Vanguard Magazine

www.vanguardcanada.com JUNE/JULY 2020 37 and Your Military Operations ARtIfIcIAl INtellIgeNce peRSpectIVe T he Canadian Armed Forces (CAF) are currently in the pro- cess of procuring a number of Intelligence, Surveillance, and Reconnaissance (ISR) plat- forms and sensors, which will greatly im- prove its ability to collect data in support of operations at home and abroad. Collection value is not solely about the availability of platforms or sensors however, but the ca- pability and capacity to transform massive amounts of raw data into decision-quality information and intelligence. The CAF's second in command, Lieutenant-General Rouleau, among a host of other senior leaders, have regularly stressed this point, but a viable solution to this challenge is yet to be clearly identified. At Xtract AI, we believe Artificial Intelligence (AI) tech- nologies are the most viable. Reducing the fog of war Computer vision and machine learning, two of the most relevant fields of AI to this discussion, have evolved rapidly over the last number of years. Their military ap- plications, however, have lagged civilian applications because of the unique con- ditions in which militaries operate. These challenges, most of which are loosely at- tributed to the "The Four V's of Big Data" - Volume, Variety, Velocity, and Veracity - contribute to the AI version of the Clausewitzian "fog of war". However, there are obstacles to realizing the present and future value of AI, most of which are unique to the environmental con- ditions that military organizations operate in. These include, but are not limited to, insufficient volumes of quality data to train models to the required confidence levels, labour-intensive data labelling requirements and complex systems integration require- ments. Significant resistance to change is to be expected from within military orga- nizations as well, especially given that the real-world operational value of AI for Full Motion Video (FMV) interpretation has yet to be demonstrated.There are a host of other challenges that must also be addressed, such as the fact that C4ISR infrastructure is typi- cally insufficient to meet the computational demands of AI. Nowhere is this deficiency more apparent, or relevant, than in support of real-time decision-making at the edge. Extracting intelligence out of data To overcome these obstacles, the "FMV interpretation challenge" must be ap- proached by pursuing the augmentation — not replacement — of FMV analysts. Trust, whether human or machine, is earned through a well-structured process of force development, generation and em- ployment. An agile engineering process that iteratively delivers incremental im- provements to capabilities, allowing them to evolve from minimally viable to fully operational is also anticipated. AI technologies must then integrate into the existing operational systems architec- ture. If these conditions can be met, AI will have its best chance of evolving be- yond a lab environment to its procure- ment, integration and adoption by mili- tary organizations. Joint Terminal Attack Controllers (JTAC) need help to manage their ROZ (Restricted Operation Zone) while keeping an eye on the action with the FMV. AI can be the tool to help reduce their workload. Whether it be building AI into an ISR platform itself to help reduce bandwidth requirements when transmitting FMV, a Remote Video Terminal (RVT) to support Sensor Operators and JTAC, or into a suite of tools used by FMV analysts at a national reach back capability like the Joint Targeting Intelligence Centre, the opportunities are vast. Innovative solutions To operationalize AI, industry experts must be able to work hand-in-hand with the operators and analysts living these challenges on a daily basis. The Innovation for Defence Excellence and Security (IDEaS) program provides one such opportunity. Xtract AI has already progressed through the IDEaS com- ponent 1A selection phase on multiple proj- ects, and is now in execution of component 1B projects and are excited to be part of the innovative solutions being proposed to the Canadian Armed Forces. Xtract AI is developing core AI tech- nologies targeted at helping solve ISR in- terpretation challenges for the Canadian Military. To maximize value, Xtract AI will soon be seeking sponsors within the Canadian Armed Forces and business part- ners to collaboratively pursue solutions that will help decision-makers leverage FMV to make better decisions faster, to achieve a decision-action cycle advantage over our adversaries. Note: Xtract AI is an approved vendor for key procurement groups, including Government of Canada Approved Source List Tier 1, 2, & 3 AI Vendor. Cornell P. Pich is the Vice President of Business Development at Xtract AI, responsible for advancing growth through the development, imple- mentation and execution of the strategic plan, as well as business and government relationship development. Contact: cornell@xtract.ai 604-653-8628 Sponsored Content Photo (top): The RQ-21 Blackjack belongs to Marine Unmanned Aerial Vehicle Squadron 2. (U.S. Marine Corps photo by Pfc. Nicholas P. Baird/Released).

• Cover