AU2021103022A4 - A Method of Controlling Traffic Light Based on Fog Computing and Reinforcement Learning - Google Patents
A Method of Controlling Traffic Light Based on Fog Computing and Reinforcement Learning Download PDFInfo
- Publication number
- AU2021103022A4 AU2021103022A4 AU2021103022A AU2021103022A AU2021103022A4 AU 2021103022 A4 AU2021103022 A4 AU 2021103022A4 AU 2021103022 A AU2021103022 A AU 2021103022A AU 2021103022 A AU2021103022 A AU 2021103022A AU 2021103022 A4 AU2021103022 A4 AU 2021103022A4
- Authority
- AU
- Australia
- Prior art keywords
- intersection
- time
- traffic
- vehicles
- adjacent intersections
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/07—Controlling traffic signals
- G08G1/08—Controlling traffic signals according to detected number or speed of vehicles
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/065—Traffic control systems for road vehicles by counting the vehicles in a section of the road or in a parking area, i.e. comparing incoming count with outgoing count
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
- G16Y10/00—Economic sectors
- G16Y10/40—Transportation
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a method of controlling traffic light based on fog computing and
reinforcement learning. Firstly, the fog nodes at the intersection collect the traffic condition of
the intersection and broadcast it to the fog nodes at the adjacent intersections, collect the
number of vehicles at the intersection and calculates the required green light time for vehicles
to pass through current intersection and calculate the required green light time for the traffic
flow at the adjacent intersections to pass through the intersection, the required green light time
for the traffic flow at the adjacent intersections to pass through the intersection, the combined
number of vehicles at the current intersection and adjacent intersections and the total green
light time needed for the current intersection. The Agent of the current intersection acts the
calculated green light time on the traffic signal, observes the feedback payoff signal, and the
environmental state transferred to the next moment. The invention can allocate the green light
time according to the real-time number of vehicles at the intersection, which can reduce the
parking time and the number of parking, improve the road regulation efficiency, and can
effectively reduce traffic congestion.
1/4
FIGURES
anna~n
Figure 1 fi
server
Figure 2
Figure
Description
1/4
anna~n
server
Figure 1 fi
Figure 2
Figure
A Method of Controlling Traffic Light Based on Fog Computing and Reinforcement
Learning
The present invention belongs to the field of intelligent transportation technology, and specifically relates to a method of controlling traffic light based on fog computing and reinforcement learning.
With the accelerated urbanization and the improvement of people's living standard, there are more and more vehicles, resulting in overburdened urban roads and increasingly serious traffic congestion. Every year, countries around the world suffer huge economic losses due to traffic congestion, which shows that traffic congestion has seriously restricted economic development. According to the survey, traffic accidents caused by traffic congestion are still on the rise every year, and there are a series of energy problems and environmental problems that follow; traffic congestion not only causes waste of transportation resources, but also greatly reduces transportation efficiency, consumes excessive social costs, and seriously hinders the development of cities, so it is urgent to alleviate traffic congestion. Telematics aims to address the current challenging new needs in the field of transportation systems; vehicles in Telematics can collect status information about themselves and their environment through sensors, radio frequency identification technology, roadside units and other devices, and use the Internet and computer technology to transmit, analyze and process the current information to achieve intelligent traffic management. Telematics aims to address the current challenging new needs in the field of transportation systems; vehicles in Telematics can collect status information about themselves and their environment through sensors, radio frequency identification technology, roadside units and other devices, and use the Internet and computer technology to transmit, analyze and process the current information to achieve intelligent traffic management.
Fog computing is a highly virtualized network platform, which consists of a large number of edge terminal nodes and routing devices, and provides computing, storage and other services between the cloud and network terminal devices, and its essence is decentralized edge computing. Fog computing has the advantages of cognitive, efficient and low latency, which can realize information sharing through the fog platform and allow data to "talk ". Artificial intelligence is of interest because of its ability to handle large amounts of data in complex situations; reinforcement learning is based on theories such as animal learning and adaptive control of parameter perturbations, and is one of the methods of machine learning. Reinforcement learning is "trial-and-error learning", in which an intelligence evaluates good or bad behavior by means of payoff signals; the goal of the intelligence is to find the optimal strategy in each discrete state and expects to obtain a discounted payoff. The goal is to find the optimal strategy in each discrete state, expecting to obtain the maximum discounted payoff.
The purpose of the present invention is to provide a method of controlling traffic light based on fog computing and reinforcement learning, which solves the problem that the existing traffic signal control method has poor real-time performance and the traffic information at each intersection cannot be shared in real time. The technical solution used in the present invention is a method of controlling traffic light based on fog computing and reinforcement learning, implemented in the following steps: Step 1: the current intersection's fog nodes collect intersection traffic conditions and broadcast them to the adjacent intersections' fog nodes. Step 2: the current intersection's Agent collects the number of vehicles at the intersection and calculates the required green light time Tg for vehicles to pass through the
current intersection. Step 3: the current intersection's Agent selects the corresponding impact factor value based on the number of vehicles at the adjacent intersections and calculates the green light timeTighb,,, required for the traffic to pass through the intersection at the adjacent intersections. Step 4, the current intersection calculates the green light time igb, required for the traffic flow of the adjacent intersections to pass through the intersection. Step 5: the current intersection combines the number of vehicles at its own intersection and adjacent intersections to calculate the total green light time T,,,, it needs.
Step 6: the Agent of the current intersection acts the calculated green light duration on the traffic signal, observes the feedback payoff signal, and the environmental state transferred to the next moment.
Step 7: the Agent at the current intersection updates the Q table according to the update rule. The invention is also characterized by: The testing of intersection traffic conditions in step 1 includes: The number of vehicles driving to the next intersection. The time T,,,,, at which the upstream intersection of the test intersection starts
releasing traffic flow. The time Tend at which the upstream intersection of the test intersection ends the
release of the traffic flow. The time T7,,,, at which the traffic flow at the current intersection reaches the next
intersection, as in equation (1) below:
arrive S 11 S'1 2 (1) V
In equation, IS- St2 |is the distance between two intersections and Vis the average travel speed of the traffic flow. In step 2 in which the current intersection calculates its own required green light time T g based on its own number of vehicles, as follows in equation (2): T=d+2*NumVeh (2)
In equation, d is the delay time of starting the vehicles and NumVeh is the number of vehicles.
T In step 2, when g is not less than the maximum green time , the traffic light green T. time is set to the maximum green time, and go to step 6; when g is not greater than the T. maximum green time, there is no vehicle at the adjacent intersections, g is set to the actual green time of the intersection, and go to step 6; when there is a vehicle at the adjacent intersections, the go to step 3. In step 3, the impact factor 8 value indicates the degree of impact; the greater the
,8 value, the greater pressure for the adjacent intersections to the current intersection traffic, according to the number of vehicles at the adjacent intersections of the current intersection to take the corresponding value, the 8 value is used to regulate the traffic
flow between the current intersection and its adjacent intersections.
In step 5 the equation of otaI is shown in equation (3) as follows:
Ta,1, = T+ T+eighbor (3)
In the step 5 when T,,,,, is not less than the maximum green time, the traffic light
green time is set to the maximum green time, and go to step 6; when T,,,a, is not greater
than the maximum green time, T,,,,, is set to the actual green time of the intersection,and
go to step 6. In step 7, the Q table is a matrix, and the update rule is the following equation:
Q,(s,a)=(1-a)Q,(s,a)+art,l+7maxQ(s,a) (4)
In equation, s is the state, a is the action, r is the return value, a is the learning rate,
7is the discount factor, and Q(s,a) is the Q value of the state selection action at the t
moment.
In step 7, determining whether the Q matrix converges, the Q matrix converges, the program ends, and the Q matrix does not converge and go to step 2. The beneficial effects of the present invention are: A traffic light control method based on fog computing and reinforcement learning used in the present invention has the advantage that, compared with the traditional traffic light control method, the traffic light control method based on fog computing and reinforcement learning of the present invention has the advantage that it can allocate the green light time according to the real-time number of vehicles at the intersection, which can reduce the stopping time and the number of stops, improve the road regulation efficiency, and can effectively reduce traffic congestion.
Figure 1 is a fog platform-based traffic light management system from the the method of controlling traffic light based on fog computing and reinforcement learning of the present invention. Figure 2 is a schematic diagram for multiple intersections from the method of controlling traffic light based on fog computing and reinforcement learning of the present invention. Figure 3 is a flowchart of a traffic light control algorithm for multiple intersections from the method of controlling traffic light based on fog computing and reinforcement learning of the present invention. Figure 4 is a diagram of the average vehicle throughput from the method of controlling traffic light based on fog computing and reinforcement learning of the present invention; and Figure 5 is a graph of the average vehicle delay time from the method of controlling traffic light based on fog computing and reinforcement learning of the present invention Figure 6 is a graph of average vehicle queue length from the method of controlling traffic light based on fog computing and reinforcement learning of the present invention.
The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Between cloud computing and personal computing, fog computing is a semi virtualized, distributed service computing architecture model that is close to the ground. Compared to cloud computing, its required time to transmit information is short and its computing speed is fast, which meets the requirement of high real-time traffic light.
Reinforcement learning is a kind of machine learning, in the process of interacting with the environment, the Agent can achieve the set goals through learning strategies, the traffic flow does not have a specific law, and a scheme with a certain law cannot be used to control the traffic light, while reinforcement learning can interact with the environment in real time with better reliability, so the present invention combines the features of fog computing and reinforcement learning to control the traffic light, i.e., FRTL (Fog Reinforcement Traffic Light) control method. It comprises the following steps: Q learning algorithm in reinforcement learning is used to develop the traffic light control algorithm, and the three elements of Q learning algorithm are state, action, and payoff. The invention uses the number of vehicles as the traffic state, the green light time as the action, and establishes the payoff function based on the vehicle delay time, which is obtained directly from the delay time detector from VISSIM. A positive payoff is taken, i.e., the smaller the delay, the larger the payoff value. Equation (5) is the traffic state expression and Equation (6) is the payoff function.
Sj ={NumVeh ,NumVel4,NumVeh ,.,NumVeh.} (5)
Inequation, S is the state of the ith intersection at moment t andNumVeht is the
number of vehicles from intersection i to intersection j at moment t.
10,DelayTime =0 8 ,0< DelayTime 15 6 ,15 < DelayTime 30 4 ,30<DelayTime 45 2 ,45 < DelayTime ! 60 r at, a' S', Sj)= 0 ,60 < DelayTime ! 75 (6) -2,75< DelayTime 95 -4,95 < DelayTime ! 115 -6,115 < DelayTime ! 135 -8,135 < DelayTime ! 155 -10,DelayTime>155
The present invention provides the the method of controlling traffic light based on fog computing and reinforcement learning, implemented in the following steps:
Step 1: As shown in Figure 1, the fog nodes deployed at each intersection will collect the traffic information of the current intersection and broadcast the traffic information of the current intersection to the fog nodes at adjacent intersections for information sharing through the fog computing platform, and the broadcasted information includes: The number of vehicles driving to the next intersection. The time T,,,, at which the upstream intersection of the test intersection starts
releasing traffic flow. The time Tnd at which the upstream intersection of the test intersection ends the
release of the traffic flow. The time T,, at which the traffic flow at the current intersection reaches the next
intersection, as in equation (1) below:
arriv -' S'H (1 V
In equation, ISt- St12 is the distance between two intersections and Vis the average travel speed of the traffic flow. Step 2: The Agent at the current intersection calculates its own required green time based on its own number of vehicles, as shown in Equation (2). T=d+2*NumVeh (2)
In equation, d is the delay time of starting the vehicles and NumVeh is the number of vehicles. T. When g is not less than the maximum green time , the traffic light green time is set T. to the maximum green time, and go to step 6; when g is not greater than the maximum T. green time, there is no vehicle at the adjacent intersections, g is set to the actual green time of the intersection, and go to step 6; when there is a vehicle at the adjacent intersections, the go to step 3. Step 3: The Agent at the current intersection selects the corresponding impact factor values/J from Table 1 based on the number of vehicles at the adjacent intersections and
calculates the green light time T.,,,b, required for the traffic at the adjacent intersections
to pass through the intersection.
Table 1: The Correspondence between values and the number of vehicles at adjacent intersections Value 8 Number of vehicles at adjacent intersections 0.8 NumVeh > 60
0.7 50 <NumVeh < 60
0.6 40 <NumVeh < 50
0.5 30 <NumVeh < 40
0.4 20 < NumVeh < 30
0.3 10 <NumVeh < 20
0.2 NumVeh < 10
As the traffic flow of the adjacent intersections will cause traffic pressure on the current intersection, as shown in Figure 2, so the impact factor # is used to indicate the degree of impact. The larger the value p, the greater the traffic pressure on the current intersection from the adjacent intersections. The value p and the number of vehicles at the adjacent intersections correspond to the table shown in Table 1. The role of the value is used to regulate the traffic flow between the current intersection and its adjacent intersections, according to the number of vehicles at the adjacent intersections of the current intersection to take the corresponding value, thus avoiding traffic congestion caused by releasing too many vehicles from the number of adjacent intersections. Step 4: The current intersection uses equation (3) to calculate the green light time Tneighbor required for traffic flowing through the intersection at the adjacent intersections. Step 5: The current intersection integrates the number of vehicles at its own intersection and adjacent intersections to calculate the total green time T,,,, needed for
itself, with the formula shown in Equation (3).
To,=Tamve+TgTn eighbor (3)
When T,,,, is not less than the maximum green time, the traffic light green time is
set to the maximum green time, and go to Step 6; when T,,,,, is not greater than the
maximum green time, T,,,, is set to the actual green time of the intersection,and go to
Step 6. Step 6: The Agent of the current intersection acts the calculated green light time to the traffic signal, observes the feedback payoff signal, and the environmental state transferred to the next moment.
Step 7: The Agent at the current intersection updates the Q table according to the update rule, Q table is a action criterion table, that is, a matrix, as shown in Table 2: s denotes the environment state, a denotes the action. According to the value Q, it can calculate the payoff obtained by choosing a certain action in that state, the update rule is shown in Equation (6).
Q, (s,a)=(1-a)Q,(s,a)+a r,y+ maxQ,(s,a)l (4)
In equation, s is the state, a is the action, r is the return value, a isthelearning
rate, 7 is the discount factor, and Q(s,a) is the Q value of the state selection action
atthe t moment.
Finally determining whether the Q matrix converges; If the Q matrix converges, the
procedure ends; if the Q matrix does not converge, procedure goes to step 2. Table 2: Table Q
Q ai a2
Si 1 3
S2 2 4
As shown in Figure 3, the proposed adaptive traffic light control method is simulated using the VISSIM-Excle VBA-MATLAB integrated simulation platform according to the traffic light control algorithm flowchart, and finally the simulation results are analyzed with the parameter settings shown in Tables 3 and 4, respectively. Table 3 Learning system parameter settings
Parameter name Parameter value
8 0.3
a 0.6
0.5
Table 4 VISSIM parameter settings Road network setting VISSIM parameter settings Traffic rule right-hand traffic Shape of the intersection Cruciform Width of the lane 3.5 m
Number of the inlet lane Bi-directional single lane Number of vehicles per lane 200 - 2000 vehicles Simulation time 3600s Random seed 15 Simulation speed 10.0 Sim.sec./s Simulation identification 5 Time steps(s)/Sim.sec. Traffic light 16 (4 per intersection) Number of phases Two phases (East-West Direct, North-South Direct) Stream detector At the parking line of each entry lane Based on the above simulation conditions, the following simulation scenarios are performed: Three metrics: average vehicle throughput, average vehicle delay time and average vehicle queue length at the intersection, are used to evaluate the performance of the FRTL control method. Embodiment 1: As shown in Figure 4, the horizontal coordinate is the number of vehicles per hour in each inlet lane, and the vertical coordinate is the average throughput, from the figure, we can see that when the number of vehicles is lower than 300veh/h, the average vehicle throughput of the intersection under the three control methods is almost equal. When the number of vehicles is between 300veh/h and 600veh/h, the throughput of the intersection under the trunk road control method and FRTL control method is slightly higher than that under the dayparting control method. Since the trunk road control method only considers the traffic flow on the trunk road and ignores the impact of non-trunk road traffic flow on the trunk road, so when the number of vehicles exceeds 1500veh/h, the difference between the regulation effect of the trunk road control method and the time-sharing control method is very small, and the maximum regulation effect is basically achieved. However, when the number of vehicles is greater than 600veh /h, the average vehicle throughput of the intersection under FRTL control mode is the largest and the regulation effect is the best. Embodiment 2: As shown in Figure 5, the horizontal coordinate is the number of vehicles per hour and the vertical coordinate is the average vehicle delay time; from the figure, we can see that the average vehicle delay time under the FRTL control method is slightly higher than under the trunk road control method when the number of vehicles is about 1500veh/h, which is caused by the lower traffic volume on the secondary roads during that time period (the trunk road control method has to set priority to the roads and that is why the concept of non-trunk road exists; whereas the FRTL control method considers the roads to have the same priority), but, overall, the FRTL control method has the lowest average vehicle delays. Embodiment 3: As shown in Figure 6, the horizontal coordinate is the number of vehicles per hour for each inlet lane. It can be seen from the figure that the vehicle queue length is longer at multiple intersections than at single intersections, which is caused by the mutual influence of vehicles between adjacent intersections. The simulation results show that the control effects of the three control methods do not differ much in the case of relatively small traffic flow, and the difference in the average vehicle queue length is very small. With the increase of traffic flow, when the vehicle input is 1500veb/h or 2000veh/h; the FRTL control method has a better control effect on reducing the queue length, and overall, the average vehicle queue length under FRTL control method is the shortest.
Claims (9)
1. A method of controlling traffic light based on fog computing and reinforcement learning characterizes in that it is implemented in the following steps: Step 1: the current intersection's fog nodes collect intersection traffic conditions and broadcast them to the adjacent intersections' fog nodes. Step 2: the current intersection's Agent collects the number of vehicles at the intersection and calculates the required green light time T for vehicles to pass through the
current intersection. Step 3: the current intersection's Agent selects the corresponding impact factor value based on the number of vehicles at the adjacent intersections and calculates the green light time eighbor required for the traffic to pass through the intersection at the adjacent
intersections. Step 4, the current intersection calculates the green light time Tighb, required for the
traffic flow of the adjacent intersections to pass through the intersection. Step 5: the current intersection combines the number of vehicles at its own intersection and adjacent intersections to calculate the total green light time T,,,,it needs.
Step 6: the Agent of the current intersection acts the calculated green light duration on the traffic signal, observes the feedback payoff signal, and the environmental state transferred to the next moment. Step 7: the Agent at the current intersection updates the Q table according to the update rule.
2. The method of controlling traffic lights based on fog computing and reinforcement learning according to claim 1 characterizes in that the step 1 of testing traffic conditions at an intersection comprises: The number of vehicles driving to the next intersection. The time T,,, at which the upstream intersection of the test intersection starts
releasing traffic flow. The time Tdd at which the upstream intersection of the test intersection ends the
release of the traffic flow.
The time Tr,,, at which the traffic flow at the current intersection reaches the next
intersection, as in equation (1) below:
T rri_ Stfl -Sfl 2 (1 V
In equation, IS- St12 |is the distance between two intersections and Vis the average travel speed of the traffic flow.
3. The method of controlling traffic lights based on fog computing and reinforcement learning according to claim 1 characterizes in that the step 2 in which the current
intersection calculates its own required green light time T based on its own number of vehicles, as follows in equation (2): T=d+2*NumVeh (2)
In equation, d is the delay time of starting the vehicles and NumVeh is the number of vehicles.
4. The method of controlling traffic lights based on fog computing and reinforcement T. learning according to claim 1 or 3 characterizes in that the step 2 when g is not less than the maximum green time , the traffic light green time is set to the maximum green time, T. and go to step 6; when g is not greater than the maximum green time, there is no vehicle T at the adjacent intersections, g is set to the actual green time of the intersection, and go to step 6; when there is a vehicle at the adjacent intersections, the go to step 3.
5. The method of controlling traffic lights based on fog computing and reinforcement learning according to claim 1 characterizes in that in the step 3 the impact factor # value
indicates the degree of impact; the greater the # value, the greater pressure for the
adjacent intersections to the current intersection traffic, according to the number of vehicles at the adjacent intersections of the current intersection to take the corresponding # value, the 8 value is used to regulate the traffic flow between the current intersection and its
adjacent intersections.
6. The method of controlling traffic lights based on fog computing and reinforcement
learning according to claim 1 characterizes in that in step 5 the equation of Total is shown in equation (3) as follows:
7ot, ,=T,+T,,,z+Teighbor (3)
7. The method of controlling traffic lights based on fog computing and reinforcement learning according to claim 1 or 6 characterizes in that in the step 5 when T,,, is not less
than the maximum green time, the traffic light green time is set to the maximum green time, and go to step 6; when T,,,,, is not greater than the maximum green time, T is
set to the actual green time of the intersection,and go to step 6.
8. The method of controlling traffic lights based on fog computing and reinforcement learning according to claim 1 characterizes in that the step 7 in which the Q table is a matrix, i.e., a Q matrix, and the update rule is the following equation:
Q,±(s,a)=(1-a)Q,(s,a)+art,+ymaxQ,(s,a)] (4)
In the equation, s is the state, a is the action, r is the return value, a isthelearning
rate, 7 is the discount factor, and Q(s,a) is the Q value of the state selection action at
the t moment.
9. The method of controlling traffic lights based on fog computing and reinforcement learning according to claim 1 or 8 characterizes in that the step 7 determines whether the Q matrix converges, the Q matrix converges, the program ends, and the Q matrix does not converge and go to step 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| AU2021103022A AU2021103022A4 (en) | 2021-06-02 | 2021-06-02 | A Method of Controlling Traffic Light Based on Fog Computing and Reinforcement Learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| AU2021103022A AU2021103022A4 (en) | 2021-06-02 | 2021-06-02 | A Method of Controlling Traffic Light Based on Fog Computing and Reinforcement Learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| AU2021103022A4 true AU2021103022A4 (en) | 2021-07-22 |
Family
ID=76858482
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| AU2021103022A Active AU2021103022A4 (en) | 2021-06-02 | 2021-06-02 | A Method of Controlling Traffic Light Based on Fog Computing and Reinforcement Learning |
Country Status (1)
| Country | Link |
|---|---|
| AU (1) | AU2021103022A4 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113838296A (en) * | 2021-09-17 | 2021-12-24 | 中山大学 | Traffic signal control method, device, equipment and storage medium |
| CN115083149A (en) * | 2022-05-19 | 2022-09-20 | 华东师范大学 | Real-time monitoring reinforcement learning variable-duration signal lamp control method |
-
2021
- 2021-06-02 AU AU2021103022A patent/AU2021103022A4/en active Active
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113838296A (en) * | 2021-09-17 | 2021-12-24 | 中山大学 | Traffic signal control method, device, equipment and storage medium |
| CN115083149A (en) * | 2022-05-19 | 2022-09-20 | 华东师范大学 | Real-time monitoring reinforcement learning variable-duration signal lamp control method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108510764B (en) | Multi-intersection self-adaptive phase difference coordination control system and method based on Q learning | |
| CN111951549B (en) | Self-adaptive traffic signal lamp control method and system in networked vehicle environment | |
| CN111047884A (en) | Traffic light control method based on fog calculation and reinforcement learning | |
| AU2021103022A4 (en) | A Method of Controlling Traffic Light Based on Fog Computing and Reinforcement Learning | |
| CN106960584A (en) | A kind of traffic control method and device of self adaptation crossroad traffic signal lamp | |
| CN109887289A (en) | A kind of network vehicle flowrate maximization approach of urban traffic network model | |
| CN101593419A (en) | A kind of city road network traffic flow intelligent coordination control method with public traffic in priority | |
| CN107507430A (en) | A kind of urban road crossing traffic control method and system | |
| CN107730886A (en) | Dynamic optimization method for traffic signals at urban intersections in Internet of vehicles environment | |
| CN105047057B (en) | A kind of high through street Macro-traffic Flow emulation mode of multi-user's multilane | |
| US10891855B2 (en) | Method to schedule intelligent traffic lights in real time based on digital infochemicals | |
| CN104916142A (en) | Adaptive intersection traffic signal control method of trunk road priority | |
| CN109816978B (en) | Regional group traffic guidance system and method considering dynamic response behaviors of drivers | |
| CN105551250A (en) | Method for discriminating urban road intersection operation state on the basis of interval clustering | |
| CN114495547B (en) | Signal intersection cooperative passing method for automatically-driven automobile | |
| Guerrero-Ibanez et al. | A policy-based multi-agent management approach for intelligent traffic-light control | |
| CN112365714A (en) | Traffic signal control method for intersection of intelligent rail passing main branch road | |
| CN105743783B (en) | Car networking network node screening technique based on BS-TS and autoencoder network | |
| CN114743396A (en) | Traffic guidance scheduling system based on particle swarm optimization algorithm and scheduling method thereof | |
| Wu et al. | A V2I communication-based pipeline model for adaptive urban traffic light scheduling | |
| Wang et al. | TLB-VTL: 3-level buffer based virtual traffic light scheme for intelligent collaborative intersections | |
| Dong et al. | Analysis and control of intelligent traffic signal system based on adaptive fuzzy neural network | |
| Du et al. | Dynamic capacity estimation of mixed traffic flows with application in adaptive traffic signal control | |
| CN112884194A (en) | Variable lane switching and signal timing method based on signalized intersection running condition | |
| Gao et al. | Coordinated control method of intersection traffic light in one-way road based on V2X |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| FGI | Letters patent sealed or granted (innovation patent) |