The intelligent traffic management system takes a step forward and helps evaluate the road traffic conditions (in real-time), forecasts, and informs public/private commuters about the traffic situation.
The solution presented here is a rugged AI platform for traffic management systems. It relies on Lanner's rugged platform LEC-2290 and the Hailo H8 PCI Acceleration Card (Industrial-grade SKU). The combination of this appliance with the PCI card enables machine vision and video analytics for traffic management.
Challenges of traffic management systems
A real-time and smart traffic management system could help cities improve commuter travel time (private and public transport), enhance air quality, control traffic speed, reduce accidents, protect sensitive roadside areas, and a lot more. But despite these technological benefits, implementing such an efficient and hyper-connected traffic management system can be quite challenging.
The most noticeable challenges include:
- Network limitations. Even though technologies like Vehicle to Everything (V2X) and 5G mobile broadband could help improve smart traffic infrastructure communications, the massive amount of data generated by distributed smart city cameras and sensors could overwhelm any network. The networks become a limiting factor, primarily if the intelligence, AI, Machine Vision, or data analytics are hosted in the cloud.
- Cybersecurity. Traffic management systems are prone to cyber-attacks, risks, or threats, especially in the field. Field devices and their communications are vulnerable. Anyone roaming around street posts could attempt to steal or control traffic data without the additional physical barriers. The traffic signal, CCTV, transport, and vehicle-to-Infrastructure (V2I) must be protected and ready to respond.
- Time-sensitive application. A traffic management system may use many distributed roadside devices that gather massive amounts of data which usually needs to be processed fast. Most traffic applications are time-sensitive. For instance, vehicles on the road and pedestrians need immediate responses. A traffic management and monitoring system should be able to produce responses in real-time or near real-time.
- High maintenance and operation costs. Interconnecting all kinds of different technologies to collect data, transport it, process it, and analyze it can become an expensive investment. In particular, the networking requirements for a traffic management system can be costly when low latencies and high bandwidth are required.
Solution: The Rugged AI Platform for Traffic Management.
The proposed Rugged AI Platform for Traffic Management solution can help deal with the abovementioned challenges. In order to reduce the need for faster, more cost-efficient networks and insecure on-site devices, the intelligence for traffic management must be moved as close as possible to where data collectors such as video cameras or sensors are deployed (rather than on the cloud).
The edge (roadside, vehicles, pedestrians, traffic lights, etc.), which is the closest to where data is being generated, must be empowered. For instance, the street poles with active electronics compartments can be used to host edge computers capable of collecting and processing edge data. Other optimal network roadside edges can also be network operators or telcos edges, such as broadband antennas with distributed cloud or multi-access edge computing (MEC) equipment.
Hailo H8 PCI Acceleration Card (Industrial-grade SKU)
Recently, Lanner collaborated with leading AI chipmaker Hailo to design and create the Falcon H8— the first Hailo-8™ AI-powered PCIe accelerator card. This collaboration enables scalable and robust intelligent video analytics for applications like traffic management and Intelligent Transport Systems (ITS).
The Hailo Falcon H8 is one of the most cost-efficient PCI acceleration cards on the market. It is known for its low power consumption and ultra-high Tera Operations Per Second (156 TOPS). This unparalleled level of performance allows deep learning applications on edge servers located anywhere in the smart city, such as smart light poles or telco’s MEC.
Hailo’s Falcon H8 acceleration card requires a standard PCI interface included in an NVR or edge AI appliance such as the Lanner’s LEC-2290. Lanner’s LEC-2290 is an Nvidia NGC-ready edge AI appliance with support for Intel® Core i7-8700T/i7-8700 CPU (codenamed Coffee Lake S). The appliance is designed for AI video analytics and machine vision for smart city solutions like physical security or traffic monitoring. Having LEC-2290 running the Falcon H8 allows running video-intensive traffic management and monitoring workloads.
How does the solution work?
The edge AI appliance gathers data from the IP video cameras mounted anywhere from lighting poles, buildings, or smart poles deployed across the city. Having an edge AI platform allows traffic management and monitoring applications to become independent from cloud computing.
The video stream collected from these cameras gets processed through AI video analytics and machine vision. The output is then sent to a high-end server. The high-end server, in turn, can run a Video Management System (VMS) such as Hailo’s for further monitoring or management.
Benefit.
The solution proposed here (Lanner’s LEC-2290 + Hailo H8 PCI Acceleration Card) is an AI-based traffic video analytics and monitoring system for smart cities. Overall, it allows the implementation of traffic management applications such as adaptive traffic lights, vehicle prioritization, parking access and detection, electronic tolling, and more.
Aside from the apparent benefits, the solution can also provide the following advantages.
-
Autonomously monitor traffic cameras. The traffic management solution allows real-time AI-video analytics and reactive/proactive alerting capabilities. It can help monitor and detect accidents with more precision, help reduce congestion (and pollution), and even optimize public transportation infrastructure.
- Improve remote road intersection operations and management. Allow operators to remotely manage traffic and introduce automation to traffic lights so that they can adapt and cooperate to reduce traffic or allow safe passage for first responders.
- Introduce smart traffic management applications. Having access to AI on edge facilitates allows many smart traffic applications such as ALPR (Automatic License Plate Recognition), which can be helpful for enforcement, entrance to parking lots, or automatic tolls. Other applications might include traffic enforcement, such as monitoring traffic and detecting traffic violations.
- Allow other mission-critical smart city machine vision applications. Aside from facilitating traffic management and monitoring, the smart AI-based camera system can also introduce different MV (Machine Vision) applications such as real-time surveillance and pedestrian abnormal behaviors detection.
Next Steps.
For more information on other embedded intelligence edge appliances, or the rugged AI platform for traffic management solution please contact Lanner’s sales representative.