms; not ideal for this application.
Bratanov says, “The role of the IMU in DAA is crucial for intruder detection. The IMU data is used initially to compensate for aircraft motion in real-time; then an advanced temporal and spatial filtering is in charge of detection and tracking of targets. We needed an iMAR-like capability but at much lighter weight, size, and power.”
As part of the IMU selection process, the research team conducted a comprehensive trade study of currently available products that might meet the size, weight, power, and cost requirements with comparable iMAR capabilities for the UAS market. Apart from specific accuracy criteria essential for the DAA, the team sought a system that could provide operational robustness and interoperability for the onboard system of a highly dynamic UAV.
Throughout the study, the Project ResQu team flight-tested a number of solutions, including the system they selected: a NovAtel OEM615 dual-frequency, SPAN®, RTK capable GNSS receiver along with Analog Devices ADIS 16488 microelectromechanical system (MEMS) IMU.
The ADIS 16488 IMU features low noise gyros and accelerometers in a small, lightweight, environmentally sealed enclosure. When integrated with NovAtel’s SPAN technology, it enables precision measurements for applications that require low cost, high performance, and rugged durability in a very small form factor.
With tightly coupled SPAN integration, NovAtel is able to provide a low cost, lightweight MEMS solution to achieve superior positioning performance in open sky situations.
Traditionally, accurate navigation using inertial technology was only possible with high-end ring-laser or fiber optic gyroscope technology. New applications are increasingly requiring smaller, lighter and more cost effective solutions for use on smaller vehicles or within sensor assemblies. MEMS technology has improved dramatically over the last few years to answer this need. Compared to their higher grade counterparts, MEMS sensors have relatively large error characteristics that need to be accurately estimated before they can be used in a navigation solution. Key error sources include bias, bias instability, scale factor, and angular or velocity random walk errors. Using a tightly–coupled approach, NovAtel is able to effectively estimate these errors and provide an accurate position, velocity, and attitude solution.
“Compared to other solutions tested during the initial study, the NovAtel SPAN-ADIS solution was the best performer for our particular requirements,” said Bratanov.
The team also used NovAtel’s Waypoint® Inertial Explorer® for reconstruction of the most accurate solution by processing airborne and ground data in differential mode. Waypoint’s Inertial Explorer maximizes the performance of GNSS/INS hardware by ensuring position, velocity, and attitude accuracy.
The tightly–coupled integration of GNSS and IMU data delivers precise results, even when lower grade inertial sensors are used. “Waypoint provided valuable reference data to evaluate required and demonstrated real-time capabilities,” Bratanov said. “The technology provided us with a critical functionality for the range of applications that we’re considering.”
Synchronized and Stabilized
In early March 2013, ARCAA received its first SPAN-ADIS system. Once the NovAtel-ADIS system was delivered, the Project ResQu team commenced evaluation and flight testing of the GPS/INS systems onboard the ScanEagle. “We managed to 3D design the components of the payload, developed the required software, performed integrated comprehensive over system tests, passed flight readiness reviews and received encouraging early performance results,” said Bratanov. “The most exciting thing was that it took us only one week from physical delivery of the just released SPAN-ADIS to the system being successfully flight tested onboard the ScanEagle platform.”
The Project ResQu DAA system is made up of a single board computer with an integrated multicore Graphic Processing Unit (GPU). The computer acquires precisely synchronized data from the multi-pixel Electro-Optical (EO) sensor and the NovAtel positioning system. The data then undergoes multiple stages of processing, which involves stabilization, spatial/temporal filtering, decision making, and taking action.
Later in 2013, the integrated DAA system was successfully tested onboard the ScanEagle UAV using multiple collision-course (and avoidance) encounters with the ASL.
The series of flight tests were performed in non-segregated civilian airspace near an airport located in Queensland, Australia. The tests allowed the detection performance of the system to be assessed under a variety of environmental conditions and refined key trade-off dependencies of the DAA system for a small UAV.
In the world’s first successful flight trials for a UAV of this size, the DAA system successfully detected and immediately reported the collision threat to the ground-based remote pilot of the ScanEagle UAV who implemented a collision avoidance maneuver.
Bratanov said, “During the project, we benefited from regular ADIS-SPAN system updates and refinements, leading to improved GPS/INS system accuracy as the project evolved. This contributed to better image stabilization, which is a key task in the DAA system.”
“The SPAN system provided accurate positioning and attitude data, which is vital to collision avoidance. Moreover, the system provided stabilization of the images acquired by the COTS camera on board,” Bratanov said. “This is another critical part of DAA, as one system requirement is that before the image is processed it must be stabilized and compensated for aircraft ego-motion.”
The research team plans to publish a study comparing inertial measurements techniques and other stabilization methods for the DAA system later this year.