printer friendly version

Traffic doesn’t have to be this way

I thought I was being clever when I moved into a townhouse just 8 kilometres from where I work at the Technews Publishing offices in Randburg, to spare me the 50 kilometre, roughly hour-long commute from the East Rand I’d become used to. The reality is, that the 8 kilometre journey typically takes me about 45 minutes. If it’s raining, I can add another 30 minutes. Load shedding adds about the same. If it’s raining and there’s load shedding at the same time, I might as well phone into the office and ask for the day off.

They say the traffic in Cape Town is even worse. I’m not qualified to comment on the general truth of that claim, but I once had the bad luck to land at Cape Town International Airport for a conference on the night when Justin Bieber held his first concert in South Africa, and my hotel was around the corner from the concert venue. I wasn’t exactly a fan before, but that experience made me a passionate un-Belieber for life.

More effective traffic management is something that would save us all a lot of frustration and wasted time, and it’s one of the areas where artificial intelligence (AI) and big data can have a particularly significant effect. It’s therefore worth considering how the current capabilities of video analytics, coupled with AI, can make our roads more efficient, and safer.

Vanessa Tyne.

According to Axis Communications’ senior key account manager and team lead, Vanessa Tyne, it is possible with current state-of-the-art algorithms for traffic analytics to identify incidents/accidents and traffic jams within seconds, and even prevents them in the best-case scenario. They can be loaded as an application on existing cameras or directly inside smart cameras and other traffic sensors. She says that the ideal would be to have any AI/smart analytics link into a mobile app for drivers that would alert them to potential delays or accidents up ahead. Display boards with the information would also have the same function.

Analytics lie at the core

“With quick detection of incidents using AID (Automatic Incident Detection), it is easier for control centres to be alerted to incidents in real time and then to start preparing for potential congestion. Using this analytic to alert other drivers of the accident via a mobile app can avert secondary accidents as well as alleviate congestion by rerouting drivers. AID is also able to send video clips for investigation at a later stage,” she elaborates.

Intersection control can also be better managed, and congestion alleviated by an analytic optimising the operation of traffic lights, reducing the waiting or queue time at intersections. Parking violation detection is another edge-based analytic that will decrease congestion. In addition to detecting illegally parked vehicles, this analytic can help identify areas and times when violations are most likely to occur, and dispatch staff accordingly. This can make a tremendous impact on compliance – and ultimately your revenue.

Just as in so many aspects of our daily lives, AI can play a major part in the ability to better manage traffic and/or individual vehicles, by making sense of chaotic traffic patterns due to incidents/accidents or just old-fashioned peak hour traffic. Europe has extensive deployment of AID and other traffic monitoring systems and they have alleviated a lot of congestion,” says Tyne.

“As mentioned above, by using intersection analytics to ensure better flow and optimisation of traffic lights, peak time queueing or gridlock situations can be avoided. Deep learning will allow for identification of congestion patterns as well as time patterns – it is all about the type of information being fed into the system. This is also extensively deployed globally. Again, as mentioned previously, this also applies to occurrences of parking violations.”

Good AI depends on good cameras

Tyne insists that cameras need good quality imaging for AI technology to be optimal, as this will increase the performance of the analytics in terms of detection rate and false alarm rate. Good lighting or cameras with built-in low-light technology will assist in ensuring all incidents are detected timeously.

Due to potential IT infrastructure constraints, cameras that have built-in bandwidth optimisation technology will also benefit any deployment. Added to this, scalability and flexibility for expansion and the ability to adapt the solution are important factors to consider, as this will allow for a reduction in operations and maintenance costs.

Axis communications owns a company called Citilog which has the following management solutions for the market (amongst others):

• Automatic incident detection for tunnels, bridges, and highways.

• Traffic data collection. The various products allow for collection of data from the sensors deployed, which collates this information into a user-friendly dashboard.

• Intersection control. Axis’ sensors are used to detect pre-gridlock situations and effectively prevent gridlock at strategic intersections, including intersections with bus rapid transit and tramways.

• New to be launched is parking violation detection – minimising parking violation is beneficial because it results in traffic flow improving over time.

Solutions from a country with epic traffic

If you live in a major South African city like Johannesburg or Cape Town, and you think we have it bad with traffic congestion, do yourself a favour and google how bad it can get in China, a country with more than 1,3 billion people and an estimated 340 million motor vehicles, including 250 million cars. In fact, let me save you the effort: this one ( www.securitysa.com/*jam, redirects to https://www.youtube.com/watch?v=O3kL6nMap2s) looks particularly fun.

Chinese camera manufacturer, Dahua Technology, therefore knows a thing or two about traffic challenges, and offers a range of solutions that cater to improving them. Its traffic solutions work for highways and ordinary streets, with cameras providing accurate data in real time, as well as road map information. They also collect data for transit authorities, and improve road safety by avoiding incidents and reducing accidents on the road. Another core system function is information dissemination, which can help guide traffic with additional information sources such as restrictions, weather, road status, and emergencies.

Specifically, these solutions provide up to 4-lane traffic data counting that distinguishes between vehicle types like passenger cars, large, medium-sized and micro trucks, saloon cars and micro buses. Statistical data gathered includes time, date, statistical period, lane, traffic flow, occupancy, headway, average speed, queue length, and vehicle type, for each lane. They can be applied to traffic signal control, real-time traffic status, parking violations, bus lane enforcement, speeding detection and red light enforcement.

Traffic data is collected by traffic flow cameras using non-intrusive virtual loops, where detection is done through advanced computer algorithms and does not involve any roadworks, as in the case of an inductive loop. It can cover one to three lanes at a time, and has a 99% detection accuracy at speeds below 80 kmph.

Dahua recently launched the DHI-ITC231-RU1A-(IR), a 2 megapixel traffic flow analytics camera. This camera is mainly used in roadway traffic flow analytics, devoted to traffic data counting with an 80 metre range. Detection accuracy is over 99% when the speed is below 80 kmph and covering one to three lanes.

For more information contact:

• Axis Communications, +27 11 548 6780, sasha.bonheim@axis.com, www.axis.com

• Dahua Technology South Africa, +27 10 593 3242, sales.za@dahuatech.com, www.dahuasecurity.com/sa

Share this article:

Further reading: