Introduction
Motivation, objectives and approach
Vietnam's transportation system is currently grappling with significant challenges, particularly in major cities like Hanoi, where issues such as congestion, accidents, and chaotic traffic conditions arise from narrow roads, a growing number of vehicles, and a lack of adherence to traffic rules While various solutions have been proposed and implemented to enhance the transportation system and raise public awareness, many of these strategies demand substantial financial investments to be effective Consequently, there is a pressing need for cost-effective methods to improve the transportation situation, which motivated me to focus my thesis on this critical theme.
In developed countries, transportation planners must have a strategic vision to effectively develop transport systems, often utilizing traffic simulators to gather essential data These simulators enable policymakers to identify cost-effective solutions for traffic infrastructure development The trend of employing multi-agent systems to simulate societal behavior is increasingly common in addressing transportation challenges Following this trend, we initiated the development of the Vietnam Traffic Simulator (VTS) in 2010, guided by Assc Prof Bui The Duy, with the primary goal of enhancing the accuracy and reliability of the VTS.
The primary objective of this thesis is to complete the evaluation of the simulator, which involves both proactive and reactive strategies Drawing on foundational knowledge from research on Multi-Agent Systems (MASs), I enhanced the structure of the Vehicle Traffic Simulator (VTS) to better accommodate the unique traffic conditions in Vietnam Specifically, I collected data from various sources, introduced new functionalities, corrected simulator behaviors, and proceeded to the evaluation phase.
Outline of the thesis
This thesis includes a comprehensive literature review on traffic simulation models and the Multi-agent approach in Chapter 2 Chapter 3 highlights key features of the Vietnam Traffic Simulator, detailing recent enhancements made since the short paper presented at PRACSYS.
2010 [17] The evaluation steps will be presented in chapter 4 The last chapter is the conclusion and future research.
Literature Review
Models of traffic simulation
Traffic simulation serves multiple purposes, including identifying solutions for traffic system issues, evaluating new transportation facility designs prior to resource allocation, analyzing system safety, and training traffic management personnel Given the complexity of transportation systems, modeling can be approached in two ways, focusing on scope and time This section will introduce foundational models used in the Vietnam Traffic Simulator, categorized into three areas: time, scope, and multi-agent systems.
Simulation models of traffic can be categorized by level of detail: macroscopic
Macroscopic models provide a low-level overview of systems, such as traffic streams represented by aggregate statistics like flow rate and density In contrast, microscopic models offer a detailed depiction of both system entities and their interactions Mesoscopic models strike a balance, representing most entities in detail while simplifying their interactions Meanwhile, nanoscopic models focus on intricate aspects of human behavior, such as drivers' steering actions and perception-reaction times, to accurately simulate human performance.
The Visual Traffic Simulation System (VISSIM), developed by Thomas Fotherby, is a versatile mesoscopic simulation tool designed for various traffic systems It offers comprehensive features for designing transportation infrastructure, providing detailed insights into vehicle flow, including the quantity of cars and trucks as well as their speeds.
The system comprises four key application components: Road Network Designer, Traffic Modelling Designer, Visual Simulation, and Application Results Each of these components offers essential features that serve as valuable suggestions for the Vehicle Traffic Simulation (VTS) process.
Road Network Designer This section of the application should allow a user to quickly design simple schematic road diagrams (road networks)
- The application should be to-scale (e.g x screen pixels per metre)
- The application should start by showing a drawing panel as a blank designing area (representing a x*x m square area)
- Assume terrain is always flat (a simplification)
- Lanes are drawn on the panel in straight-line sections Each new section of the lane follows on from the previous section
- Any existing lane should be able to be extended with another identical lane next to it (space permitting)
- Existing lanes should be able to be deleted
- When the ends of three or more lane-sections overlap a junction should form
- Lanes should also be able to pass over or under other lanes Therefore there are 4 options at any point where lanes cross other lanes
- There should be buttons for: create Lane, create Road, Add Lane, Delete Lane
- Road designs should be able to be saved and loaded
A Traffic-Modelling Tool requires a valid road network to function effectively To ensure realistic simulations and meaningful outcomes, users must input accurate traffic data, which can be sourced from electronic detection devices and traffic surveys.
- A junction has inputs and outputs
- Each junction has unique input and output traffic-flow intensities
- The input traffic-intensity of one junction will be a function of the output traffic-intensities of other junctions
- A certain amount of cars will enter the system according to some kind of control element
- A car can start at any input into the system and go to any output
- All cars should eventually exit the system (No infinite loops)
Users can specify the average or exact number of cars per minute entering the system for each input This process necessitates the labeling of roads within the designed road network.
To enhance the simulation experience, it is essential to include an option for randomizing car input data with each run Without this feature, the simulation will consistently operate with identical data, resulting in the same number of cars entering at the same time.
- Traffic-flow models should be able to be saved and loaded
Visual simulation requires a valid road network to function effectively This section features animated graphics depicting scaled vehicles navigating through the system's geometry The animated traffic is generated and managed based on statistics provided by the traffic modeling tool.
- Cars obey a speed limit This is their "top speed" An example maybe between 50 and 60 kilometres/hour (31-37mph)
- Cars enter the system at top speed at positions and times according to a set traffic-model specified by the "traffic-modelling tool"
- Cars can only enter the system if they are in a valid position (Not a collision)
- Cars can only change lanes at junctions (No U-turns)
- Cars will always try to go at their top speed when possible but their speed is governed by the "car-following model" described below
Cars adjust their speed through acceleration and deceleration Acceleration is typically a constant value, such as 5 m/s², while deceleration must be realistic, ensuring that vehicles cannot come to an immediate stop.
Cars do not make independent decisions; their travel routes and lane choices are solely determined by their starting position and the statistical outcomes at the junctions they encounter.
- A Car will travel at its top speed limit unless it is within 10m of another car
- It must de-accelerate to match the other cars speed by the time there is a 3m distance
- It must never go within 1m of another car on the same lane
- Cars follow this model when pulling up to red lights, give-way signs or if there is stopped traffic ahead
- At a suitable distance before the obstruction the car will de-accelerate with a constant value to stop in time
- The project is simplified to not include overtaking
- A car will only change lane at junctions according to the junction traffic-model statistics
Vehicle behaviour at give-way junctions
- Cars on the main route are unaffected and travel as normal according to the car-following model
- Cars on the slip roads "pull up" to the give-way line to check for oncoming traffic
Cars merging from the slip-road onto the main route must do so without obstructing existing traffic This requires a sufficiently large gap in the main route traffic, adhering to the gap-acceptance model.
Vehicle behaviour model at signalled junctions:
- Signals are independent for each input lane
- Cars will "pull-up" to the stop line if the signal is red
- The signal is two-phase Go is green, stop is red
- On a green signal the car is specified an output lane (according to the traffic-model of the junction) and will travel to the output lane in a direct route
- Traffic light timing intervals will be initially split fairly between different sets Later, traffic lights can be re-programmed to be more intelligent
The traffic light's color will be represented on the screen by the corresponding stop line color in each lane Additionally, when a lane's light is green, arrows will appear at the junction, indicating the directions available for vehicles to proceed.
- For each input to the system there should be a control to increase or decrease the traffic entering at that input
Application Results Each component of the simulated traffic system should log data:
- Each input and output of the system should have a log of how many cars passed through
- Every junction should log how many cars passed through each input and output lane
- Each traffic-light junction should store the timing intervals of each light
- There should be traffic flow data for the system as a whole (Number of cars passed through per second)
- There should be a value estimating total surface area of road surface used in the current network design
*All of these features are implemented in the VTS
Time serves as a fundamental independent variable in traffic simulation models Continuous simulation models illustrate how system elements evolve continuously over time due to ongoing stimuli, while discrete simulation models depict real-world systems with abrupt state changes at specific moments Discrete models can be categorized into two types: discrete time models, which compute state changes within defined time intervals, and discrete event models, which execute calculations triggered solely by the occurrence of events.
* Intelligent Traffic light control system
In this subsection we introduce a simulator named Green Light District Simulator developed by Utrecht University (Netherland) [7] This is a system which supports the determination in duration of traffic lights
Figure 3.Traffic light simulation system
The system functions as a microscopic traffic simulation, incorporating discrete events influenced by factors like traffic density and average vehicle speeds This data is utilized to automatically recommend optimal traffic light durations.
The main components of this system are:
- Drive Lane consists of two parallel lines
- Road made by 2 Drive Lane It includes information about direction, incoming and outgoing gates which form the transportation network
- Node is the term describing cross cuts between conjuction and crossroad
- EdgeNode describes areas in which cars go in and out
- Sign describes the traffic lights These places are the points where the duration adjustment algorithm is deployed automatically
- Cars play an important role in the simulation However, due to the main target of this simulation is automatic traffic light adjustment, the model of moving vehicles is simplified
During the simulation phase, the system collects statistical data, including vehicle density and the count of incoming and outgoing vehicles, to inform the duration adjustment algorithm for traffic lights.
2.1.3 Multi-agent system for traffic simulation
Conclusion
Recent research highlights key features of various traffic simulation systems, primarily focusing on microscopic systems that intricately simulate driver behavior These systems offer users the flexibility to design customized road networks and include comprehensive reporting components By utilizing a defined "scenario," such as a road system or highway configuration, these simulations generate results in both statistical and graphical formats The statistical outputs deliver quantitative predictions of potential outcomes, while the graphical and animated results offer valuable insights into the underlying behavior of the traffic system.
In the next section, we will briefly introduce our model used in a simulator named VTS which applies the multi-agent based model.
Vietnam Traffic Simulator
Introduction to multi-agent system
In this section, we would like to introduce a few basic concepts of agents and multi-agent based systems These are relatively new concepts which attract many researchers
There are many concepts of agent given, but so far none of them has been considered as a standard concept for the agent, for example:
The term 'agent' typically describes an entity that operates continuously and independently within an environment populated by other processes and agents (Shoham, 1993).
“An agent is an entity that senses its environment and acts upon it” (Russell, 1997);
The term "agent" encompasses two distinct yet interconnected concepts: firstly, it refers to the agent's capacity for autonomous execution, and secondly, it highlights the agent's ability to engage in domain-oriented reasoning.
Intelligent agents are autonomous software entities that perform tasks on behalf of users or other programs, utilizing knowledge of the user's goals and desires to operate effectively.
An autonomous agent is a system that operates within its environment, capable of sensing and interacting with it This agent pursues its own objectives while influencing future conditions based on its observations and actions.
However, we think that this concept given by Wooldridge and Jennings [11] is the most sufficient one:
A computer system, whether hardware or software-based, possesses key characteristics that define its functionality These systems exhibit autonomy, allowing agents to operate independently without human intervention while maintaining control over their actions and internal states They also demonstrate social ability, enabling interaction with other agents and potentially humans through a specialized communication language Furthermore, these agents are reactive, capable of perceiving environmental changes and responding promptly Lastly, they showcase pro-activeness, engaging in goal-directed behavior by taking initiative rather than merely reacting to their surroundings.
For a further description of the agents, we would like to present some of its characteristics:
• Each agent in the environment has separate attributes Based on what they get from the environment, they take action based on the current status of their attributes
• Each agent has a number of rules governing behavior and the ability to make their decision [4]
• Agent capable of responding to the environmental impact by bringing out certain actions However, the agent did not just take direct action of its target
To achieve the goal the agent needs to perform a sequence of different actions The determination of the sequence of actions is decided by implementing the plan [6]
• Agent is active They have the ability to take action, work independently without influenced from external factors
• Agent is mobility They have the ability to learn, remember and make behavioral response based on their experience [4]
Those are some major foundation of agent concepts We will take a closer look to a complicated level of agent, which is Multi Agent Systems
A Multi-Agent System (MAS) is a framework characterized by the presence of multiple agents that interact within a shared environment This system comprises both the agents and the environment, facilitating mutual interactions among the agents.
Agents can engage in various interactions, including competition, conflict, cooperation, and coordination to achieve shared objectives They possess the capability to communicate with one another through a specific protocol, allowing them to send and receive messages Additionally, agents can recognize the information received by other agents, enhancing their collaborative efforts.
In MASs, each agent has a limited view; we call it the perception of the agent
Agents operate with limited information about their environment and other agents within the system, leading to actions that influence specific aspects of both the environment and the interactions with other agents.
The behavior and properties of agents in Multi-Agent Systems (MAS) can vary significantly, as they possess organizational abilities that enable them to operate within groups, adhering to shared constraints and focusing on common goals Various organizational models exist within MAS, including the hierarchical model, where decision-making authority is concentrated among top agents at each level, facilitating interactions primarily among agents at the same or adjacent levels Alternatively, the market model distinguishes roles, with some agents providing products or services while others utilize these offerings.
Traffic simulators utilize road systems, traffic lights, and signals to create a dynamic environment where each participant, referred to as an agent, navigates the network to reach specific destinations These agents possess unique attributes, including gender, age, and experience, resulting in diverse driving behaviors Interaction among traffic participants occurs within a defined range, facilitated by communication methods such as honking and using turn signals to signal lane changes.
The agent-based traffic simulation system is an effective method for modeling complex societal structures, particularly transportation systems This approach is emphasized as the primary focus of this thesis due to its ability to capture the intricacies of traffic dynamics.
ABM (Agent-Based Modeling) is one of the computer model used to simulate the action of heterogeneous entities in an autonomous environment and the interaction between them
Agent-Based Modeling (ABM) simulates the interactions between individual agents to reconstruct or predict complex phenomena, making it an essential tool for studying intricate systems The macro-level complexities cannot be fully understood through a single micro unit, highlighting the importance of micro-level principles in generating macro phenomena By developing specific plans for each agent, we can analyze the overall transportation system, identifying congestion points that emerge during experimental simulations.
To develop an Agent-Based Model (ABM), it is essential to first identify its purpose Researchers then analyze the system to determine its components and the relationships between them The model is subsequently applied to conduct if-then experiments Finally, the model's usefulness is assessed and compared with other models based on the results obtained.
The steps to build an ABM [4]:
- Identify agent: Define the type of agent entities, attributes and their behavior
- Define the environment in which the agent will "live"
- Identify ways in which the properties of the agent are updated in response to the interactions between agent-agent and agent- environment
- Add methods to control the interaction between agent-agent and agent- environment
Modeling
In this section, we describe our agent based simulation system for the traffic in Vietnam The system comprises of two main components:
- the road system and permitted travel directions in the road system,
- the agents representing the drivers of motorbikes and cars together with their vehicles in the road system
The core aspect of the system lies in how agents formulate and implement their travel plans This article will explore in detail the process of plan creation and examine how various agent profiles influence this planning.
The road system is built up from multiple road areas
Road systems consist of basic components known as Areas, each featuring entries and exits referred to as Gates Each Gate is defined by two points, connected by a road line that may represent a single road segment or a series of segments Within each Area, road segments provide crucial information regarding pavements and allowed travel directions, which agents utilize to formulate their travel plans.
Connecting areas together to build a road system
A road system can be formed by connecting multiple areas, where one area's entry must align in position and size with another area's exit For instance, in the connection depicted in Figure 2, the exit of Area1 connects to the entry of Area2 Area1 features one entry and one exit, while Area2 has one entry and two exits, resulting in a total of one entry, two exits, and two road lines within the system.
The design of road areas enables the creation of diverse road systems with flexible shapes Additionally, segmenting a road system into distinct areas enhances the efficiency of calculations involved in agent planning.
In an agent-based simulation system, a crucial element is the identification of agents In our model, traffic vehicles are represented by agents, specifically car drivers and motorbike drivers, allowing for a comprehensive analysis of vehicular interactions.
Traffic agents must execute proactive and reactive actions to effectively control their vehicle's movement These actions aim to navigate towards a target while avoiding obstacles, such as other vehicles and pavements Our simulation system categorizes these actions into two distinct types, ensuring a comprehensive approach to vehicle maneuvering.
- speed adjustment (including accelerating and braking),
- steering, which involves not only changing lanes but moving to any adjacent available space
In our simulation system, each agent makes decisions based on the current situation, yet different agents may choose distinct actions even in similar circumstances The behavior of these agents is influenced by several key attributes.
The attributes of agents are utilized to formulate plans based on specific traffic conditions Agents are categorized into groups, assuming that those within the same group share similar attributes These groups are established based on factors such as age and render, and each group is defined by a profile that outlines the attribute values relevant to that particular group.
This section outlines an agent's planning algorithm designed to determine optimal travel routes within specific traffic conditions Additionally, it explores how the agent's attributes influence its movement from one location to a target within a road network.
The control cycle for agents in our system operates by continuously calculating a feasible plan for a specified time frame until the agent reaches its target As long as the plan remains viable and has not exceeded its time limit, the agent executes the next action in the established plan.
The calculation of a plan for an agent contains three steps:
(2) Detecting possible collisions on optimal route,
(3) If there are collisions, determining alternative route to avoid collisions
The optimal route is the fastest path an agent should take to reach its target without any obstructions In our simulation system, this route is represented by a series of sampled points The agent's optimal path to its destination is calculated using a greedy algorithm, ensuring efficiency in navigation.
The distance between two continuous sampling points is represented as ∆l, while v denotes the current speed of an agent The time taken for the agent to move from one sampling point to the next can be calculated based on these parameters.
Because an agent can only plan for a certain amount of time ahead, the number of sampling points on the planned ideal route is: n = plan time ÷ ∆t
At the outset, the agent faces three options: continue straight, turn left, or turn right, leading to three potential points for determining the optimal route (refer to Figure 8) The chosen point will be the one closest to the target destination.
Detecting possible collisions on the optimal route
To ensure safe navigation along the optimal route, the agent must assess potential collision risks with other agents By monitoring the position, direction, and speed of nearby agents, the agent can predict possible collisions For instance, as depicted in Figure 9, agent A determines that while the first and second positions on the route are clear of collisions, a potential conflict arises at the fourth position due to proximity to agent B In response to this risk, agent A can either reduce speed or alter its course to prevent a collision.
A’s current speed is higher than its safe speed limit, it will reduce speed Otherwise, it will decide to steer
Determining alternative route to avoid collisions
When there might be collisions in the planned route, an alternative route is calculated so that the alternative route is in parallel with the optimal route.
Improvement
The VTS is effectively designed with robust function tests, ensuring smooth operation in representing traffic flows However, additional features like Traffic Light control and enhanced data input/output capabilities are necessary for improvement Furthermore, while the agent-based model is functional, some attributes do not align well with the unique traffic conditions in Vietnam This section will detail the proposed enhancements.
Traffic lights play a crucial role in transportation systems, serving various types of roads, including railroads and air routes This simulator focuses specifically on land road traffic lights, featuring two essential states: red and green.
We developed a strategy for agents to identify the two traffic light states When approaching a red light, agents must reduce their speed, and the first agent to reach the red signal must come to a complete stop They are permitted to accelerate again once the light turns green.
In the next step, we added variables in the object TrafficLight and built up controlled function for it In the end, we designed the interface for this new fuction
Figure 10 Interface and simulation of the traffic light
Some corrections have been made to increase the reality of the simulation for the traffic in Vietnam
Mistaken perceptions of drivers based on outdated models can lead to inaccurate assessments of their behavior on the roads Research by Sameh El Hadouaj and Alexis Drogoul indicates that a driver's perception significantly influences their actions, while factors such as age and gender have minimal impact Consequently, the traditional categorization of driver types is becoming increasingly unrealistic.
Therefore, I implemented all the drivers with only one category of attributes However, each driver’s properties should be different from others Consequently,
I added random properties in each of these attributes to make them different agents with different properties
Table 1 An example of randomized parameters
In our experimental phase, we discovered that Vietnamese people often struggle to accurately assess safe travel distances As a result, they tend to drive without considering the appropriate space between vehicles To address this issue, we implemented a random safety distance attribute for each vehicle, ranging from 0 to 2 meters.
We also increase the range of the steering arc from 1.20 n/s to 1.6 n/s It allows vehicles to move more freely between the lanes.
Evaluation
Method
To evaluate the accuracy and realism of our simulations, we utilized the VTS simulator to conduct experiments, drawing from a diverse array of real data sources This included videos captured by mobile phones, digital cameras positioned on high buildings, online traffic footage from VOV, and archived recordings from the Hanoi Transportation Department However, we encountered challenges in selecting suitable data for our experiments, as many collected videos were compromised by limited traffic lanes, poor quality, or obstructed views.
Figure 11 Some examples of real time traffic data
The experiments were conducted at the Khuat Duy Tien – Tran Duy Hung crossroad, which features a roundabout, making it an ideal site for our research The parameters of this location are detailed in the table below.
Table 2 Parameters of KDT – TDH crossroad
Abbreviation: KDT: Khuat Duy Tien, PH: Pham Hung, TDH: Tran Duy Hung, HL: Hoa Lac
Figure 12 the Khuat Duy Tien – Tran Duy Hung crossroad in the simulator
Figure 13 the Khuat Duy Tien – Tran Duy Hung crossroad captured by the traffic camera
Traffic inflow and outflow data were collected from daily video recordings captured by VOV’s camera at 7 a.m., with vehicle counts obtained through video analysis The goal is to compare the traffic density from this statistical data with simulation results, ensuring that the number of vehicles entering matches the simulation's input and that the outflow corresponds to the output However, the statistical data is limited to a portion of the intersection and lacks details on the lane's start and end points, prompting calculations based only on vehicles at the camera's corners The study focuses on counting motorbikes as small vehicles and cars and trucks as large vehicles, as bicycles are infrequently observed Due to the difficulty of counting vehicles in real-time, the approach involves pausing the video every three seconds to tally the total number of vehicles passing through an entrance.
From … Density (number of vehicles going to the crossroad per 3 seconds) Avg Velocity
Go straight Turn left Turn around
Table 3 The information query form
From KDT Density (number of vehicles going to the crossroad per 3 seconds) Avg Velocity
Go straight 8 4 6 1 6 2 9 6 5 9 5 6 8 1 3 9 5 4 4 5 2 6 45 Turn left 3 2 3 2 1 1 1 3 3 2 3 2 5 1 3 8 4 7 2 3 3 1 35 Turn around 1 0 0 0 1 0 2 0 1 0 0 0 0 1 0 0 0 0 2 0 0 0 5
Table 4 An example of query data
The parameters of the agents are summarized as below
Table 5 Default parameters of the simulation
In further details, we considered the bus as a kind of big vehicles though it may not function as a normal truck or car
The numbers of inflow vehicles per minute is 423 small vehicles and 79 big vehicles which are distributed in the following graphs
Figure 14 Distribution of inflow vehicles in real data
The incoming statistics counted are illustrated as the graphs below:
Figure 15 Timegraph of inflow inflow vehicles in real data
From KDT From TDH From PH From HL
Incoming small vehicles per block 3s
Incoming big vehicles per block 3s
Results
As settings above, we compare the outflow rate of statistical data with the finished vehicles out of the crossroad in the simulator
The result we achieved is described as following graphs:
Figure 16 The worst case of achieved results
Figure 17 The best case of achieved results
After conducting 20 simulations, we found that while the output data often did not align with real-world data, in 20% of the cases, the results closely matched actual data Notably, the simulation recorded a total of 382 outgoing vehicles, which is comparable to the real-life figure of 398.
If we do normalization to all the data achieved during evaluation, we could reach a graph of comparison like below
Figure 18 Normalization of achieved results
In our study, we increased the number of vehicles on the road to assess its capacity limits As the vehicle count rose, the average travel speeds significantly declined With 600 vehicles—200 more than typical real-life traffic—the onset of congestion occurred more rapidly This trend is clearly illustrated in the accompanying graph, which highlights the decreasing velocity rates.
Figure 19 The decrease rate of velocity
This section will introduce an evaluation of traffic light simulation Accurately counting the number of vehicles stopping at red lights poses a challenge; therefore, we must explore alternative methods for collecting viable data to ensure the evaluation can proceed effectively.
Figure 20 Traffic light data observation
Discussion
The simulation demonstrated that maintaining the same input statistical data allows us to achieve a consistent relative outflow rate Furthermore, the statistical component report indicated that the average velocity of vehicles upon completion of their journey is approximately 40 km/h, closely aligning with the average speed of small vehicles observed at real crossroads.
The simulation features a randomized setting, leading to varied output data Although the collected data represents only one minute of real-life traffic, it closely mirrors actual traffic conditions This suggests that the simulator is capable of effectively handling real data.
Assessing road capacity is crucial for policymakers, but a lack of real data samples hinders the scientific validation of simulators Therefore, further research is necessary to address this gap.
The evaluation of traffic lights is crucial, as it requires a survey to determine the number of individuals crossing from the opposite lane while waiting for the red light.
Conclusion
Conclusion
This thesis introduces a novel solution to Vietnam's traffic challenges through the development of the VTS model, a cutting-edge traffic simulation tool This model assists transportation planners in identifying effective strategies to address issues like congestion and enables cost-effective testing of new designs prior to the construction of transportation infrastructure.
After gaining experience in state-of-the-art traffic simulation and multi-agent systems, I collaborated on the development of VTS, enhanced its functionality, and conducted experiments to assess its accuracy.
Using real data collected from video footage stored by VOV traffic, we can assess key features of the VTS and validate the accuracy of the simulator.
However, there are still many issues left due to the lacking of experiment data
We will work for it in the future development presented in the section below.
Future development
To enhance the traffic light system, we aim to gather more specific data for conducting targeted experiments The lack of suitable data currently limits our progress Additionally, there are several areas for improvement, including the system interface and the methods for inputting statistical data.
[1] Adina Magda Florea Introduction to Multi-Agent Systems, In Proc of Continuous Education Program on Intelligent Agents Technology and Knowledge Processing, Bucharest, 2001, pp 49-60
[2] Agent Based Modeling FAQ
[3] Charles M Macal & Michael J North Introduction to Agent-based Modeling and Simulation
[4] Charles M Macal & Michael J North Tutorial on Agent-Based Modeling and Simulation Part 2: How to model with Agents Proceedings of the 38th conference on Winter simulation, 2006, pp 73-83
[5] Differences between objects and agents
[6] FERBER, J., 1999 Multi-agent Systems: Introduction to Distributed Artificial Intelligence England: Addison Wesley
[8] James J Odell Objects and Agents Compared Journal of Object Technology, vol.1, no.1, 2002, pp 41-53
[9] Katia P Sycara Multiagent Systems AI Magazine, 1998
[10] Matti Pursula, Simulation of Traffic System – An Overview Journal of Geographic Information and Decision Analysis, vol.3, no.1, 1999, pp 1-8.