Imagine a world where a lamppost does more than just illuminate streets; it actively contributes to a smarter, safer, and more sustainable community. Using Amazon Web Services (AWS) and NVIDIA technologies, Omniflow is turning this vision into a reality.
Omniflow’s smart lampposts use artificial intelligence and machine learning (AI/ML) algorithms and computer vision to help communities become more sustainable and energy efficient. By adopting edge computing and a cloud-based technology stack, the company has achieved significant operational benefits—and is helping its customers reduce carbon emissions with lighting powered by wind and solar patented technology.
Opportunity | Using AWS and NVIDIA Services to Power Advanced AI/ML Use Cases for Sustainability with Omniflow
Founded in 2012, Omniflow provides smart-city solutions to government entities as well as private businesses. Traditional lampposts require a significant amount of energy to operate and can be costly to maintain. To reduce grid power consumption, Omniflow retrofits old lights with its smart lamppost technology. This solution is not only more sustainable than traditional options but also transforms lampposts into valuable sources of data.
“Our smart lampposts are powered by wind and solar at the edge and completely integrated into a single infrastructure,” says Pedro Ruão, founder and CEO of Omniflow. “We use these assets to provide many other services, such as air-quality monitoring, computer vision, and traffic analytics.”
With Omniflow’s lampposts, its customers can track various data points related to their environment and make decisions to improve efficiency, sustainability, and public safety. As the company started to incorporate more features, such as cameras and sensors, it needed more computational power to process data and run AI/ML algorithms. To resolve this challenge, the company joined NVIDIA Inception, a free program to help startups evolve faster through cutting-edge technology.
“During NVIDIA Inception, we were able to purchase some initial hardware for testing,” says Ruão. “We liked the results, so we started working with the NVIDIA and AWS teams more closely.” To dive deeper into AWS services, Omniflow also joined the AWS Sustainable Cities Accelerator program—an initiative designed to help mature startups accelerate their impact, access additional resources, and expand their reach.
By gathering data at the edge on AWS, we can help our customers optimize their environments.”
Founder and CEO, Omniflow
Solution | Adopting Amazon EC2 P4 Instances and NVIDIA Jetson Nano to Process Computer Vision Data at the Edge
Omniflow migrated to a new system built on Amazon Elastic Compute Cloud (Amazon EC2) P4 Instances. These instances are powered by NVIDIA A100 Tensor Core GPUs and deliver industry-leading high throughput and low-latency networking. Omniflow is currently using Amazon EC2 P4 Instances with NVIDIA TAO Toolkit, a service that helps quickly train and customize object detection models powered by AI/ML at the edge.
Within each smart lamppost, Omniflow has embedded NVIDIA Jetson Nano, which brings accelerated AI performance to the edge. After an AI/ML model is trained, it is deployed to NVIDIA Jetson Nano devices in each lamppost. The models are then used to perform different functions, such as analyzing traffic footage or assessing pollution levels. “We were already performing edge computing, but computer vision requires a much more powerful standard of performance,” says Ruão. “With NVIDIA Jetson, each of our lampposts has the computing power to process two to four cameras and extract all the necessary metrics, plus perform air-quality monitoring.”
To send this data back to AWS, Omniflow uses AWS IoT Core, a service that helps developers connect billions of Internet of Things (IoT) devices and route trillions of messages to AWS services without managing infrastructure. This information is processed using functions running on AWS Lambda—a serverless, event-driven compute service—and stored in Omniflow’s database. These AWS Lambda functions activate specific actions based on the incoming data, such as abnormal air-quality levels or unusual traffic patterns.
For instance, if the air-quality sensors detect elevated levels of pollutants, an AWS Lambda function activates an alert to municipal authorities. Similarly, if the computer vision models identify a traffic jam or an accident, the system automatically alerts traffic management centers for immediate action. “By gathering data at the edge on AWS, we can help our customers optimize their environments,” says Ruão.
Although the lampposts are designed to run continuously, they don’t require constant energy…