Method and system for adjusting variable traffic sign in ice and snow weather

Abstract

This invention discloses a method and system for adjusting variable traffic signs in icy and snowy weather. The method includes: First, collecting various environmental parameters affecting driver behavior at the location of the variable traffic sign using monitoring equipment, and generating a corresponding environmental state vector based on the collected parameters. Then, inputting the generated environmental state vector into a pre-trained deep reinforcement learning model, which determines the dynamic adjustment amount of the variable traffic sign. The dynamic adjustment amount can include positional adjustments between adjacent signs, or angular adjustments of the variable traffic sign in various directions in three-dimensional space. Finally, determining the deployment location and spacing of the variable traffic sign based on the dynamic adjustment amount output by the deep reinforcement learning model ensures that drivers can quickly and accurately identify traffic signs and operate their vehicles safely, thus guaranteeing traffic safety in the diversion zone.

Description

Technical Field

[0001] This invention relates to the field of control system technology for controlling the traffic movement of road vehicles, and specifically to a method and system for adjusting variable traffic signs in icy and snowy weather. Background Technology

[0002] Once deployed, traffic signs along highways and urban roads cannot be adjusted and must remain fixed in their installation locations throughout the operational period. Fixing traffic signs on the main road sections generally does not affect their ability to control drivers. However, the merging zones between highways and urban expressways are accident-prone areas with severe traffic congestion. Drivers have different requirements for the spacing of traffic signs depending on environmental conditions. Especially in adverse weather conditions such as ice and snow, the road surface friction coefficient is significantly reduced, and vehicle braking distances increase. Drivers' behavior in merging zones is affected by factors such as road surface friction coefficient and icing conditions, requiring a greater spacing for traffic signs than in normal weather. Fixing the positions of traffic signs would compromise traffic safety in merging zones and increase the risk of accidents. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention proposes a method and system for adjusting variable traffic signs in icy and snowy weather. This method can rationally adjust the position of traffic signs according to different traffic environment conditions to meet drivers' requirements for traffic sign deployment under varying traffic conditions, thereby improving road safety. The specific technical solution is as follows: In a first aspect, a method for adjusting variable traffic signs in icy and snowy weather is provided. In a first implementable mode of the first aspect, it includes: Obtain the environmental state vector of the area where the variable traffic sign is located; Based on the environmental state vector, the dynamic adjustment amount of the variable traffic sign is determined by a trained deep reinforcement learning model. The position of the variable traffic sign is adjusted according to the dynamic adjustment amount.

[0004] In conjunction with the first possible implementation of the first aspect, in the second possible implementation of the first aspect, the environmental state vector includes: road surface friction coefficient, ambient temperature, icing thickness, road visibility, weather level, snowfall, traffic flow and / or current speed limit.

[0005] In conjunction with the first implementable method of the first aspect, the third implementable method of the first aspect also includes: Based on the environmental state vector, the variable speed limit of the variable traffic sign is determined by a trained deep reinforcement learning model, and the speed limit of the variable traffic sign is adjusted according to the variable speed limit.

[0006] In conjunction with the first possible implementation of the first aspect, in the fourth possible implementation of the first aspect, the action evaluation function of the deep reinforcement learning model includes: A safety function is constructed based on the current speed limit, road surface friction coefficient, ambient temperature, ice thickness, road visibility, weather level, snowfall, and traffic flow. A traffic efficiency function is constructed based on vehicle flow, maximum road capacity, and road segment design speed. The control cost function is constructed based on the current speed limit and dynamic adjustment amount.

[0007] In conjunction with the first possible implementation of the first aspect, in the fifth possible implementation of the first aspect, a deep deterministic policy gradient algorithm is used to train the decision network of the deep reinforcement learning model.

[0008] In conjunction with the first possible implementation of the first aspect, in the sixth possible implementation of the first aspect, the variable traffic sign includes a traffic sign and a movable rail, the traffic sign being mounted on a slide of the movable rail.

[0009] Secondly, a variable traffic sign adjustment system for icy and snowy weather is provided. In a first feasible implementation of this second aspect, it includes: The perception module is configured to acquire the environmental state vector of the area where the variable traffic sign is located. The decision module is configured to determine the dynamic adjustment amount of the variable traffic sign based on the environmental state vector using a trained deep reinforcement learning model. The adjustment module is configured to adjust the position of the variable traffic sign according to the dynamic adjustment amount.

[0010] In conjunction with the first possible implementation of the second aspect, in the second possible implementation of the second aspect, the decision module includes a speed limit decision unit configured to determine the variable speed limit value of the variable traffic sign based on the environmental state vector using a trained deep reinforcement learning model.

[0011] In conjunction with the first possible implementation of the second aspect, in the third possible implementation of the second aspect, the decision module includes a training unit configured to train the deep reinforcement learning model using a deep deterministic policy gradient algorithm.

[0012] Beneficial Effects: The variable traffic sign adjustment method and system for icy and snowy weather of this invention can sense the environmental state vector of the diversion zone where the variable traffic sign is located. Based on the environmental state vector, a pre-trained deep reinforcement learning model can determine the dynamic adjustment amount of the variable traffic sign under the current weather conditions, and adjust the position and spacing of the variable traffic sign accordingly. This enables real-time dynamic automatic adjustment of the position and spacing of speed limit signs during operation, meeting drivers' requirements for traffic sign deployment under different traffic environments and improving road traffic safety. Attached Figure Description

[0013] To more clearly illustrate the specific embodiments of the present invention, the accompanying drawings used in the specific embodiments will be briefly described below. In all the drawings, the elements or parts are not necessarily drawn to scale.

[0014] Figure 1 A flowchart of a method for adjusting variable traffic signs in icy and snowy weather according to an embodiment of the present invention; Figure 2 This is a system block diagram of a variable traffic sign adjustment system for icy and snowy weather, provided as an embodiment of the present invention. Detailed Implementation

[0015] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention.

[0016] like Figure 1 The flowchart shown illustrates a method for adjusting variable traffic signs in icy and snowy weather. This method includes: Step 1: Obtain the environmental state vector of the area where the variable traffic sign is located; Step 2: Based on the environmental state vector, determine the dynamic adjustment amount of the variable traffic sign using a trained deep reinforcement learning model; Step 3: Adjust the position of the variable traffic sign according to the dynamic adjustment amount.

[0017] Specifically, firstly, monitoring equipment can be used to collect various environmental parameters affecting driver behavior at the location of variable traffic signs, such as ambient temperature and road surface friction coefficient, and generate corresponding environmental state vectors based on these parameters. Then, the generated environmental state vectors can be input into a pre-trained deep reinforcement learning model, which determines the dynamic adjustment amount of the variable traffic signs. The dynamic adjustment amount can include positional adjustments between adjacent signs, or angular adjustments of the variable traffic signs in various directions in three-dimensional space. Finally, the deployment location and spacing of the variable traffic signs can be determined based on the dynamic adjustment amounts output by the deep reinforcement learning model, thereby ensuring that drivers can quickly and accurately identify traffic signs and operate their vehicles safely, thus guaranteeing traffic safety in the diversion zone.

[0018] For example, adjusting the spacing between two adjacent hard shoulder clearance signs can adjust the permitted length of the hard shoulder. Alternatively, adjusting variable speed limit signs forward can alert drivers to the speed limit in the divergence zone, allowing them to slow down in advance.

[0019] In this embodiment, optionally, the environmental state vector includes: road surface friction coefficient, ambient temperature, icing thickness, road visibility, weather level, snowfall, traffic flow, and / or current speed limit.

[0020] Specifically, the environmental state vector includes the road surface friction coefficient, ambient temperature, icing thickness, road visibility, weather level, snowfall, traffic flow, and current speed limit. These factors all affect the speed limit for the next time period; therefore, they can be selected as parameters to determine the variable speed limit adjustment.

[0021] In this embodiment, optionally, it also includes: Based on the environmental state vector, the variable speed limit of the variable traffic sign is determined by a trained deep reinforcement learning model, and the speed limit of the variable traffic sign is adjusted according to the variable speed limit.

[0022] Specifically, variable traffic signs can be variable speed limit signs. Based on collected environmental state vectors, the dynamic adjustment determined by a trained deep reinforcement learning model can include not only positional adjustments but also the variable speed limit value of the sign. The speed limit value displayed on the variable speed limit sign can be adjusted according to the variable speed limit, thereby adaptively adjusting the speed limit in the diversion zone based on the current environmental conditions. This ensures that drivers can quickly drive at an appropriate speed, guaranteeing traffic safety in the diversion zone.

[0023] In this embodiment, optionally, the action evaluation function of the deep reinforcement learning model includes: A safety function is constructed based on the current speed limit, road surface friction coefficient, ambient temperature, ice thickness, road visibility, weather level, snowfall, and traffic flow. A traffic efficiency function is constructed based on vehicle flow, maximum road capacity, and road segment design speed. The control cost function is constructed based on the current speed limit and dynamic adjustment amount.

[0024] Specifically, when designing the action evaluation function for the deep reinforcement learning model, road safety, traffic efficiency, and control cost were comprehensively considered. The designed action evaluation function includes a safety function, a traffic efficiency function, and a control cost function. The specific expression of the designed action evaluation function is as follows: ; in, , and These are the weights corresponding to road safety, traffic efficiency, and control costs, used to adjust the balance between road safety, traffic efficiency, and control costs. Let the environment state vector be... This is a dynamically adjustable amount.

[0025] For security functions, the specific expression is: ; in, Based on the current speed limit Road surface friction coefficient Ambient temperature and ice thickness Road visibility Weather level Snowfall Traffic flow The constructed accident occurrence probability function is specifically expressed as follows: ; in, , , , , , , , These are the weighting coefficients corresponding to the current speed limit, road surface friction coefficient, ambient temperature and icing thickness, road visibility, weather level, snowfall, and traffic volume. When calculating the probability of an accident, the environmental parameters can first be normalized to eliminate dimensional differences between them. Then, combined with the weighting coefficients corresponding to each environmental parameter, the probability of an accident is calculated using the accident probability function.

[0026] The traffic efficiency function is expressed as follows: ; in, For vehicle traffic flow, The maximum traffic capacity of the diversion zone. Design speeds for sections of roads in the diversion zone.

[0027] The specific expression for the cost function is as follows: ; in, It is a variable speed limit value. To adjust the weighting coefficients, This is the amount of position adjustment.

[0028] In this embodiment, optionally, the decision network of the deep reinforcement learning model is trained using a deep deterministic policy gradient algorithm.

[0029] Specifically, the simulation can model vehicle traffic conditions under varying road surface friction coefficients, ambient temperatures, icing thickness, road visibility, weather conditions, snowfall, traffic flow, and current speed limits. During training, the environmental state vector for each time step is acquired, and the variable traffic signs are adjusted dynamically according to the simulation strategy. The environmental state vector for the next time step is then generated, and the action reward is evaluated using an action evaluation function. Subsequently, experience replay and policy optimization methods can be used to update the policy parameters of the policy network in the deep reinforcement learning model to maximize the expected cumulative reward. .

[0030] in, Indicates according to strategy Expected value of the sample. For strategy parameters, Indicates time, The discount factor, used to measure the importance of future rewards, is used to calculate the policy gradient update formula: ; in, For learning rate, The policy gradient can be calculated using forward propagation and directional propagation based on the chain rule.

[0031] In this embodiment, optionally, the variable traffic sign includes a traffic sign and a movable guide rail, with the traffic sign mounted on a slide of the movable guide rail.

[0032] Specifically, variable traffic signs include a moving guide rail and a traffic sign. The moving guide rail can be laid along the roadside, and the traffic sign can be fixedly installed on a slide on the moving guide rail. Once the dynamic adjustment amount of the variable traffic sign under the current environmental conditions is determined by a deep reinforcement learning model, the moving guide rail can be driven to move the traffic sign according to the position adjustment amount in the dynamic adjustment, thereby adjusting the position of the variable traffic sign. This ensures that drivers can quickly and accurately identify the traffic sign and operate their vehicles safely, thus guaranteeing traffic safety in the diversion area.

[0033] like Figure 2 The diagram shown is a system block diagram of a variable traffic sign adjustment system for icy and snowy weather. The system includes: The perception module is configured to acquire the environmental state vector of the area where the variable traffic sign is located. The decision module is configured to determine the dynamic adjustment amount of the variable traffic sign based on the environmental state vector using a trained deep reinforcement learning model. The adjustment module is configured to adjust the position of the variable traffic sign according to the dynamic adjustment amount.

[0034] Specifically, the adjustment system includes a perception module, a decision-making module, and an adjustment module. The perception module uses monitoring equipment to collect various environmental parameters affecting driver behavior at the location of variable traffic signs and generates corresponding environmental state vectors based on these parameters. The decision-making module inputs the generated environmental state vectors into a pre-trained deep reinforcement learning model, which then determines the dynamic adjustment amount of the variable traffic signs. This dynamic adjustment can include positional adjustments between adjacent signs or angular adjustments of the variable traffic signs in various directions in three-dimensional space. The adjustment module determines the deployment location and spacing of the variable traffic signs based on the dynamic adjustment amounts output by the deep reinforcement learning model, thereby ensuring that drivers can quickly and accurately identify traffic signs and operate their vehicles safely, thus guaranteeing traffic safety in the diversion zone.

[0035] In this embodiment, optionally, the decision module includes a speed limit decision unit, configured to determine the variable speed limit value of the variable traffic sign based on the environmental state vector using a trained deep reinforcement learning model.

[0036] Specifically, the dynamic adjustment output of the decision-making module can include not only the positional adjustment of variable traffic signs, but also the traffic sign information displayed by the variable traffic signs. For example, for variable speed limit signs, the dynamic adjustment can include the variable speed limit value that the variable speed limit sign will display. The adjustment module can adjust the speed limit value displayed by the variable speed limit sign based on the variable speed limit value output by the decision-making module, thereby controlling drivers to drive at appropriate speeds and ensuring traffic safety in the diversion zone.

[0037] In this embodiment, optionally, the decision module includes a training unit configured to train the deep reinforcement learning model using a deep deterministic policy gradient algorithm.

[0038] Specifically, the decision-making module includes a training unit that can simulate vehicle traffic conditions under different road surface friction coefficients, ambient temperatures, icing thicknesses, road visibility, weather conditions, snowfall, traffic flow, and current speed limits in a simulated environment. During training, the environmental state vector for each time step is acquired, and the variable traffic signs are adjusted dynamically according to the simulation strategy. The environmental state vector for the next time step is then generated, and the reward for this action is evaluated using an action evaluation function. Subsequently, experience replay and policy optimization methods are used to update the policy parameters of the policy network in the deep reinforcement learning model, thereby training a system capable of accurately determining the optimal deployment location and information for traffic signs under the current environmental conditions.

[0039] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for adjusting variable traffic signs in icy and snowy weather, characterized in that, include: Obtain the environmental state vector of the area where the variable traffic sign is located; Based on the environmental state vector, the dynamic adjustment amount of the variable traffic sign is determined by a trained deep reinforcement learning model. The position of the variable traffic sign is adjusted according to the dynamic adjustment amount.

2. The method for adjusting variable traffic signs in icy and snowy weather according to claim 1, characterized in that, The environmental state vector includes: road surface friction coefficient, ambient temperature, ice thickness, road visibility, weather level, snowfall, traffic flow and / or current speed limit.

3. The method for adjusting variable traffic signs in icy and snowy weather according to claim 1, characterized in that, Also includes: Based on the environmental state vector, the variable speed limit of the variable traffic sign is determined by a trained deep reinforcement learning model, and the speed limit of the variable traffic sign is adjusted according to the variable speed limit.

4. The method for adjusting variable traffic signs in icy and snowy weather according to claim 1, characterized in that, The action evaluation function of the deep reinforcement learning model includes: A safety function is constructed based on the current speed limit, road surface friction coefficient, ambient temperature, ice thickness, road visibility, weather level, snowfall, and traffic flow. A traffic efficiency function is constructed based on vehicle flow, maximum road capacity, and road segment design speed. The control cost function is constructed based on the current speed limit and dynamic adjustment amount.

5. The method for adjusting variable traffic signs in icy and snowy weather according to claim 1, characterized in that, The decision network of the deep reinforcement learning model is trained using a deep deterministic policy gradient algorithm.

6. The method for adjusting variable traffic signs in icy and snowy weather according to claim 1, characterized in that, The variable traffic sign includes a traffic sign and a movable rail, with the traffic sign mounted on a slide of the movable rail.

7. A variable traffic sign adjustment system for icy and snowy weather, characterized in that, include: The perception module is configured to acquire the environmental state vector of the area where the variable traffic sign is located. The decision module is configured to determine the dynamic adjustment amount of the variable traffic sign based on the environmental state vector using a trained deep reinforcement learning model. The adjustment module is configured to adjust the position of the variable traffic sign according to the dynamic adjustment amount.

8. The variable traffic sign adjustment system for icy and snowy weather according to claim 7, characterized in that, The decision-making module includes a speed limit decision unit, configured to determine the variable speed limit value of the variable traffic sign based on the environmental state vector using a trained deep reinforcement learning model.

9. The variable traffic sign adjustment system for icy and snowy weather according to claim 7, characterized in that, The decision module includes a training unit configured to train the deep reinforcement learning model using a deep deterministic policy gradient algorithm.

Images