AI is a powerful tool that has the potential to disrupt and transform both the conduct of military operations and the management of corporate functions.  At all levels of command, algorithmic technologies are enabling new capabilities which are faster and more powerful than those deliverable by human agents alone. Enabled by ever-growing volumes of data, these technologies can enhance situational awareness and decision support across all domains. At the operational level, AI and the use of machine learning (ML) tools can augment human capabilities to monitor, predict, target, and accelerate the capacity to detect and respond to an adversary. AI can also be used to automate logistics and predict the need for repairs, improving operational readiness. At the organizational level, advanced analytics allow organizations to see and understand themselves and their processes better, identifying potential efficiencies of cost and time. AI-augmented automation can take over repetitive tasks, freeing personnel to take on more complex and demanding activities.

Leveraging AI is central to the priorities of DND/CAF, and its allies and adversaries. Canada’s defence policy, Strong, Secure, Engaged, and the DND/CAF Data Strategy describe a desired end state in which data is leveraged to provide increased efficiency and an information advantage in military operations and corporate applications. Analyzing and interpreting large volumes of data is now well beyond the capacity of human agents alone and achieving this end state will require assistance from AI and other automated decision systems. Defence priority areas such as NORAD modernization, National Defence Operations and Intelligence Centre (NDOIC), and Joint Intelligence, Surveillance, and Reconnaissance (JISR) envisage the incorporation of AI and related technologies in the development of a modernized Command, Control, Communications, Computers, Cyber, Intelligence, Surveillance and Reconnaissance (C5ISR) spine. Canada's defence partners are rapidly defining their own approaches to AI; while China has announced plans to achieve global AI dominance by 2030.

But DND/CAF is not yet positioned to adopt and take advantage of AI. At present, AI initiatives within DND/CAF are fragmented, with each command and environment addressing AI independently. AI maturity varies across DND/CAF, with pockets of significant expertise and lower levels of skills and capacity elsewhere. No roadmap exists to move the organization towards leveraging AI effectively to ensure that investments are coordinated and appropriately governed, or to develop the capabilities, attitude, and skills to implement AI effectively, safely, and responsibly. Without such an approach, DND/CAF risks missing many appropriate opportunities to responsibly employ AI in the conduct of CAF operations, thus failing to realise its many benefits, including operational advantage over potential adversaries. Equally, it risks creating or perpetuating harms through algorithmic or data biases and unanticipated system effects, and the loss of opportunities to improve the business of defence and our corporate stewardship through the capabilities enabled by AI.

To move forward on AI, DND/CAF requires an AI Strategy to guide and cohere efforts towards enabling operations and defence business with AI. This AI Strategy lays out five lines of effort to accelerate the adoption of responsible AI within DND/CAF. The lines of effort described within this document have associated activities to advance implementation in the near term, including the creation of a DND/CAF AI Centre (DCAIC) as a centre of excellence for DND/CAF. This AI Strategy will be followed by an implementation directive laying out responsibilities and timelines for Strategy implementation.

Machine learning can predict system failures on Royal Canadian Navy ships

AI capability needed: Identify impending failure in ship systems using sensor data

AI techniques used: Supervised and unsupervised machine learning, predictive analytics

Value added by AI: Prediction of system failures using Integrated Platform Management System (IPMS) sensor data

Failures in machines can lead to tragic events on board ships, putting sailors in danger and the success of sea operations at risk. Therefore, being able to predict failures in marine systems and to replace equipment before failure occurs offers major user benefits for the Royal Canadian Navy (RCN).

For that reason, in 2018, the RCN reached out to the Defence Research and Development Canada Centre for Operational Research and Analysis (CORA) to explore whether data from IPMS could be used to predict failures on board RCN ships. IPMS was installed during a mid-life refit to help monitor propulsion, electrical, and damage control machinery. Its network of sensors on each ship records data every 0.5 seconds, giving trillions of data points on the ship’s condition and performance. 

CORA took three years of data from four systems aboard select RCN ships—the propulsion diesel engine, the diesel generator, the shaft-line and controllable reverse pitch propellers, and the gear box. It aggregated the data on a five-minute scale and categorized the data as being either normal or preceding a failure. CORA then used corrective maintenance logs and operational deficiency reports to corroborate the failure events.

CORA used this data to train autoencoder neural network algorithms to pinpoint system anomalies associated with failure events. The initial results showed that the algorithms could predict the need for corrective maintenance up to a week in advance 75 percent of the time.

The system’s performance was not perfect, with false positives also produced, but the initial results were promising enough to warrant further testing.

Page details

Date modified: