DSTO & R&D: MAGIC aims at autonomy | ADM Nov 2010
Jointly run by the Defence Science and Technology Organisation (DSTO) and the Research Development & Engineering Command (RDECOM) in the US, the Multi Autonomous Ground-robotic International Challenge (MAGIC) has seen the field narrowed down to five final contenders.
Steve Butler, DSTO | Adelaide
The next big challenge in the use of unmanned ground vehicles (UGVs) in land operations is autonomy and cooperation; enabling them to operate and conduct specified tasks with minimal human control.
The use of autonomous unmanned vehicles has the potential to greatly reduce the risk to soldiers required to carry out hazardous duties such as detecting improvised explosive devices (IEDs).
Robots have the potential to play an increasingly significant role in replacing and supporting soldiers by undertaking intelligence, surveillance and reconnaissance operations as well as reducing the overall workload of the soldiers by performing routine and laborious tasks.
Australia’s Defence Science and Technology Organisation (DSTO) has long recognised a technology gap in the application of un-manned ground robotic systems in land operations and has looked for ways to accelerate the development of the technology with a view to field such vehicle systems in our future operations.
DSTO decided to stage an international competition which is designed to challenge the minds of the world’s leading robotics researchers and developers to focus on the development of next generation fully autonomous vehicle systems that can operate effectively in military operations and civilian emergency situations.
This competition known as MAGIC 2010 (Multi-Autonomous Ground-robotic International Challenge) is jointly organised by DSTO and RDECOM of the US Army.
The MAGIC Grand Challenge will be held at the Adelaide Showgrounds in early November 2010.
Steve Quinn, Chief, Land Operations Division, DSTO, is the Australian lead and said that the MAGIC 2010 initiative was designed to test the ability of multi-vehicle cooperatives to autonomously and dynamically coordinate, plan and execute an intelligence, surveillance and reconnaissance mission in a changing urbanised environment while simultaneously providing a unified situational awareness picture.
“The MAGIC goals support DSTO’s enabling research program in Force Protection by leveraging the technology push into operations,” Quinn said.
“We wanted to encourage the generation of the most innovative proposals from around the world with the widest prospect of achieving advanced solutions in autonomous multi-vehicle systems.”
The overall program goals include:
• Acceleration of the development of autonomous and unmanned vehicle technology in areas that include: task allocation, multi-UVS control, tactical behaviour, machine intelligence, dynamic planning and re-planning, data and sensor fusion, human-machine interfaces, multi-aspect situational awareness, and systems integration.
• Improvement in the current human-to-vehicle control ratios by demonstrating that multi-UVS cooperatives can operate effectively with limited supervision by humans in realistic environments.
• Shifting the perceptions within the technical and operational communities regarding the state-of-the-art of autonomous multi-vehicle control by demonstrating the augmentation of realistic military activity in changing environments.
• Attracting and energising a wide community of participants to bring fresh insights to the problem of developing robust autonomous multi-vehicle cooperatives and to identify and transition technologies to meet emerging requirements.
Of the culminating MAGIC Grand Challenge itself, Quinn, said: “It was quite a challenge we set the competitors.
“The mission tasks included mapping the environment, detecting, locating, classifying, recognising, disabling and tracking a number of friendly and hostile targets.”
The MAGIC 2010 Challenge was launched by the then Minister for Defence Personnel, Materiel and Science Greg Combet on July 13, 2009.
Combet’s announcement included the details of the prize money given as research grants - $US750,000 for the winner, $US250,000 for the runner-up and $US100,000 for third place.
In addition, the initial short listed teams received seed funding to develop initial proposals into prototypes.
In all, US$1.85 million was outlaid in research grants.
A total of 23 entries were received from the US, Canada, Poland, Japan, South Korea, Turkey and Australia.
The entrants were required to submit a highly detailed technical submission.
Those submissions were analysed by MAGIC’s Technical Assessment Panel (TAP) which consisted of robotics experts from Australian and US departments of defence.
Initially the following 12 teams were short listed, of which 10 were funded and a further two were invited to take part using their own resources:
Australia
• MAGICian: A consortium of University of Western Australia (Robotics and Automation
Laboratory, Adaptive Systems Research Group), Flinders University (Artificial Intelligence and Intelligent Systems Laboratories), Edith Cowan University (Artificial Intelligence and Software Engineering Cluster), Thales Australia (D3S&A, Naval Division), ILLIARC Pty Ltd.
• STRATEGIC ENGINEERING (an Australian robotics company specialising in advanced sensing, vision systems, autonomous vehicles, industrial robotics and field robotic platforms), with the University of Adelaide.
Canada
• NORTHERN HUNTERS – Canada (Comprising Amtech Aeronautical Ltd, Scientific Instrumentation Ltd, Royal Military College of Canada.
Turkey
• CAPPADOCIA – Turkey (Comprising ASELSAN (Turkish military electronics company) with Bilkent University, Bogazici University, Middle East Technical University from Turkey, and Ohio State University (Control & Intelligent Transportation Research Lab) of the USA.
US
• RASR - Reconnaissance and Autonomy for Small Robots Team – (Lead: Robotics Research, LLC; with Industry Partners: General Dynamics Robotic Systems, Qinetic-NA, Del Services, Cedar Creek Defence University, Carnegie Mellon Robotics Institute, Embry-Riddle Aeronautical University, University of Michigan.)
• TEAM CORNELL, Cornell University – (School of Mechanical and Aerospace Engineering.)
• TEAM MICHIGAN – (Comprising SoarTech with research support from the University of Michigan.)
• TEAM VaCAS – (Department of Mechanical Engineering, Virginia Tech)
• UNIVERSITY OF PENNSYLVANIA – (With BAE Systems as auxiliary team members.)
The 2 teams short listed for self-funding were both from Australia:
• NUMINANCE - Australia (Comprising Numinance Pty Ltd and LaTrobe University with support from 10 small companies).
• UNIVERSITY OF NEW SOUTH WALES – Australia (Teaming with Kumamoto University, Japan, and University of Western Sydney)
The aim was to assess the technical progress made by short listed teams mid way through the year where they needed to demonstrate technical competency in specified areas.
An exhaustive assessment was conducted in June and July this year when the TAP visited each of the 12 short listed teams across five countries.
Presentations and demonstrations focussed on convincing the TAP members that teams had a good extant knowledge of: autonomous unmanned ground vehicle design; multi-UGV autonomy and coordination; signal, sensor and image processing; navigation and mapping; processing and fusion from the simulated UAV feed; situational awareness and information fusion; human-machine interfaces; and mission operations relevant to the challenge.
On July 27 the five finalists to compete at the MAGIC Grand Challenge were announced.
These were: Cappadocia, Magician, RASR, Team Michigan, and University of Pennsylvania.
These teams gathered at the main arena of the Royal Adelaide show grounds from Saturday November 6 to compete in the week long Grand Challenge.
To complete the challenge they were required to: (i) accurately and completely explore and map the challenge area; (ii) correctly locate, classify and recognise all simulated threats; and (iii) complete all phases within 3.5 hours.
The competition was conducted in three increasingly complex phases.
The teams were required to field a minimum of three robots and demonstrate cooperation and autonomy – while conducting a series of complex ISR, mapping, navigation and object location/designation operations.
With the Grand Challenge complete, the next phase of MAGIC is the most important phase where the results will be analysed to answer such questions such as how successful was MAGIC in achieving the specified objectives, how much did we learn from the exercise and how much did it teach us that we didn’t know.
The success of MAGIC will depend on answers to these questions.
The winner of MAGIC will be announced at the Land Warfare Conference in Brisbane this month.