Analysis and Comparative Study of Signalized and Unsignalized Intersection Operations and Energy-Emission Characteristics Based on Real Vehicle Data

Abstract

With the development of the economy, urban road transportation has been continuously improved, and the number of motor vehicles has also increased significantly, leading to serious energy consumption issues. As critical nodes in the urban road transportation network, intersections have become a focal point of research on vehicle energy consumption. To investigate whether traffic signal lights affect fuel consumption and emissions, this study analyzed the operating characteristics, fuel consumption, and emissions of intersections with and without traffic signal lights using real-world vehicle data. The data from the INTERACTION dataset for both signalized intersection VA and unsignalized intersection MA are used in the study, with a time duration of 3200 s. The VT-micro energy consumption and emissions model was applied to calculate and comprehensively analyze the vehicle flow, fuel consumption, and emissions. Additionally, the study compared the fuel consumption and emissions for different driving scenarios, including straight through, left turn, right turn, and U-turn, within a single traffic signal cycle. The results revealed that at signalized intersections, the average fuel consumption per vehicle was 26.54 L/100 km, NO_x emissions were 68.76 g/100 km, and CO₂ emissions were 61.07 g/100 km. In contrast, at unsignalized intersections, the average fuel consumption per vehicle was 46.88 L/100 km, NO_x emissions were 149.26 g/100 km, and CO₂ emissions were 107.16 g/100 km. The study indicated that for traffic volumes between 50 and 103 vehicles per 100 s, signalized intersections demonstrated better fuel consumption and emission performance than unsignalized intersections. Signalized intersections could accommodate larger traffic volumes and provide enhanced traffic safety. In conclusion, the findings of this study are important for urban traffic planning and environmental policies. They provide a scientific basis for reducing fuel consumption and emissions and improving road traffic efficiency. Due to the advantages of signalized intersections in terms of energy consumption and emissions, future urban traffic planning should consider more signal light controls to achieve energy savings, emission reduction, and improved traffic operation efficiency.

1. Introduction

The seriousness of environmental pollution has attracted global attention [1,2,3]. Climate change, air pollution, and water pollution have become increasingly severe issues [4], forcing governments worldwide to take proactive measures to promote energy conservation and emission reduction. Particularly, petroleum consumption and tailpipe emissions from motor vehicles are significant contributors to environmental pollution [5,6], profoundly impacting air quality and human health [7]. Furthermore, with the continuous growth in the number of motor vehicles, their energy consumption has also dramatically increased [8]. The energy consumption issue of motor vehicles has become a critical challenge that is difficult to address in urban road traffic systems [9]. To tackle this challenge, many countries and cities have been formulating and implementing various policies and measures to reduce the energy consumption and tailpipe emissions of motor vehicles, driving forward the process of energy conservation and emission reduction.

In urban road traffic networks, intersections are one of the key research areas for vehicle energy consumption and emissions. Intersections, due to their unique geographical locations and traffic organization characteristics, often lead to frequent acceleration, deceleration, and frequent stops of vehicles [10]. These operational features make the issue of vehicle energy consumption and emissions at intersections more prominent, becoming a crucial focus for optimizing traffic operations and implementing environmental protection measures. In previous research, many scholars have focused on the energy consumption and emissions issues of motor vehicles at intersections. However, for the study of signalized intersection, most studies have been concentrated on topics such as eco-driving strategies at signalized intersections [11,12,13] and the optimization of traffic signal timing [14,15]. For instance, Ren et al. [16] developed the CVIS intersection control model based on traffic efficiency and vehicle energy consumption. Li et al. [17] proposed an optimized speed-trajectory control method based on braking energy recovery efficiency to reduce energy consumption caused by frequent braking at signalized intersections. Wen et al. [18] proposed an optimal control method for connected and automated vehicles to pass through signalized intersections, aiming to reduce energy consumption in mixed traffic scenarios with conventional vehicles and connected and automated vehicles. Yan et al. [19] optimized traffic signal-control strategies according to the spatio-temporal characteristics of different traffic scenarios to alleviate urban traffic congestion. In addition, the ecological traffic concept is integrated into the traffic signal control, aiming at reducing traffic exhaust emissions. Boukerche et al. [20] explored traffic signal control through ecological driving technology to improve vehicle fuel efficiency. As for the study of unsignalized intersections, the current research focuses on the decision-making based on game theory [21,22,23] and the establishment of interactive behavior model between vehicles and pedestrians [24,25] without studying vehicle energy consumption and emissions. The original intention of setting traffic lights at intersections is to improve traffic efficiency and ensure traffic safety more effectively. Nevertheless, there have been no studies to explore whether the introduction of traffic lights under similar traffic flow conditions will have a positive impact on environmental protection, and there is a lack of comparative studies on energy consumption and emissions between signalized and unsignalized intersections, as shown in

Table 1. Previous research and research gaps.

	Signalized Intersection	Unsignalized Intersection
Previous research	Eco-driving strategies [11]	Decision-making based on game theory [21,22]
	Optimization of traffic signal timing [14]	Establishment of interactive behavior model between vehicle and pedestrian [24,25]
Research gaps	There have been no studies to explore whether the introduction of traffic lights under similar traffic flow conditions will have a positive impact on environmental protection, and there is a lack of comparative studies on energy consumption and emissions between signalized and unsignalized intersections.

.

Therefore, studying the energy consumption and emission characteristics of vehicles at intersections and exploring the impact of traffic signal lights on vehicle energy consumption and emissions can help identify solutions to reduce pollution and enhance traffic operational efficiency. This study aims to conduct a comparative analysis of energy consumption and emissions between signalized and unsignalized intersections, which can further aid in understanding the influence of traffic signals on vehicle energy consumption and emissions. The findings of these research efforts hold significant implications for urban traffic planning and environmental policies, providing a scientific basis for reducing traffic energy consumption, mitigating environmental pollution, and enhancing road traffic efficiency.

In the extraction of information such as vehicle speed and position from real vehicle data, multiple algorithms will be involved while simultaneously introducing errors and error accumulation. In data-driven analytical research, this aspect becomes particularly critical, and previous studies extensively explore this domain. Xia et al. [26] proposed a comprehensive automated driving system data acquisition and analytics platform, providing support for trajectory extraction and evaluation for connected automated vehicle. By leveraging deep-learning-based object detection and multi-object tracking algorithms, this platform achieved cooperative perception fusion and fine trajectory reconstruction through post-processing methods. Additionally, the pivotal role of localization was underscored by integrating inertial dead reckoning with 3D LiDAR-based map matching, showcasing the method’s robustness against environmental uncertainties and sensor noise, which is verified through urban experiments [27]. The pursuit of precise vehicle state estimation continued with a novel consensus-based algorithm introduced by Xia et al. [28], enhancing observability and accuracy for sideslip angle and heading estimation. This algorithm synthesized vehicle kinematics and dynamics, further heightening accuracy and observability in dynamic scenarios. Liu et al. [29] introduced a kinematic-model-based method for vehicle slip-angle estimation, seamlessly merging data from GNSS and IMU sources. Collectively, these studies advanced our comprehension of data processing, localization, and dynamic control. The INTERACTION real vehicle dataset utilized in this study has undergone video stabilization, alignment, detection, data association, tracking, and smoothing processes [30], ensuring data validity and reliability, providing a robust foundation for this analysis.

In summary, real-world vehicle data from the INTERACTION dataset is used, specifically focusing on data from two types of intersections: signalized intersection (referred to as VA) and unsignalized intersection (referred to as MA). The Virginia tech microscopic (VT–Micro) energy consumption and emission model is employed to perform calculations, allowing for a comprehensive analysis of vehicle flow, fuel consumption, and emissions. Furthermore, a comparison of fuel consumption and emissions is conducted for different operational scenarios within a single cycle, including straight-through, left-turn, right-turn, and U-turn movements.

The paper is organized as follows: Section 2 introduces VT–Micro energy and emission model and INTERSECTION dataset. In Section 3, the total operation characteristics and fuel consumption between signalized and unsignalized intersections are analyzed. In Section 4, one cycle operation characteristics and fuel consumption between signalized and unsignalized intersections are analyzed. Finally, the conclusions of this paper are drawn in Section 5, as shown in Figure 1.

2. Model and Data

2.1. VT–Micro Energy and Emission Model

The VT–Micro energy and emission model is a microscopic dynamic emission and fuel consumption calculation model obtained through the application of linear regression [31]. It was proposed by Kyoungho Ahn and Hesham Rakha et al., based on fuel consumption data obtained from nine small cars using chassis dynamometers in a laboratory setting [32]. This model considers vehicle speed and acceleration as variable parameters to calculate emissions and energy consumption for each vehicle. Subsequently, the scope of data collection was expanded from nine small passenger cars to include a total of 60 small cars and trucks [33]. Cluster analysis techniques were employed to obtain the operating states and energy consumption data for these vehicles, resulting in refinements to the calculation formula of VT–Micro energy and emission model. While the theoretical energy consumption and emission values derived from the VT–Micro model may exhibit some discrepancies from actual values, the trends in energy consumption it reflects are generally consistent with empirical observations. Hence, the VT–Micro energy and emission model is used to conduct a comparative analysis of energy consumption and emissions at signalized and unsignalized intersections.

The basic form of the VT–Micro model is shown as Equation (1).

$$\ln \left(MO{E}_e\right)=\{\begin{array}{c}{\displaystyle {\displaystyle \sum }_{i=0}^3}{\displaystyle {\displaystyle \sum }_{j=0}^3}(K_{i,j}^e\times V_n^i\times a_n^j),a\ge 0\\ {\displaystyle {\displaystyle \sum }_{i=0}^3}{\displaystyle {\displaystyle \sum }_{j=0}^3}(K_{i,j}^{\prime e}\times V_n^i\times a_n^j),a<0\end{array}$$

(1)

where, MOE_e is the instantaneous energy consumption rate for the nth car. The unit is mL/s. v_n is the instantaneous speed of the nth car and the unit is km/h. a_n is the instantaneous acceleration of the nth car, and the unit is km/h/s. i and j are the power of velocity and acceleration, respectively. $K_{i,j}^e$ is the correlation regression coefficient when the acceleration is positive value and $K_{i,j}^{\prime e}$ is when the acceleration is negative value.

The total fuel consumption and emissions of the nth vehicle in the time T period are respectively recorded as E_fuel, E_NOx and E_CO2, and the specific calculation formula is as shown in Equation (2):

$$E_e=\frac{{\displaystyle {{\displaystyle \sum }}_{t=0}^T}MO{E}_e}{L_n}$$

(2)

where, L_n is the distance traveled by the nth car in the time T period. E_e is the total fuel consumption or emission, in which the unit of fuel consumption E_fuel is L/100 km, the unit of NO_x emission E_NOx is g/100 km, and the unit of CO₂ emission E_CO2 is kg/100 km.

In 2014, Li et al. [34], based on the VT–micro model studied by Ahn et al., used the vehicle trajectory data in NGSIM (the main source of the data was the measured vehicle data of Los Angeles Highway 101 on 15 June 2005, 7:50–8:35 a.m.) to further fit the parameters of the model. The correlation regression coefficients corresponding to fuel consumption, NO_x emissions, and CO₂ emissions were obtained, respectively, as shown in

Table 2. Correlation coefficient of fuel consumption MOE_e (The unit is L/s).

${\mathit{K}}_{\mathit{i},\mathit{j}}^{\mathit{e}}/ {\mathit{K}}_{\mathit{i},\mathit{j}}^{\prime \mathit{e}}$		j = 0	j = 1	j = 2	j = 3
a ≥ 0	i = 0	−7.735	0.2295	−5.61 × 10−3	9.773 × 10−5
	i = 1	0.02799	0.0068	−7.722 × 10−4	8.38 × 10−6
	i = 2	−2.228 × 10−4	−4.402 × 10−5	7.90 × 10−7	8.17 × 10−7
	i = 3	1.09 × 10−6	4.80 × 10−8	3.27 × 10−8	−7.79 × 10−9
a < 0	i = 0	−7.735	−0.01799	−4.27 × 10−3	1.8829 × 10−4
	i = 1	0.02804	7.72 × 10−3	8.375 × 10−4	3.387 × 10−5
	i = 2	−2.199 × 10−4	−5.219 × 10−5	−7.44 × 10−6	2.77 × 10−7
	i = 3	1.08 × 10−6	2.47 × 10−7	4.87 × 10−8	3.79 × 10−10

,

Table 3. Correlation coefficient of NO_x emission MOE_e (The unit is mg/s).

${\mathit{K}}_{\mathit{i},\mathit{j}}^{\mathit{e}}/ {\mathit{K}}_{\mathit{i},\mathit{j}}^{\prime \mathit{e}}$		j = 0	j = 1	j = 2	j = 3
a ≥ 0	i = 0	−1.080	0.2369	1.47 × 10−3	−7.822 × 10−5
	i = 1	1.791 × 10−2	4.053 × 10−2	−3.75 × 10−3	1.052 × 10−4
	i = 2	2.412 × 10−4	−4.078 × 10−4	−1.284 × 10−5	1.52 × 10−6
	i = 3	−1.06 × 10−6	9.418 × 10−7	1.86 × 10−7	4.419 × 10−9
a < 0	i = 0	−1.080	0.2085	2.193 × 10−2	8.816 × 10−4
	i = 1	2.111 × 10−2	1.067 × 10−2	6.550 × 10−3	6.265 × 10−4
	i = 2	1.63 × 10−4	−3.23 × 10−5	−9.429 × 10−5	−1.008 × 10−5
	i = 3	−5.832 × 10−7	1.83 × 10−7	4.473 × 10−7	4.573 × 10−8

and

Table 4. Correlation coefficient of CO₂ emission MOE_e (The unit is mg/s) [35].

${\mathit{K}}_{\mathit{i},\mathit{j}}^{\mathit{e}}/ {\mathit{K}}_{\mathit{i},\mathit{j}}^{\prime \mathit{e}}$		j = 0	j = 1	j = 2	j = 3
a ≥ 0	i = 0	6.916	0.217	2.354 × 10−4	−3.639 × 10−4
	i = 1	0.02754	0.968 × 10−2	−0.175 × 10−2	8.35 × 10−5
	i = 2	−2.070 × 10−4	−1.0138 × 10−4	1.966 × 10−5	−1.02 × 10−6
	i = 3	9.80 × 10−7	3.66 × 10−7	−1.08 × 10−7	8.50 × 10−9
a < 0	i = 0	6.915	−0.032	−9.17 × 10−3	−2.886 × 10−4
	i = 1	0.0284	8.53 × 10−3	1.15 × 10−3	−3.06 × 10−6
	i = 2	−2.266 × 10−4	−6.594 × 10−5	−1.289 × 10−5	−2.68 × 10−7
	i = 3	1.11 × 10−6	3.20 × 10−7	7.56 × 10−8	2.95 × 10−9

.

Using Equations (1) and (2), combined with the real-time speed and acceleration of the vehicle, the real-time fuel consumption and emission value of each vehicle can be calculated.

2.2. Real Car Data Introduction

2.2.1. Data Introduction

The INTERSECTION dataset is a new dataset of naturalistic vehicle trajectories recorded at German intersections [36]. In the dataset, intersection locations with and without traffic signals are screened. By comparing the road geometries and traffic flow, two locations are selected for the study: an unsignalized intersection named “USA_Intersection_MA” (hereafter referred to as MA) and a signalized intersection named “TC_Intersection_VA” (hereafter referred to as VA). The animation length and number of vehicles recorded in the two scenes in the dataset are shown in

Table 5. INTERACTION dataset.

Scenarios	Locations	Folders	Video Length (min)	Number of Vehicles
Unsignalized intersection	USA_Intersection_MA	DR_USA_Intersection_MA	107.37	2982
Signalized intersection	TC_Intersection_VA	TC_Intersection_VA	60	3775

. The maps are shown in Figure 2.

The signalized intersection has a main road with five lanes and six lanes, and a minor road with three lanes. In contrast, the unsignalized intersection has a main road with five lanes, which subsequently reduces to four lanes in the northbound direction, and a minor road with three lanes. The road geometries are generally similar, but the signalized intersection has a higher traffic volume. The signalized intersection was captured by a fixed camera within the recording range as indicated by the box outline in Figure 2. To facilitate a better and consistent comparison of the characteristics between signalized and unsignalized intersections, the internal area of the unsignalized intersection captured by a drone is also selected. The area does not include the road section leading to the intersection before entering it.

The file “DR_USA_Intersection_MA” consists of 22 Excel data sheets, with each sheet recording approximately 300,800 milliseconds, equivalent to 5 min. However, the file “TC_BGR_Intersection_VA” contains four Excel data sheets, each recording 899,900 milliseconds, which corresponds to 15 min. Each Excel data sheet primarily includes information such as track ID, vehicle types, vehicle sizes, current positions, current velocities, and orientations, as shown in

Table 6. Field names and meanings of dataset.

No.	Field	Meaning	Explain
1	track_id	The ID of the agent.
2	frame_id	The frames when the agent appears in the video.
3	timestamp_ms	The time when the agent appears in the video.	The unit is ms.
4	agent_type	The types of tracked agents.	For example, it can be a car, a truck and so on.
5	x	The x position of the agent at each frame.	The unit is m.
6	y	The y position of the agent at each frame.	The unit is m.
7	vx	The velocity of the agent along x-direction at each frame.	The unit is m/s.
8	vy	The velocity of the agent along y-direction at each frame.	The unit is m/s.
9	psi_rad	The yaw angle of the agent at each frame.	The unit is rad.
10	length	The length of the agent.	The unit is m.
11	width	The width of the agent.	The unit is m.

. A partial data display is presented in

Table 7. Partial data display.

Track_id	Frame_id	Timestamp_ms	Agent_Type	x	y	vx	vy	psi_rad	Length	Width
29	281	28,100	car	1009.283	1019.697	0.09	−3.198	−1.543	5.04	2.14
29	282	28,200	car	1009.293	1019.377	0.091	−3.169	−1.542	5.04	2.14
29	283	28,300	car	1009.302	1019.06	0.091	−3.112	−1.542	5.04	2.14
29	284	28,400	car	1009.312	1018.749	0.09	−3.027	−1.541	5.04	2.14
29	285	28,500	car	1009.322	1018.446	0.088	−2.918	−1.541	5.04	2.14
29	286	28,600	car	1009.332	1018.155	0.085	−2.787	−1.54	5.04	2.14
29	287	28,700	car	1009.341	1017.876	0.081	−2.635	−1.54	5.04	2.14
29	288	28,800	car	1009.349	1017.612	0.077	−2.467	−1.54	5.04	2.14
29	289	28,900	car	1009.357	1017.366	0.071	−2.285	−1.54	5.04	2.14
29	290	29,000	car	1009.364	1017.137	0.065	−2.093	−1.54	5.04	2.14
29	291	29,100	car	1009.37	1016.928	0.058	−1.894	−1.54	5.04	2.14

.

At the VA intersection, a complete traffic light cycle lasts for 100 s, as shown in Figure 3. The table data are processed to retain complete traffic light cycles, with each table containing eight complete cycles, resulting in a total of 32 complete cycles, lasting for 3200 s. The traffic flow for each 800 s is 793, 858, 838, and 826 vehicles, respectively, totaling 3316 vehicles. On average, 103 vehicles pass through the intersection in each cycle of 100 s. To make a comparison with the unsignalized intersection while ensuring the animation length remains equal and traffic flow is as similar as possible, 300 s are selected from each table. The 11 tables with the highest traffic flow are chosen, and the final table is truncated to 200 s to match the duration of the VA intersection. For the MA intersection, the traffic flow for every 200 s is 181, 161, 155, 152, 147, 145, 142, 142, 141, and 141 vehicles. The last table with 200 s of traffic flow has 95 vehicles, resulting in a total of 1602 vehicles passing through the unsignalized intersection over 3200 s. On average, 50 vehicles pass through the intersection in each 100 s.

2.2.2. Data Preprocessing

Zhan et al. [30] processed the collected video through three steps: video stabilization and alignment, detection, data association, and tracking and smoothing. The bounding boxes are very accurate, and the paths are smooth. The processed data are shown in Table 7.

The speed in the original table is v_x and v_y; that is, the partial speed of the car along the x direction and the y direction. Equation (3) is used to calculate the instantaneous speed of the vehicle at a certain moment, and the unit is km/h.

$$v=\sqrt{v_x{}^2+v_y{}^2}\times 3.6$$

(3)

Equation (4) is used to obtain the instantaneous acceleration of the vehicle at a certain moment and the unit is km/h/s. Outliers with accelerations greater than or less than 12.6 are discarded.

$$a_i=\frac{v_{i+1}-v_i}{fram{e}_{i+1}-fram{e}_i}/0.1$$

(4)

Equation (5) is used to obtain the total distance traveled by the vehicle at all recorded moments, and the unit is mm.

$$L={\displaystyle \sum }_{i=1}^T\sqrt{{(x_i-x_{i-1})}^2+{(y_i-y_{i-1})}^2}\times 1000$$

(5)

3. Comparative Analysis of Total Operation Characteristics and Fuel Consumption between Signalized and Unsignalized Intersections

3.1. Analysis of Traffic Flow Characteristics

3.1.1. Analysis of Traffic Flow Characteristics at Signalized Intersection

By utilizing Python for data analysis and visualization, it was observed that vehicle movement at intersections can experience delays due to signal light restrictions, resulting in some temporal gaps. For instance, at 320,600 milliseconds, a vehicle with carID 1411 was positioned in the left-turn lane. There were no vehicles obstructing its left-turn path in the adjacent oncoming lanes or intersecting straight lanes, as illustrated in Figure 4a. However, data indicated that carID 1411 could only begin its left-turn maneuver at ts 335,200 milliseconds, as depicted in Figure 4b. During the approximately 14.6-s interval from 320,600 ms to 335,200 ms, carID 1411 and other vehicles waiting to make a left-turn had to remain stationary, waiting for the left-turn signal to turn green. This phenomenon illustrates that due to signal light restrictions, vehicles that could safely make left turns are unable to do so until the left-turn signal is green, resulting in a wasted time of about 15 s. A similar observation is demonstrated in Figure 5. At ts 359,000 milliseconds, carID 1478 had no other vehicles obstructing its left-turn path, yet it remained stationary until ts 390,700 milliseconds when it finally executed the left-turn maneuver, incurring a delay of 31.7 s.

As shown in Figure 6, at 55,400 milliseconds, there were no impediments for northbound-to-southbound vehicles; however, they only began to pass through the intersection at 60,400 milliseconds, wasting 5 s of passage capacity.

3.1.2. Analysis of Traffic Flow Characteristics at Unsignalized Intersection

Using Python in conjunction with the 11 tables from the MA intersection, it is possible to visualize the traffic flow at the unsignalized intersection. Analyzing the traffic flow animation reveals that the unsignalized intersection experiences brief congestion. For example, as shown in Figure 7, at 11,500 milliseconds, Vehicle 25 starts queuing and waiting. At this moment, there is heavy traffic in all four directions. It continues to wait until 41,500 milliseconds before vehicles start moving into the intersection. This waiting time lasts for 30 s, and the brief congestion results in relatively longer waiting times compared to when signal lights are present.

3.2. Analysis of Energy and Consumption Characteristics

3.2.1. Analysis of Energy and Consumption Characteristics at Signalized Intersection

Using the VT–Micro energy and emission model Equations (1) and (2), the total fuel consumption for the four files at the signalized intersection is calculated to be 21,012.19 L/100 km, 22,594.39 L/100 km, 22,001.61 L/100 km, and 22,360.49 L/100 km, as shown in Figure 8. The average fuel consumption for each cycle of 100 s is 2749.02 L/100 km, and the average fuel consumption per vehicle is 26.54 L/100 km.

The total NO_x emissions for each 800 s at the signalized intersection are calculated to be 56,819.88 g/100 km, 57,136.75 g/100 km, 55,396.08 g/100 km, and 58,384.32 g/100 km, as shown in Figure 8. The average NO_x emissions for each cycle of 100 s is 7116.78 g/100 km, and the average NO_x emissions per vehicle is 68.76 g/100 km.

The total CO₂ emissions for each 800 s at the signalized intersection are calculated to be 48,322.76 kg/100 km, 51,994.83 kg/100 km, 50,647.47 kg/100 km, and 51,475.77 kg/100 km, as shown in Figure 8. The average CO₂ emissions for each cycle of 100 s is 6326.28 kg/100 km, and the average CO₂ emissions per vehicle is 61.07 kg/100 km.

Using the VA003 file as an example, the fuel consumption and emissions for each vehicle are shown in Figure 9. Vehicles passing through the signalized intersection generate varying levels of fuel consumption. Due to the restrictions imposed by the traffic signals, some vehicles can pass through the intersection without the need for speed changes, resulting in lower fuel consumption and emissions, as illustrated in Figure 10 for carID 3029. However, vehicles that require speed adjustments while passing through the intersection generate higher fuel consumption and emissions, as shown in Figure 10 for carID 2795. This diversity in how vehicles pass through the intersection leads to significant differences in fuel consumption when vehicles traverse the signalized intersection.

When a large number of vehicles gather at a signalized intersection, it can lead to road congestion, as illustrated in Figure 11. Road congestion inevitably causes vehicles to experience abrupt accelerations and decelerations, resulting in the highest fuel consumption. Therefore, when congestion occurs at a signalized intersection, fuel consumption rapidly increases.

3.2.2. Analysis of Energy and Consumption Characteristics at Unsignalized Intersection

Using the VT–Micro energy and emission model Formulas 1 and 2, the total fuel consumption for the 11 files at the unsignalized intersection is calculated to be 7182.28 L/100 km, 8800.01 L/100 km, 8144.10 L/100 km, 6984.76 L/100 km, 6242.37 L/100 km, 6366.74 L/100 km, 6706.50 L/100 km, 6451.03 L/100 km, 6848.45 L/100 km, 7381.90 L/100 km, and 4096.87 L/100 km (for the 200 s interval), as shown in Figure 12. The average fuel consumption for each cycle of 100 s is 2341.01 L/100 km, and the average fuel consumption per vehicle is 46.88 L/100 km.

The total NO_x emissions for each 300 s at the unsignalized intersection are calculated to be 23,913.26 g/100 km, 26,522.91 g/100 km, 25,807.78 g/100 km, 22,916.62 g/100 km, 20,033.96 g/100 km, 21,032.22 g/100 km, 21,262.30 g/100 km, 21,135.09 g/100 km, 20,398.04 g/100 km, 23,325.24 g/100 km, and 12,858.19 g/100 km (for the 200 s interval), as shown in Figure 12. The average NO_x emissions for each cycle of 100 s is 7443.48 g/100 km, and the average NO_x emissions per vehicle is 149.26 g/100 km.

The total CO₂ emissions for each 300 s at the unsignalized intersection are calculated to be 16,329.96 kg/100 km, 20,114.82 kg/100 km, 18,476.88 kg/100 km, 15,909.17 kg/100 km, 14,401.01 kg/100 km, 14,560.62 kg/100 km, 15,380.11 kg/100 km, 14,726.63 kg/100 km, 15,734.95 kg/100 km, 16,771.25 kg/100 km, and 9418.75 kg/100 km (for the 200 s interval), as shown in Figure 12. The average CO₂ emissions for each cycle of 100 s is 5349.50 kg/100 km, and the average CO₂ emissions per vehicle is 107.16 kg/100 km.

Using the MA20 file as an example, the fuel consumption and emissions for each vehicle are depicted in Figure 13. Vehicles passing through the unsignalized intersection exhibit relatively small variations in fuel consumption. Without the restrictions imposed by traffic signals, there are fewer instances where vehicles experience significant fuel consumption due to prolonged stops. Instead, most vehicles either accelerate directly through the intersection, as seen in Figure 14 for carID 56, or decelerate and pass through after other vehicles, as illustrated in Figure 14 for carID 60 and carID 66. As a result, the fuel consumption generated by vehicles passing through the unsignalized intersection appears to be quite similar.

When vehicles pass through an unsignalized intersection, they encounter more instances of rapid speed changes. Without signal guidance, vehicles rely more on the drivers’ own driving behavior when traversing the intersection. When vehicles meet at the intersection, one may accelerate through, while the other urgently brakes and slows down. These abrupt speed changes result in higher fuel consumption, leading to a higher average fuel consumption for vehicles passing through unsignalized intersections compared to signalized ones. As shown in Figure 15, when carID 56 encounters carID 62, carID 56 accelerates through the intersection, while carID 62 urgently decelerates and waits for carID 56 to pass through the intersection before accelerating again.

When a large number of vehicles gather at an unsignalized intersection, it can lead to road congestion, as shown in Figure 16. Road congestion inevitably causes vehicles to experience abrupt accelerations and decelerations, resulting in the highest fuel consumption. Therefore, when congestion occurs at an unsignalized intersection, fuel consumption rapidly increases.

3.2.3. Comparison Analysis between Signalized and Unsignalized Intersections

Based on Figure 8 and Figure 12, along with Equations (1) and (2), it is evident that vehicle fuel consumption, NO_x emissions, and CO₂ emissions are all influenced by the amount of traffic flow. The total fuel consumption, NO_x emissions, and CO₂ emissions for vehicles at signalized and unsignalized intersections, along with the averages per cycle of 100 s and per vehicle, are shown in Figure 17.

At signalized intersections, the traffic flow is larger than at unsignalized intersections, sometimes even twice as much. The average traffic flow within one cycle of 100 s is also twice as much at signalized intersections compared to unsignalized ones. However, surprisingly, the average NO_x emissions within one cycle at unsignalized intersections even exceed those at signalized intersections. Additionally, the average fuel consumption, NO_x emissions, and CO₂ emissions per vehicle are higher at unsignalized intersections compared to signalized ones. Consequently, it can be concluded that unsignalized intersections have lower traffic capacity, higher fuel consumption, and higher emissions compared to signalized intersections. Signalized intersections not only ensure safety but also save energy and produce lower emissions compared to unsignalized ones.

4. Comparative Analysis of One Cycle Operation Characteristics and Fuel Consumption between Signalized and Unsignalized Intersections

One traffic signal cycle (100 s) at the VA intersection was selected. There were 108 vehicles passing through, including 66 vehicles driving straight, 18 vehicles making left turns, 21 vehicles making right turns, and three vehicles making U-turns. At the MA intersection, within 100 s, there were 61 vehicles passing through, including 37 vehicles driving straight, 20 vehicles making left turns, four vehicles making right turns, and zero vehicles making U-turns.

At the VA intersection, during one traffic signal cycle, the average speed of 66 vehicles driving straight is 33.74 km/h, with an average fuel consumption of 22.35 L/100 km, NO_x emissions of 77.21 g/100 km, and CO₂ emissions of 50.93 kg/100 km. The average speed of 18 vehicles making left turns is 16.75 km/h, with an average fuel consumption of 42.33 L/100 km, NO_x emissions of 66.14 g/100 km, and CO₂ emissions of 98.44 kg/100 km. The average speed of 21 vehicles making right turns is 13.85 km/h, with an average fuel consumption of 23.13 L/100 km, NO_x emissions of 33.75 g/100 km, and CO₂ emissions of 53.96 kg/100 km. The average speed of 3 vehicles making U-turns is 2.34 km/h, with an average fuel consumption of 114.38 L/100 km, NO_x emissions of 140.02 g/100 km, and CO₂ emissions of 265.46 kg/100 km.

At the MA intersection, during one traffic signal cycle, the average speed of 37 vehicles driving straight is 12.86 km/h, with an average fuel consumption of 47.39 L/100 km, NO_x emissions of 153.97 g/100 km, and CO₂ emissions of 107.48 kg/100 km. The average speed of 20 vehicles making left turns is 12.25 km/h, with an average fuel consumption of 44.91 L/100 km, NO_x emissions of 131.34 g/100 km, and CO₂ emissions of 102.95 kg/100 km. The average speed of four vehicles making right turns is 10.89 km/h, with an average fuel consumption of 43.55 L/100 km, NO_x emissions of 103.98 g/100 km, and CO₂ emissions of 102.41 kg/100 km.

4.1. Analysis of Operation Characteristics at Signalized and Unsignalized Intersections

The average velocity of driving straight, turning left, turning right, and turning around in a period of signalized and unsignalized intersection is shown in Figure 18.

As shown in Figure 18, at signalized intersections, the average speed of vehicles driving straight is the highest, followed by vehicles making left turns. The speeds of vehicles making right turns and U-turns are relatively similar, but the U-turn speed is very low. When the green light for driving straight is on, there is usually no safety hazard, so vehicles accelerate through the intersection, resulting in the highest speed. For left turns, vehicles need to enter the waiting area before accelerating through the intersection, while right turns are not restricted by traffic lights, but they must yield to pedestrians, leading to relatively lower speeds. U-turning vehicles typically need to yield to vehicles driving straight and may have conflicts with vehicles making right turns at large turning radii, resulting in a stop-and-go situation and, therefore, lower average speeds. However, at unsignalized intersections, the speeds of vehicles driving straight, making left turns, and making right turns are similar. Without the restriction of traffic lights, these vehicles usually proceed while observing and waiting for no oncoming traffic before quickly passing through the intersection, following the “first come, first served” rule and yielding to vehicles making turns, which results in similar speeds among these vehicles.

4.1.1. Analysis of Operation Characteristics at Signalized Intersection

At signalized intersections, the speed of 66 vehicles driving straight through the intersection is shown in Figure 19. On average, it takes 5.12 s for vehicles to pass through the intersection. Most vehicles accelerate through the intersection, with a few even reaching speeds exceeding 50 km/h, as shown in Figure 20. In this case, the green light is about to end, and carID 551 only takes 2.6 s to pass through the intersection. There are also some vehicles that decelerate to almost zero speed before accelerating through the intersection. This is because they encountered U-turning vehicles ahead, causing a brief congestion, and they had to reduce their speed and wait, as shown in Figure 21. The average vehicle speed is 33.74 km/h.

At the signalized intersection, the speed of 18 vehicles making left turns through the intersection is shown in Figure 22. On average, it takes 13.4 s for vehicles to make a left turn through the intersection. Most vehicles slow down as they enter the intersection and then accelerate to pass through it. For left turns from the main road, there is a waiting area, so vehicles must come to a stop and wait until there are no oncoming vehicles before proceeding, as shown in Figure 23. For example, carID163 had to wait for 40 s before making the left turn. However, left turns from the minor road directly proceed when the left-turn signal is green, resulting in acceleration through the intersection. The average vehicle speed for left turns is 16.75 km/h.

At the signalized intersection, the speed of 21 vehicles making right turns through the intersection is shown in Figure 24.

On average, it takes 4.82 s for vehicles to make a right turn through the intersection. Most vehicles slow down as they enter the intersection and then accelerate to pass through it. Right turns usually involve conflicts with pedestrians, and the right-turn signal is typically steady green. When encountering pedestrians, vehicles slow down and wait until the pedestrians have crossed before accelerating through the intersection. The average vehicle speed for right turns is 13.85 km/h. Out of the 21 vehicles making right turns, only carID 362 had a speed exceeding 25 km/h.

4.1.2. Analysis of Operation Characteristics at Unsignalized Intersection

At the unsignalized intersection, the average time needed to pass through the intersection is 16.84 s for driving straight, 15.36 s for making left turns, and 14.27 s for making right turns. The speeds of vehicles passing through the intersection for driving straight, making left turns, and making right turns are shown in Figure 25, Figure 26 and Figure 27, respectively.

The average speeds for driving straight, making left turns, and making right turns are similar, and the time needed is also similar. The trend for all three movements is to slow down upon entering the intersection and then accelerate to pass through it.

4.2. Analysis of Fuel Consumption and Emission Characteristics at Signalized and Unsignalized Intersections

Using left turns and right turns at the signalized intersection as examples, the average speed, total fuel consumption, and total emission values for each vehicle are shown in Figure 28 and Figure 29. Combining the analysis from the previous speed graphs (Figure 22 and Figure 24), it can be observed that vehicles with slower speeds have higher fuel consumption and emissions. Vehicles with similar driving conditions generally exhibit similar fuel consumption and emission levels.

Using straight-through and left turns at the unsignalized intersection as examples, the average speed, total fuel consumption, and total emission values for each vehicle within a 100 s cycle are shown in Figure 30 and Figure 31. Vehicles with similar driving conditions generally exhibit similar fuel consumption and emission levels. However, carID22 and carID29 show significantly higher NO_x emissions due to their higher acceleration rates.

4.3. Comparison Analysis between Signalized and Unsignalized Intersections

The average speed, average fuel consumption, and emission levels of vehicles within a single cycle at signalized and unsignalized intersections are shown in Figure 32. The highest fuel consumption and CO₂ emissions occur during U-turns. At signalized intersections, left turns have the highest fuel consumption and CO₂ emissions, while right turns and straight-through movements are similar in this regard. However, for NO_x emissions, straight-through movements have the highest levels compared to left and right turns. At unsignalized intersections, the fuel consumption and emissions are generally similar across all movement types. In terms of NO_x emissions, straight-through movements have the highest levels compared to left and right turns, possibly due to their longer travel time. It is inferred that the higher emissions during left turns at signalized intersections are caused by the secondary deceleration and stop required in the left-turn waiting zone before accelerating again, leading to increased emissions.

Comparing vehicles at signalized and unsignalized intersections within one cycle, regardless of whether they are driving straight, turning left, or turning right, signalized intersections have lower fuel consumption and emissions than unsignalized intersections. Additionally, vehicles spend less time passing through signalized intersections; this type of intersection provides a higher level of safety assurance. Therefore, within the scope of this study, when the average traffic volume ranges from 50 to 103 vehicles in a 100-s cycle, signalized intersections perform better in terms of fuel consumption and emissions compared to unsignalized intersections.

5. Conclusions

In the INTERACTION dataset, two signalized (VA) and unsignalized (MA) intersections with similar road shapes were selected to study vehicle operation characteristics, fuel consumption, and emissions characteristics. In the signalized intersection, a total of 3316 vehicles passed through during the captured 3200 s. In the unsignalized intersection, a total of 1602 vehicles passed through during the captured 3200 s. The transportation capacity at the signalized intersection is much greater than that at the unsignalized intersection.

At the signalized intersection, the total fuel consumption was 87,968.68 L/100 km, the total NO_x emissions were 227,737.03 g/100 km, and the total CO₂ emissions were 202,440.82 kg/100 km. At the unsignalized intersection, during the same period, the total fuel consumption was 75,205.00 L/100 km, the total NO_x emissions were 239,205.59 g/100 km, and the total CO₂ emissions were 171,824.15 kg/100 km. When the traffic flow at the unsignalized intersection is half that of the signalized intersection, the total NO_x emissions at the unsignalized intersection even exceed that of the signalized intersection. In more detail, at the signalized intersection, the average fuel consumption per vehicle was 26.54 L/100 km, with an NO_x emission of 68.76 g/100 km and a CO₂ emission of 61.07 g/100 km. Conversely, at the unsignalized intersection, the average fuel consumption per vehicle was 46.88 L/100 km, with an NO_x emission of 149.26 g/100 km and a CO₂ emission of 107.16 g/100 km. The average fuel consumption, NO_x emissions, and CO₂ emissions per vehicle at the unsignalized intersection are all higher than at the signalized intersection. This leads to the conclusion that the capacity, fuel consumption, and emissions at the unsignalized intersection are inferior to those at the signalized intersection. The signalized intersection ensures both safety and energy conservation while also having lower emissions compared to the unsignalized intersection.

One traffic-signal cycle (100 s) at the VA intersection and the same 100 s at the MA intersection are selected to study. There were 108 vehicles passing through at the VA intersection, and there were 61 vehicles passing through at the MA intersection within 100 s. Comparing vehicles at signalized and unsignalized intersections within one cycle, regardless of whether they are driving straight, turning left, or turning right, at signalized intersections have lower fuel consumption and emissions than at unsignalized intersections. Additionally, vehicles spend less time passing through signalized intersections, suggesting that signalized intersections not only improve road traffic efficiency but also reduce energy consumption and emissions. Therefore, within the scope of this study, when the average traffic volume ranges from 50 to 103 vehicles in a 100-s cycle, signalized intersections outperform unsignalized intersections in terms of transportation capacity, energy consumption, and emissions.

In our subsequent research, we aspire to conduct experiments for data collection and utilize algorithms to process and obtain vehicle speed and position information. We intend to explore the potential errors that may arise during the application of various algorithms, including computer vision, optical-flow estimation, object tracking, object detection, deep learning, vehicle calibration, sensor fusion, and Kalman filtering. Our goal is to analyze the impact of these errors and their cumulative effects on the final results.