Suncor Energy Inc.

Synopsis

Fastloop architected, developed and automated, hybrid AI infrastructure designed for mission-critical anomaly detection across Suncor’s fleet of drones. This solution bridges the gap between edge computing intelligence and cloud scalability while deploying AI technology on top of autonomous drones to lower operating costs and improve site security and risk detection simultaneously.

Highlights

Deployed visual intelligence via an Edge-to-Cloud infrastructure

Scaled AI technology on top of autonomous drone hardware and software

Reduced the window from data sync to notification to just 45 seconds

Achieved 100% recall in anomaly detection

Challenges

Suncor has operations across North America, including highly regulated environments in remote locations. Combined with the vast surface areas, weak network / signal strength, inaccessible locations, and challenging weather conditions, tasks such as perimeter security, site monitoring, and wildlife encounter significant complexities. Suncor’s ability to follow, detect and manage regulatory compliance has traditionally taken significant time, manpower and equipment to solve.

Core challenges include:

Limited bandwidth forces systems to rely on edge processing. Hardware constraints at the edge limits on model performance while maintaining a continuous operation.

Detect anomalies reliably and efficiently despite heavy snow, rugged terrain, and other challenges.

Moving to a production-ready state requires navigating the intersection of mandatory utility standards, global security management protocols, and strict aviation regulations.

Solution

Partnering with autonomous drone leasing provider, Drone-Lytics, Fastloop built custom AI solutions to advance Suncor’s investment in utilizing autonomous drones, video data (thermal, RGB and geospatial), imagery and on-site technologies. This enabled security detection, object identification and human and wildlife observation, among others, where multimodal AI became a team of anomaly detection experts. Due to the remote sites and limited connectivity, Edge-to-Cloud processing was required, delivering real-time anomaly triggers, detections, analytics and agentic notifications drastically lowering man hours and fixed costs while improving speed and accuracy.

Key components of the solution:

Drones collect footage of the grounds and start uploading videos as soon as were autonomously docked
Drone footage is automatically ingested into the onsite Edge Computing device for AI-driven anomaly analysis
Vision language models (VLMs) start anomaly analysis on Google Distributed Cloud
Data is transferred to Google Cloud Platform where data is aggregated with historical flights and quality assurance evaluates the results
Real-time updates are customized into dashboards
Alerts are delivered to staff members, notifying the anomaly type detected and its map location, timestamp for response

Edge to Cloud AI pipeline
Combined the real-time processing capabilities of edge computing, including a specialized database solution, and the installation of on-premise hardware, and Google Cloud’s performance and scalability for AI workloads.

Agentic notification engine
Engineered a custom front-end and autonomous agent that monitors anomalies in real-time, triggering role-based alerts to field personnel.

Strategic AI tooling mix
Achieved an AI inspector to quality assurance team by leveraging a strategic mix of pre-trained architectures. Achieved production-grade accuracy within the strict resource constraints of edge-native environments.

Results and Impact

Leveraging the state-of-art visual capability of multimodal large language models (MLLMs) and drone footage, Fastloop replaced manual, labour-intensive visual inspections with automated AI-driven analysis. This hybrid approach combines edge and cloud computing, leading to improved efficiency, automation and compliance.

Client testimonial about their experience working with Fastloop.ai and the impact of the solution.

– Client name, Job title

Why Fastloop.ai?

Drive value with LLMs in production

Grounded in engineering proficiency, Fastloop ensures that knowledge for cutting-edge LLM capabilities in theory translated into deployment, while fostering a culture of shared innovation and transparent communication.

In-house agile team that is willing to go the extra mile

Using agile methodology, the Fastloop delivery team moves beyond the standard vendor-client relationship to take true ownership at every step of the project. The team proactively problem-solves, remains flexible and responsive to evolving needs, and consistently meets the client’s expectations.

Vision for long-term partnership via fast iterations

Fastloop balances speed with quality, making sure value is created for their client as early as the proof-of-concept stage. The team works quickly and iteratively, helping the client hit quick wins while embracing the vision for a long-term partnership.

Future Outlook