Home » OPINION: The Road to Sensorless Traffic Management Using AI

OPINION: The Road to Sensorless Traffic Management Using AI

OPINION: The Road to Sensorless Traffic Management Using AI

Smart cities of the future would require smart urban traffic management. Realising this fundamental truth, Siemens reimagined and ideated a solution that is customer-centric and factors in all the variables that are unique to specific urban scenarios, writes DEEPAK GORAY.

Starting on 5th August 1914, when the American Traffic Signal Company installed an electric system at the corner of East 105th Street and Euclid Avenue in Cleveland, Ohio, the humble traffic signalling posts have become an integral part of every urban landscape around the globe. Over more than a century, although the “street furniture” in its three-lighted form has changed little, its operational doctrine has undergone sea transformation. 

While the initial versions of traffic signals were driven manually, with members of the law enforcement deciding which lane to allow, they were gradually replaced by automated variants, controlled by algorithms and a timer, allowing fixed time slices for traffic movements in each lane.  

So, the inference is that by analysing the logic that allows traffic flows, it is possible to construct better programs for orchestrating the traffic signals. For instance, let’s consider the roads connecting the residential areas with the business districts in a city. It is evident that on weekdays, traffic flow will be from the residential areas to the business districts in the morning and vice versa in the evening. It is a basic premise that must be incorporated into all control-logics to avoid traffic congestion and ensure road safety. The programme can be further tuned by factoring in variables like the time of the day and or day of the week cycles, etc. 

While such a pre-programmed, hard-wired approach can manage traffic over broader time-frames, they were insufficient to cater to real-time variations in the traffic scenarios. To align traffic control decisions better with actual ground-level situations, traffic engineers then turned to sensors in the form of pressure plates, induction loops and even RADAR guns to harness field-data for generating insights that can guide the signal-cycles. It was undoubtedly an innovative approach. The traffic loads or queue lengths in each lane on the road were measured, and the signal-cycles were adjusted based on the time taken by the maximum queue length to clear the intersection. 

However, the traffic planners and urban engineers soon realized that managing traffic movement at each junction was only addressing the tip of the iceberg. By 2040, there will be over 2 billion cars and 790 million trucks on roads around the world. If the throughput of the city’s traffic had to be increased, the controllers would have to be actively managed, and traffic flows need to be synchronised beyond the junctions, for the entire stretch of the roads.  

The advent of the digital age made it feasible. With each signal now managing the traffic flows based on inputs from sensor arrays, routed through a centralised control station, the traffic managers can visualise the loads at any given point of time and adjust the signal-cycles to increase throughputs. With the centralised control logic for the broader traffic landscape, all anomalies and timing inconsistencies can be promptly rectified. 

From a Proactive to Reactive Solution
With growing interest in and increasing use cases for automation, coupled with accurate sensing, faster response, and better throughput, we witnessed the emergence of adaptive systems that can scale according to the load. It marked the shift of the traffic management philosophy from a proactive to a reactive approach. Because now, with systems that are smart enough to communicate through centralised controls and respond effectively to the hyper-fluid tactical situation, elaborate traffic management plans were rendered redundant.

However, sensor-based traffic management comes with a caveat, that can limit its universal adoption across economies and urban landscapes. It requires substantial capital investments in expensive equipment like vehicle-borne sensors and radar guns. Moreover, their lifecycle management costs are also high, requiring frequent calibration and maintenance. While most nations across the Global North have embarked on a traffic management transition that gradually shifts from legacy hardware and technologies to more recent ones, developing nations like India can bypass the long journey by opting for more cost-effective yet efficient solutions to manage traffic in over 500,000 km stretch of its urban roads. 

Smart cities of the future need smart urban traffic management. Realising this fundamental truth, Siemens reimagined and ideated a solution that is customer-centric and factors in all the variables that are unique to specific urban scenarios. Apart from its simplicity, resilience and scalability to fit into diverse mission profiles, the product is modular, incorporating value adders that we call ‘Layers of Sophistications’. 

Our basic understanding while conceptualising the solution was that a modern urban leadership while looking for an intelligent traffic management solution would want a scalable product, considering the rate of urban sprawl in the present times; manageable, that can be driven with an optimised workforce; integration compliant, to incorporate best-in-class third party components for capability expansion; ensures maximum throughput in various diurnal phases, weather conditions, etc.; sensorless, to reduce maintenance overheads and above all, economical, to fit within the scope of the urban planning budgets. 

The heart of the system is an industrial-grade PLC that is suitable for multi-scenario applications and runs the actual control logic that has been developed through on-site traffic studies. The control module, forming the ‘First Layer of Sophistication’ can operate with equal efficiency in stand-alone mode or as part of a network, offering adequate flexibility to cater to customer requirements for a discreet or connected experience.

This control module is equipped with a communication attachment and is embedded into a control system architecture, allowing it to interact with the designated traffic command centre (CC). The framework allows the CC to communicate with all the control modules in real-time, where an operator uses a simple, intuitive graphical user interface (GUI) to set the timing of any lane at any junction, remotely. Considering the potential of the CC approach, it won’t be an overstatement to acknowledge the possibilities of managing all the junction controllers in any city from a single command post.

Factoring in Contingencies for Urban Scenarios
While designing the product, our engineers have also considered the contingencies that can surface in typical congested urban scenarios. For instance, we have created the routines to facilitate a ‘Green Corridor’ on specific roads to prioritise the movement of emergency services and first responders. Also, connectivity through standard mobile communication protocol eliminates the need for dedicated networks, and as the control modules can operate independently, with the control system architecture facilitating only supervision, the chances of catastrophic service outages are significantly minimised.  Today, most Indian cities have been brought under a surveillance net, where the live feed from cameras installed across the city are received and reviewed at a command hub. We decided to capitalise on the advantage offered by this centralised location to observe the traffic-situations at the junctions of the city in real-time for our product, making it the ‘Second Layer of Sophistication’. 

It was imperative to study the traffic load on the fly, to tune the throughput to our customer’s satisfaction. The same was achieved by accessing proprietary databases that track the accumulation and movement of mobile phone users on the roads. The data was loaded on a cloud computing platform, adding a series of validations and conditional filters to reconstruct near-life traffic scenarios. A model was built, using the output, based on neural networks that can assimilate various natural and human- made variables and the inputs from the traffic managers to arrive at an ideal time series for controlling the signal-cycles to yield the optimum throughput. 

To reinforce the decision-making process, we also analysed the video streams from the traffic surveillance network to estimate the right time frames that can be communicated directly to the control modules. Alternatively, the feed can be enriched with the outputs from our neural network model, reviewed at the command centre, and communicated to the control modules installed at the junctions using the control system architecture. While the second option may be marginally slower, it significantly improves the throughput as the intervention through artificial intelligence (AI) allows us to consider all possible scenarios and come up with the perfect suggestions. 

While our system is comprehensive, access and user experiences have been further notched up through its marriage with mobile device management. The system administrators can use a secured smartphone-based application to access the system environment for managing respective functionalities, making it the ‘Third Layer of Sophistication’. 

Thus, by leveraging the advantages of AI and real-time visual feeds, we demonstrated the feasibility of constructing a system that can eliminate the role of costly sensor arrays from the traffic management decision loop, while retaining enough precision to meet the requirements of the customer. It is heartening to observe that our proof of concepts (POCs) have been able to compete and exceed the performance horizons of world-class sensor-based traffic management systems. We intend to incorporate all the learning in our ongoing developments, as we aspire nothing less than delivering the best traffic management experience and maximum system control for all conditions, at a competitive price. 

Deepak Goray is Head of Smart Cities at Siemens India.


Leave a Reply

Your email address will not be published.