An innovative traffic light recognition method using vehicular ad-hoc networks – Scientific Reports

In this study, an innovative traffic light recognition method is proposed. VANET is used to send traffic light status from Traffic light to neighbors such as RSUs and vehicles. Current and next traffic light signal colors are disseminated to neighbors. Vehicles and traffic light (TL) controllers were connected to VANET. Drivers can be informed of their current TL status, time to change, next TL status, and a recommended current speed. The dissemination of this information is also allowed to use VANETs.

A framework for traffic light recognition using VANET is shown in Fig. 1. This framework proposes an efficient system that includes:

1.

TL recognition using dissemination
1. a.
  
  Location of current TL bounding box.
2. b.
  
  Current TL status.
3. c.
  
  Clock for time to change.
2.

Recommended vehicle speed.
3.

Dissemination of neighbors.

Traffic lights are used to disseminate TL position, time, and status using a VANET. The TL works as RSU. In Fig. 2, the TL and vehicles communicate using DSRC and are divided into three sections. In Sect. “Traffic light recognition”, the traffic light recognition and waning dissemination algorithm using a VANET are described. In Sect. “Recommended vehicle’s speed”, the prediction of the recommended vehicle speed is presented. In Sect. “Dissemination”, message dissemination to neighbors is discussed.

Table of Contents

Traffic light recognition

There are two types of roads: main and subway roads. In Fig. 2, it is assumed that the road intersection is large and that the main road has a faster average speed (90 km/h) than the subway (60 km/h). In popular TL, the light status is green, yellow, red, or green. Each road type had two opposite directions. Therefore, there are four directions. Each direction had a traffic light box on the side of the intersection. If it is assumed that the main road lasted 40 s until the green light changed to red, it is assumed that this time was more than 30 s until the green light changed to red on the subway road.

Recommended vehicle’s speed

The traffic light signal information displayed in the vehicle is the current signal state and the time until change. The indicator in the instrument panel shows the countdown timer with the predicted time taken to go green. It is blank if it cannot be predicted. The data analysis predicted the best-recommended speed. In Fig. 3, traffic light information appears on the dashboard of the vehicle. Traffic light color, the timer of seconds until changes, and currently recommended speed appear on the dashboard.

The predicted duration, which is the suggested speed message duration, suggested speed, and suggested speed message time, was calculated using Eqs. (1), (2), and (3), respectively. At each prediction duration, the suggested message is sent to the driver on the dashboard. Equation (1) shows when to suggest an appropriate speed for the driver. The suggestion time was measured in seconds. The forecast time depends on the number of messages sent during the time between traffic signal changes.

$${D}_{P}=\frac{({T}_{E}-{T}_{B})}{N}.$$

(1)

${D}_{P}$ is the prediction duration (s), ${T}_{E}$ is the TL change time, ${T}_{B}$ is the prediction start time, and N is the number of messages. The suggested speed was the distance from the current position of the car to the traffic light. Velocity equals distance over time. The predicted distance and time are small, and close to intersections. The distance is not linear in a free space such as mobile ad-hoc networks.

While vehicles in VANET have constrained mobility and can only travel along the paved road network, nodes in mobile ad hoc networks are assumed to move freely in any direction. The topology of the network changes quickly in the VANET because nodes are moving vehicles⁷. Computing the distance between the vehicle and the traffic light is made easier with the aid of a digital road map. The road digital map is mapping earth coordinate into map pixels to calculate road distance. The predicted distance is computed for the real paved road from the vehicle to the traffic light using road map in the simulator used in testing to guarantee non-linear nature of VANET model²². Vehicular network simulation (Veins) is built on top of OMNET + + and simulates urban mobility (SUMO). In SUMO, distance is calculated using Time, in the proposed algorithm, is not affected by traffic jam delays because all drivers are advised of the recommended speed which guarantees low density and low jams at traffic lights and intersections. Therefore, there is no need for adding prediction delay to computations.

The driver should pass quickly before the traffic light turned red. However, this must not exceed the speed of the road. Otherwise, a warning must be sent to the driver. If the suggested speed exceeds the road speed, a slowdown warning is sent until the driver stands at a red traffic light. To convert from meters per second to kilometers per hour, the speed is multiplied (1000/(60 × 60)) as in Equation 2.

$$\mathrm{Sc}=\frac{D*3600}{{T}_{R}*1000}.$$

(2)

S_c is the suggested speed (km/h), D is the distance to the traffic light and ${\mathrm{T}}_{\mathrm{R}}$ is the time that remains until TL changes. Where S_c must not exceed the road speed. If the speed exceeded the road speed, a slow-down warning was sent to the driver. If the suggested speed was higher than the road speed, then the next TL status time was added to the remaining time as in Equation 3.

$${T}_{R} ={T}_{R}+ {T}_{S}.$$

(3)

${T}_{R} is$ the time remaining until the traffic light changes (s) and ${T}_{S}$ is the next TL status time. Algorithm 1 shows the message sent to the dashboard of the vehicle to suggest speed limits to the driver. Traffic light enabled VANET to send time until change (${T}_{R})$ and distance to TL (D) to its neighbors in step 2. If the TL status is red, the minimum speed is provided in Step 7. If the suggested speed was higher than the road speed, the next TL status time (T + 1) was added to the remaining time (${T}_{R}$). If the TL status is green, then the minimum speed is provided in step 14. If the vehicle is far from the TL and the suggested speed is low, another road speed is calculated in step 23. Otherwise, the suggested speed message is sent to the vehicle using VANET in Step 25 to advise the driver. Table 1 shows an example row from a group member table extracted from beacon messages sent from vehicles N1, N2, and N3 to the TL. The table contains the information recording time (s) for the recorded vehicle number, as well as the location of the source vehicle of the beacon message at the x, y, and z coordinates, speed, road number, and road type. Table 2 lists the notation of the algorithm.

Table 1 Example of a group member table.

Table 2 Notations of algorithm 1 and their descriptions.

Dissemination

TL recognition methods can be divided into detection and tracking methods. Both are enabled because the TL is sent from the traffic light via VANETs. Figure 4 shows that the traffic light is supported by the VANET. The TL disseminates the traffic light status and timer until it changes to the neighboring RSU and vehicles. The TL status, warnings, and recommended speeds were sent to the vehicle and neighboring vehicles. TL’s status was green, yellow, or red. The warnings included emergencies, accidents, and closed roads.

After computing, the message is transmitted as shown in Algorithm 2. The recommended speed is computed in Section B. Intelligent traffic-light warning messages are defined in Algorithm 2. For traffic light recognition, VANET is a model in which cars move on paved roads. Messages about traffic lights are sent via VANETs to cars. They pass near stable RSUs which are used to define location of vehicles.

Table 3 lists the notation used in Algorithm 2. If there is an emergency, an ambulance will receive a warning for a road branch with a red light. A warning is delivered to automobiles behind them if there is an accident or a closed road route. If a road branch has a red light, a typical car is notified before the red light. If a road branch has a green light, typical vehicles are sent before the green light. If a road branch has a yellow light, typical vehicles are sent before the yellow signal.

Table 3 Notations of Algorithm 2 and their descriptions.