Image Credit: ESB Professional/Shutterstock.com
The attempt to control traffic was first demonstrated during the Roman Empire, but it was 1868 before the first breakthrough was made, when J.P Knight, a railway signal engineer designed the first traffic light, near Westminister Abbey in London which was powered by gas. Unfortunately, however, the device unexpectedly exploded resulting in the death of a police officer, and the use of the device was therefore discontinued.
In 1910, the first automated traffic control system was patented in Chicago in 1910 by Earnest Sirrine. The modern traffic light was discovered in 1918 when New York adopted a three-color system that was manually operated from a tower in the center of the street. Some traffic signaling devices were also created by Garrett Morgan - the inventor of the gas mask. When Morgan witnessed an accident between a carriage and a car, he felt that it was necessary to design a system that would help prevent collisions. He designed an electric traffic light system with an upright pole, which had a horizontal cross-section with the illuminated words ‘STOP’ and ‘GO’.
Following the incident with J.P Knight’s invention, the first automatic signals were not developed in London until 1926 and were activated using a timer. Subsequently, vehicle activated lights were designed in the 1930s. In this system, the weight of the car rolling over half-buried rubber tubes displaced the air in the tubes, which increased pressure producing an electric contact that activated the lights.
An inductive loop device was then developed, which is based on a wire loop embedded in the road, linked to a box used for controlling the lights. An electric current is then transmitted through the loop, and an activation signal is emitted when a steel car body travels overhead.
The traffic volume on the highway can be easily determined using a computer activated guidance system so that traffic is automatically routed onto limited-access highways. Most of today’s cars are provided with global positioning systems (GPS) that connect to a satellite helping drivers to easily and quickly identify their optimum route.
Wireless Sensors for Traffic Control
In 2010, Yousef MK carried out research on an intelligent traffic light control system using wireless sensor networks (WSN).
The WSN is designed to enable system components to communicate. Two algorithms - the traffic system communication algorithm (TSCA) and traffic signals time manipulation algorithm (TSTMA) were developed. The successful operation of the control system depends on the interaction of the algorithms as well as with other system components. The process begins with the interface of the WSN, the TSA, and the TSTMA, which results in the traffic signals changing for specific periods of time.
Communication routes between base stations and traffic sensor nodes can be identified and controlled using the TSCA. A direct routing scheme is used by the TSCA which requires that the distribution of all TSNs must be within the range of the base station. All vehicles are identified and counted using traffic sensor nodes (TSN) and this information is periodically transmitted to the base station. Based on the number of TSNs, system operation is segregated into time slots in which each TSN will operate. The aggregated traffic information is then transferred to the TSTMA in order to fix the duration of the time for traffic signals based on the number of vehicles on each traffic signal.
TSNs are installed into small holes in the center of each lane in the roadbed. These holes are designed to be safe, and protected from the condition of the roadbed and the environment and do not interfere with the operations of the TSN. Sensefields, a company that focuses on traffic technology have recently developed a wireless system for the detection and monitoring of road traffic, demonstrated in the video below:
In 2010, Alonso L carried out a study on ultrasonic sensors in urban traffic driving aid systems. This prototype development incorporated a Hexamite HXN43TR ultrasonic sensor behaving as a transmitter-receiver and its signal conditioner.
The key features of this sensor are:
- 4 kHz bandwidth
- 43.0 ± 5.0 kHz central frequency
- Narrow beam pattern of 8.5° at -3 dB
Stages involved in the signal processing are as follows:
- Noise level is reduced by filtering via a band-pass filter. This filter has a lower cut off frequency of 42 kHz and an upper cutoff frequency of 44 kHz.
- The signal’s envelope is obtained by calculating the Hilbert transform
- Parts with high amplitude are amplified selectively by increasing the signal envelope to the third power, therefore reducing the false echoes which are originated due to turbulences or irregularities in the road; i.e. improving discrimination between the noise and the signal
- Compensation of the ultrasonic wave’s attenuation via amplification using an exponential gain
The resulting signal is compared with a pre-set threshold value for acquiring the distance between the vehicles.
In 2005, Steingröver conducted a study on the reinforcement learning of traffic light controllers adapting to traffic congestion.
The conclusion of this research identified that if traffic lights are allowed to take into account the amount of traffic congestion; i.e. adding implicit traffic light co-operation; this helps algorithms to deliver better results. Future studies may include the evaluation of the chances of global co-operation between traffic light junctions. This can be done by allowing a vehicle to not only account for the next traffic light but also the others that are on its path. Another avenue for research is the issue of rapidly increasing state spaces. One may use hierarchical methods with a divide and conquer strategy for dealing with large problems. States can be categorized into higher-level states, causing easier optimization problems and smaller state spaces.
The third future avenue for research is the simulation of accidents. Local congestions caused by accidents can be solved by traffic light controllers. Adaptive traffic controllers need to be designed that can effectively route traffic around accident spots. The ultimate goal is that this and other reinforcement learning methods be studied and implemented in real traffic systems. Before this goal can be achieved, it is important to obtain more results on the robustness of the controllers, and techniques of acquiring the required sensor capabilities.
Sources and Further Reading
- Ahmad. BA. Development of a traffic light control system using programmable logic controller. Ahmad. BA. Development of a traffic light control system using programmable logic controller. Pages 1–24.
- Yousef MK, AL-karakI JN, Shatnawi MA. Intelligent Traffic Light Flow Control System Using Wireless Sensors Networks. Journal of Information Science and Engineering. 2010; 26: 753–768.
- Alonso. L, Milanés. V, Torre-Ferrero C, Godoy J, Oria JP, De Pedro D. Ultrasonic Sensors in Urban Traffic Driving. Sensors. 2010; 11: 662–672.
- Steingröver M, Schouten R, Peelen S, Nijhuis E, and Bakker B. Reinforcement learning of traffic light controllers adapting to traffic congestion. Pages 1- 8.
This article was updated on 17th February, 2020.