AI-Powered Autonomous Traffic Management Platform – an NVIDIA Solution Showcase

Challenge

> Develop a solution that’s robust enough to run detection and tracking at the edge
> Provide the reliability to handle harsh weather conditions 24/7
> Minimize power consumption
> Provide a solution with a small form factor
> Move from development to production quickly

NVIDIA Solution

> High computational power to meet specific neural network requirements
> Same OS to seamlessly share code between connected devices and the cloud
> Easy configurability using  TensorFlow and other machine vision technologies
> Reduced development cycle
> Standardization for partners who are already used to working with NVIDIA® Jetson™

Results

> The ability to identify objects and classify them with 99% accuracy
> Real-time processing of data and sending it to the cloud
> Accelerated neural network performance across many devices

Products Used

> NVIDIA Jetson TX2

Processing Engines Used

> GPU
> CPU
> Video encoder/decoder

Software Used

> NVIDIA JetPack™
> TensorFlow
> NVIDIA CUDA®> cuDNN

NOTRAFFIC IS HELPING TO ELIMINATE TRANSPORTATION HEADACHES USING THE POWER OF AI

The combination of NVIDIA’s market-leading processors with NoTraffic’s cutting-edge technology empowers customers to quickly adjust to changing requirements and circumstances. This helps cities to manage their entire grid with a more efficient and dynamic approach, solving today’s challenges while preparing for the future of connected and autonomous vehicles.

Autonomous Traffic Management Platform

By digitizing the infrastructure, cities can provide their citizens with advanced mobility services that aren’t possible with the current, disconnected infrastructure. It’s a powerful solution for customizing services for every type of road user to cut congestion, reduce emissions, and improve safety.

NVIDIA Platform

The NoTraffic solution combines hardware and software to optimize traffic flow. The hardware uses machine learning and object detection to identify the different road users approaching the intersection. The software then harnesses AI algorithms to track the various objects, as well as their speed and direction, analyze the incoming data, and manage the traffic phases accordingly. NoTraffic’s close partnership with NVIDIA made it easy to efficiently develop an edge-based AI solution that enables fast processing and accurate object identification, offering a first-of-its-kind platform that responds in real-time to all road users.