I am a member of the Emerging Talent Community as part of the SAP Supply Chain Module. Last summer, I was stuck in traffic while the never-ending upgrades to the motorways were happening. I needed to research a modern-day problem and solve it for my dissertation and this got me thinking about what was actually happening to the motorway, and what else I could add to the design. I decided to model a smart motorway by building a simulator and designing a new smart traffic light system.
As part of my Masters in IT, the dissertation consisted of two main ideas. The first was to model a smart motorway in a simulator, and the second was to model a new smart traffic light system. The simulator was designed to last for 2 hours using real data at rush hour from the M3 motorway before it was upgraded. In the designing of systems, often the focus can be on the technical level and ignore how the real-world impacts upon it. This is one of the key aspects I wanted to focus on.
The design of the smart motorway required several models which decide how the vehicles interact on the motorway. A lane changing model was designed to simulate the decision process of a driver. This model allows for when a driver randomly changes lane regardless of other factors such as speed or traffic joining the motorway. The other model was the car following model. In this model, it is presumed that the speed a vehicle is travelling at is controlled by the vehicle in front. The output from this model links directly into the lane changing model, and these models are combined to simulate motorway traffic, including the more crazy drivers out there!
Smart traffic lights
One of the key ideas in my dissertation was to design and model a new smart traffic light. The smart traffic light system is designed to continuously monitor the traffic flow on the motorway. Unlike a traditional timed traffic light, the smart aspect of the system can analyse gaps in the traffic before the slip road. The traffic queued to join the motorway is also being continuously monitored in terms of the size of the vehicles. If a large vehicle, such as a lorry is identified, the smart traffic light calculates that a larger gap in the traffic is required for the lorry to join the motorway. At this point, the traffic light is working in a ‘perfect’ world where drivers can then join the motorway. However, a driver may not join the motorway and sometimes there is no reason at all! In order to simulate this, a driver behaviour model was used.
This model is designed to simulate a driver having the final decision to join the motorway and his decision is based on both the drivers and vehicles ability. So even if the smart traffic light calculates a gap in the traffic for a car to join, the simulated driver may decide not to join and sometimes there is no valid reason. I now needed to prove with data from the simulator that firstly the models worked, and secondly, it improved traffic flow.
The traffic flow is measured as congestion alerts which are monitored by cameras and sensors. The graph below shows a comparison between the number of congestion alerts, with and without the smart traffic light system in the simulator. The lane readings were simulated to be one mile after the traffic had joined the motorway. With the smart traffic lights, there were no congestion alerts and overall, the smart traffic light system reduced the number of congestion alerts by 80%, proving that the system is ‘smart’.
Applying the smart motorway idea
The main guiding principle behind the smart motorway paper was to consider how a system adapts to random driving events. This can be applied directly into supply chain planning as often the unpredictable happens. The use of smart traffic lights can be used in warehouses, factories, construction sites or adapted to maritime, and especially in automated scenarios, where the vehicles need to interact which each other. It can be adapted to any scale and is a versatile system that adapts to people’s random actions on the road. The smart traffic light system considers how an apparent free flowing system, such as a motorway, can be made more efficient. The driver behaviour models are also something that can be applied to self-driving cars. A smart motorway is the first step towards having them on the roads.
Publishing the paper
I have turned my dissertation into a paper, in collaboration with Bournemouth University. I am being sponsored by Capgemini to present the paper in Sydney at the 13th international conference on Complex, Intelligent, and Software Intensive Systems. The conference proceedings will be published by Springer Publishing Company and I will be writing another blog after the conference on some of the new ideas raised and how they could be applied.
To find out more about the conference, click here: http://voyager.ce.fit.ac.jp/conf/cisis/2019/index.php
Ed graduated from Bournemouth University with a Masters in IT in 2018. His main area of research has been around smart transport including smart motorways and self-driving cars. Ed is a member of the ETC community in the SAP Supply Chain practice.