Vehicle-road collaborative network communication method based on real-time navigation simulation

Abstract

The invention relates to a vehicle-road cooperation network communication method based on real-time navigation simulation. The method comprises the steps of building a vehicle-road cooperation network communication system, fusing to form a three-dimensional map, forming a vehicle-mounted driving route simulation animation, and giving a driving prompt message strategy based on vehicle-mounted and cloud computing power. According to the vehicle-road collaborative network communication method, the traditional V2X technology is jumped, the vehicle-mounted system and the cloud server are adopted to respectively undertake computing power intelligent modeling, and network intercommunication of position information between portable equipment of vulnerable traffic participants and traditional roadside facilities is realized based on a three-dimensional map; according to the invention, additional roadside facilities are not required to be arranged to upload data to the server, the computing power burden is increased, the driving prompt information is determined through vehicle-mounted and cloud dual-mode calculation, and dual-mode, efficient, low-cost and accurate vehicle-road cooperative network communication is realized.

Description

Technical Field

[0001] This invention relates to a communication method for vehicle-road cooperative networks, and particularly to a communication method for vehicle-road cooperative networks based on real-time navigation simulation. Background Technology

[0002] Existing vehicle-to-everything (V2X) technologies, which enable vehicle-to-infrastructure (V2X) information collaboration, intelligent traffic safety, and real-time management, require the participation of roadside units (RSUs), on-board units (OBUs), vulnerable road users, and roadside infrastructure. The resulting massive amounts of data place high demands on cloud and satellite analytics capabilities. With the increasing number of vehicles and the continued existence of complex traffic conditions, computing power and architectural costs remain persistent challenges for researchers.

[0003] In the field of automotive systems, automakers have consistently emphasized chip upgrades. However, aside from improving user interface controls and addressing the still somewhat unclear role of automotive chips, they haven't fully utilized them. Therefore, offloading all or part of the computing power to automotive chips essentially creates independent mobile computing power for each vehicle. This aims to leverage existing 3D map technology to create simulations, replacing cumbersome and costly broadcasting methods. This would achieve a "three-self" vehicle-road cooperative solution, fully utilizing the self-simulation, self-identification, and self-alert capabilities of the automotive chip. In other words, from the perspective of each individual vehicle, roadside information acquisition takes center stage, with the cloud and satellite becoming secondary, providing timed assistance or correction for the "three-self" solution. Consequently, besides traffic lights, traffic signs, and barriers, other roadside facilities would be unnecessary.

[0004] This is because roadside facilities are fixed in location, and what they need to sense are moving road targets. When the moving targets achieve self-computing capabilities, these roadside facilities only need to provide parameters in real time. Once these parameters are set, as long as they are not shut down, especially traffic lights, their operation follows a basically unchanging time pattern. Therefore, vehicles only need to know the initial setting parameters and the updated information to obtain and adjust the status information of the roadside facilities.

[0005] Therefore, the question is how to integrate 3D mapping technology, simulation technology, and artificial intelligence prediction technology to form a new vehicle-road cooperative technology that reduces the burden of public computing power, lowers the cost of public facilities, and fully enhances the potential of vehicle-mounted equipment. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, this invention provides a vehicle-road cooperative network communication method based on real-time navigation simulation, comprising the following steps: S1 establishes a vehicle-road cooperative network communication system, including vehicles equipped with onboard systems, roadside devices, personal devices for vulnerable road users, cloud servers, and navigation satellites; the roadside devices upload parameters to the cloud server for storage through parameter settings, and update the saved parameters synchronously when the parameters are updated; S2 acquires remote sensing maps and oblique photogrammetric images via remote sensing satellites and drones, respectively. The fusion of these two data points forms a 3D map, which is stored on a cloud server. The cloud server, based on parameters set in S1 and using navigation satellite positioning of vulnerable road users' personal devices and vehicles, embeds these parameters onto the 3D map and performs simulation rendering. The real-time positioning of vulnerable road users' personal devices, roadside facilities, and vehicles is displayed on the 3D map. A large screen is set up at the location of the cloud server, allowing users to click and search to identify interested vulnerable road users' personal devices, roadside facilities, and vehicles, and to retrieve, view, and update their corresponding positioning and parameters in real time.

[0007] The S3 vehicle system downloads a 3D map and, based on the set driving route, creates a simulation animation between the target location and the current location within a preset time frame, based on the real-time vehicle speed. At the same time, the cloud server receives the current real-time vehicle speed transmitted by the vehicle system and, based on the location of the vulnerable road users' personal devices and the parameters set on the roadside, predicts the prompts that need to be given to the vehicle and sends the prompts to the vehicle system. The S4 vehicle system determines whether a reminder needs to be issued based on the simulation animation and the received prompts.

[0008] Optionally, the specific method for simulating the target position to be reached within a preset time based on the real-time vehicle speed and the current position is as follows: Q1 The vehicle system retrieves the vehicle's historical driving routes and corresponding location and speed information. For each historical driving route, a long short-term memory model is set up, with each unit representing a time node within a different preset future time period. Based on the corresponding location and speed information, the speed information at the corresponding time node is converted. Therefore, multiple speed information at the corresponding time node is collected to perform the task of partially training the model. The time nodes include month and day. The position of different time nodes in the historical driving route during the first training is used as the benchmark. The speed information of subsequent training is based on the speed corresponding to the vehicle reaching the benchmark position.

[0009] It should be emphasized that when driving the same route multiple times, at the same time points—that is, the same month and the same day of each year—speeds exhibit similar characteristic values. This is determined by driving habits, climate, and the established traffic patterns that have formed on this route over time, affecting both transit time and speed. Generally, accumulating 3-5 years of data is sufficient to train a good model.

[0010] The remaining training tasks for Q2 continue to be uploaded to the cloud server. After training is complete, the model will be sent back to the vehicle system. The Q3 in-vehicle system retrieves the current driving path, current speed, and location. It then inputs the current speed into the trained model to predict the speed at each time point. The system uses the predicted speed of the later time point between adjacent time points to create a simulated animation segment of the vehicle's form between those adjacent time points. This process is then used to stitch together the animation segments between all time points.

[0011] Optionally, the preset time is 1 second to 1 minute, and at least part of the simulation animation will be deleted after a preset number of preset times. This saves space.

[0012] Preferably, the preset number is 2-5, and the deletion is to delete all of them, or to delete 2-4 simulation animations arranged in the order of their generation time.

[0013] Optionally, the methods for predicting the prompts that need to be given to the vehicle specifically include: P1 calculates the position of the personal device of vulnerable road users within a radius of 200-300m around the vehicle in the 3D map at the corresponding time of the two previous time nodes based on the current position, and determines whether the vulnerable road user is a pedestrian or a non-motorized vehicle driver based on the position change of the personal device. P2 predicts the location of the personal equipment of vulnerable road users and / or the status of roadside facilities within 50-200m around the vehicle when the vehicle reaches the corresponding time point at the current vehicle speed. First, determine whether the personal equipment of vulnerable road users belongs to pedestrians or non-motorized vehicle drivers within a large area, and then determine the location within a smaller area. No new pedestrians or non-motorized vehicle drivers will be found within the smaller area.

[0014] When the distance between the location of the vehicle at the next time node and the location of the personal device of a vulnerable road user within 50-200m around the vehicle is within the threshold range, P3 will give the first prompt message. When the roadside facility requires the vehicle to slow down or pay attention, and / or the current driving lane is inconsistent with the lane that the vehicle should be driving in, P3 will give the second prompt message.

[0015] The threshold range is 50-70m for pedestrians and 50-80m for non-motorized vehicles; the prediction method when roadside facilities require vehicles to slow down or pay attention is: T1 continuously obtains the distance between the location of roadside facilities and the current vehicle location until the distance is less than 100-150m, at which point T2 is executed. T2 calculates whether the roadside facility is in a state that allows the vehicle to leave at the current speed. If so, no deceleration is required; otherwise, the vehicle is required to decelerate. If the roadside facilities themselves remain in a fixed state and always require vehicles to slow down or pay attention, then the prediction will be that vehicles are required to slow down or pay attention. In fact, for example, at traffic lights, even if the distance is less than 100-150m and you can cross the stop line when the light turns yellow at your current speed, you still need to slow down. However, this is a safety driving skill that drivers should be aware of, and the system does not provide any additional design reminders for this.

[0016] Optionally, the S4 vehicle system determines whether a reminder needs to be executed based on the simulation animation and the received prompt information; The S4-1 cloud server will send the location of the vulnerable road user's personal device to the vehicle system at the next corresponding time point. When the S4-2 vehicle system plays the animation to the next time point, it calculates the distance between the vehicle's position in the animation and the position of the vulnerable road user's personal device in S4-1. When a prompt message is received, if the distance is within the threshold range, the first prompt message is executed; otherwise, the first prompt message is not executed. If no prompt message is received, if the distance is within the threshold range, the first prompt message is executed; otherwise, the first prompt message is not executed. The second prompt message is executed if and only if the vehicle system receives a second prompt message.

[0017] Therefore, based on the simulation animation predicted by the vehicle system model and the cloud service area predicted by the current vehicle speed, the system combines the advantages of both methods to determine whether to execute the prompt message based on the distance to vulnerable road users. If both algorithms indicate that the distance is within the threshold range, the probability of accurate measurement is extremely high. If one algorithm is outside the prediction range, it is not adopted to prevent false prompts, and the driver's judgment is left to decide without disturbing the driver's judgment. This forms a double insurance of intelligent system and human attention.

[0018] The parameters include the precise time when the traffic light is activated, the status of the traffic light when it is activated, the time when the street light is turned on and off, the range of the collision avoidance facility, and traffic signs. For traffic lights, leaving the roadside facility means that the entire vehicle has crossed the stop line and continues to drive. For traffic signs that require deceleration or caution, step T3 is executed.

[0019] Beneficial effects This approach employs a solution where in-vehicle system simulation animation and cloud server intelligent modeling each handle their own computing power. It eliminates the need for additional roadside sensing equipment (such as cameras and speedometers) or traditional V2X equipment. It also includes 3D map-based navigation and personal devices for vulnerable road users. This achieves accurate, dual-mode, efficient, and low-cost vehicle-to-infrastructure (V2I) network communication. Attached Figure Description

[0020] Figure 1 A simplified flowchart of the steps for downloading a 3D map using a vehicle-road cooperative network communication method based on real-time navigation simulation (S1-S3 vehicle system); Figure 2 The diagram illustrates a method for an onboard system to retrieve the vehicle's historical driving routes and predict the target location to be reached within a preset timeframe based on real-time vehicle speed, along with a simulation animation showing the relationship between the target location and the current location. It also illustrates a method for predicting the necessary prompts to be sent to the vehicle. Figure 3 A flowchart illustrating the Q3 steps for generating simulation animations. Detailed Implementation

[0021] Figure 1-3 A vehicle-road cooperative network communication method based on real-time navigation simulation is presented, including the following steps: S1 establishes a vehicle-road cooperative network communication system, including vehicles equipped with onboard systems and roadside installations (such as...). Figure 2 Traffic lights in the middle), vulnerable road users (e.g.) Figure 2 Pedestrian-related personal devices, as well as cloud servers and navigation satellites (such as...) Figure 1 (As shown); The roadside settings are configured with parameters including the precise start time of traffic lights, the status of traffic lights when they are started, the turn-on and turn-off times of streetlights, the range of anti-collision facilities, and traffic signs. These parameters are uploaded to a cloud server for storage and are updated synchronously when the parameters are updated. S2 as Figure 1 As shown, remote sensing maps and oblique photography images are acquired through remote sensing satellites and drones, respectively. The two are fused to form a 3D map, which is stored on a cloud server. The cloud server embeds parameters and performs simulation rendering on the 3D map based on the parameters set in S1 and the personal devices of vulnerable road users and vehicles located by navigation satellites. The positioning of the personal devices of vulnerable road users, roadside facilities, and vehicles is displayed on the 3D map in real time. A large screen is set up at the location of the cloud server, and users can click and search to mark the personal devices of vulnerable road users, roadside facilities, and vehicles of interest, and retrieve, view, and update the corresponding positioning and parameters in real time.

[0022] S3 continues as follows Figure 1As shown, the in-vehicle system downloads a 3D map, and based on the set driving route, it predicts the target location to be reached within a preset time and the current location based on the real-time vehicle speed. At the same time, the cloud server receives the current real-time vehicle speed transmitted by the in-vehicle system, and predicts the prompt information to be given to the vehicle based on the positioning of the vulnerable road users' personal devices and the parameters set on the roadside, and sends the prompt information to the in-vehicle system. The S4 vehicle system determines whether a reminder needs to be issued based on the simulation animation and the received prompts.

[0023] The specific method for simulating the relationship between the target position and the current position within a predetermined time frame based on real-time vehicle speed is as follows: Q1 Figure 2 The vehicle system retrieves the vehicle's historical driving routes and corresponding location and speed information. For each historical driving route, a long short-term memory model is set up, with each unit representing a different time node within the next 3 seconds. Based on the corresponding location and speed information, the speed information at the corresponding time node is converted. Therefore, multiple speed information at the corresponding time node is collected to partially train the model. The time nodes include month and day.

[0024] The specific conversion method is to calculate the time of different positions based on the speed at different positions, so as to find the corresponding speed information at that time. For example, the driving distance is a function of speed S(v), then dS / dv=t, then dS=tdv (1). For a specified time t0, the two sides of formula (1) are integrated to get St0-S0=t0(Vt0-V0), so Vt0= (St0-S0) / t0+V0. Therefore, only by knowing the initial position S0 (the navigation satellite has positioning data sent to the cloud, which can be transmitted to the vehicle system by the cloud server) and the initial speed V0 (the vehicle system has records), and the position St0 of the specified time t0 (obtained in the same way as S0), that is, the vehicle system can calculate the speed Vt0 at time t0.

[0025] like Figure 2 As shown, the reference positions (represented by points) in the driving route formed during the first training are given. Taking the two time points before the current time point and the last time point as examples, the speed information of multiple vehicles arriving at the same reference positions on different months and days (vehicle speed t1, vehicle speed t2, ..., vehicle speed tn) is selected for 60-80% training, while the remaining 40%-20% is trained by the cloud server to balance the computing power between the vehicle system and the cloud server.

[0026] The remaining portion of Q2 (taking 20% ​​as an example) involves training the model, which is then uploaded to the cloud server. After training is complete, the model is sent back to the vehicle system. Figure 2 ); Q3 Figure 3 As shown, the vehicle system retrieves the current driving path, current speed, and location, substitutes the current speed into the trained model, predicts the speed at each time point, and uses the corresponding predicted speed of the later time point between adjacent time points to complete the simulated animation segment of the vehicle form between those adjacent time points, thereby stitching together the animation segments between all time points.

[0027] Figure 3 If the predicted speed from the current node to the next node is Va, then the formal distance d1 between the two nodes is calculated using the speed Va. Similarly, if the predicted speed for the next node is Vb, then the speed for the subsequent travel distance d2 in the graph is calculated using Vb.

[0028] Every three preset time intervals, two simulation animations will be generated in chronological order.

[0029] The methods for predicting the information that needs to be given to the vehicle include: P1 as Figure 2 The figure shows the position of the personal device (e.g., smartphone) of the pedestrian and non-motorized vehicle driver within a 280m radius around the vehicle in the three-dimensional map at the corresponding time points of the previous two time nodes, based on the current position (represented by a long dot and a triangle respectively in the figure), and determines whether the pedestrian or non-motorized vehicle driver belongs to the pedestrian or non-motorized vehicle driver based on the position changes of the personal device of the pedestrian and non-motorized vehicle driver. P2 predicts the location of the personal devices of vulnerable road users within 50m around the vehicle when the vehicle reaches the next time node corresponding to position A based on the current vehicle speed; the figure shows the personal device α of one pedestrian and the personal device β of another pedestrian.

[0030] When P3 reaches the next time node, the distance between position A and the position of a pedestrian's personal device within 50m of the vehicle is within the threshold range of 50m, then a first prompt message is given (e.g., on the vehicle screen). Figure 2 The system will flash two long dots on the road and / or use voice prompts to warn of pedestrians. At this time, there are no roadside facilities within 50m. When the vehicle reaches time point B, if the roadside facilities require the vehicle to slow down, and / or the current driving lane is inconsistent with the lane that the vehicle should be driving in, a second prompt message will be given.

[0031] Specifically, the predictive method for when roadside facilities require vehicles to slow down or pay attention is: T1 retrieves the distance between the location of roadside facilities and the current vehicle position. If this distance is less than 150m, T2 is executed. Figure 2 For the time node corresponding to location B (which belongs to a reference location), the traffic light is within 150m. Based on the current vehicle speed, T2 calculates whether the roadside facility status allows departure when the vehicle leaves at this speed, i.e., whether it is a green light. If so, no deceleration is required; otherwise, the vehicle is required to decelerate. It should be noted that "no deceleration is required" here means that the vehicle does not need to decelerate due to "the condition of the roadside facilities". It is a condition for checking whether a second prompt message is given, rather than actually telling the driver that it is not necessary to decelerate in advance when passing the stop line on a green light.

[0032] If the roadside facilities themselves remain in a fixed state and always require vehicles to slow down or pay attention (e.g., road signs requiring slowing down, such as sharp turns ahead, or warning of school entrances), then the prediction is that vehicles should slow down or pay attention. The S4 vehicle system determines whether a reminder needs to be issued based on the simulation animation and received prompts, including the following: The S4-1 cloud server will be as follows Figure 2 The location of the personal device of the vulnerable road user at the next time node (A) is sent to the vehicle system; When the S4-2 vehicle system plays the animation to the next time point (A), it calculates the vehicle's position in the animation and the pedestrian's personal device in S4-1 at this time. The distance between the location of the zero-person personal device β and the location of the vehicle is determined. When a prompt message is received, if the distance is within 50m, the first prompt message is executed; otherwise, the first prompt message is not executed. If no prompt message is received, if the distance is within 50m, the first prompt message is executed; otherwise, the first prompt message is not executed. The second prompt message is executed if and only if the vehicle system receives a second prompt message.

Claims

1. A vehicle-road cooperative network communication method based on real-time navigation simulation, characterized in that, Includes the following steps: S1 establishes a vehicle-road cooperative network communication system, including vehicles equipped with onboard systems, roadside devices, personal devices for vulnerable road users, cloud servers, and navigation satellites; the roadside devices upload parameters to the cloud server for storage through parameter settings, and update the saved parameters synchronously when the parameters are updated; S2 acquires remote sensing maps and oblique photography images through remote sensing satellites and drones, respectively. The two are fused to form a 3D map, which is stored on a cloud server. The cloud server embeds parameters and performs simulation rendering on the 3D map based on the parameters set by S1 and the personal devices of vulnerable road users and vehicles located by navigation satellites. The positioning of personal devices of vulnerable road users, roadside facilities, and vehicles is displayed on the 3D map in real time. A large screen is set up at the location of the cloud server. By clicking and searching, users can identify personal devices of vulnerable road users, roadside facilities, and vehicles of interest, and retrieve, view, and update the corresponding positioning and parameters in real time. The S3 vehicle system downloads a 3D map and, based on the set driving route, creates a simulation animation between the target location and the current location within a preset time frame, based on the real-time vehicle speed. At the same time, the cloud server receives the current real-time vehicle speed transmitted by the vehicle system and, based on the location of the vulnerable road users' personal devices and the parameters set on the roadside, predicts the prompts that need to be given to the vehicle and sends the prompts to the vehicle system. The S4 vehicle system determines whether a reminder needs to be issued based on the simulation animation and the received prompts.

2. The method according to claim 1, characterized in that, The specific method for simulating the relationship between the target position and the current position within a predetermined time frame based on real-time vehicle speed is as follows: Q1 The vehicle system retrieves the vehicle's historical driving routes and corresponding location and speed information. For each historical driving route, a long short-term memory model is set up, with each unit representing a time node within a different preset future time period. Based on the corresponding location and speed information, the speed information at the corresponding time node is converted. Therefore, multiple speed information at the corresponding time node is collected to perform the task of partially training the model. The time nodes include month and day. The position of different time nodes in the historical driving route during the first training is used as the benchmark. The speed information of subsequent training is based on the speed corresponding to the vehicle reaching the benchmark position. The remaining training tasks for Q2 continue to be uploaded to the cloud server. After training is complete, the model will be sent back to the vehicle system. The Q3 in-vehicle system retrieves the current driving path, current speed, and location. It then inputs the current speed into the trained model to predict the speed at each time point. The system uses the predicted speed of the later time point between adjacent time points to create a simulated animation segment of the vehicle's form between those adjacent time points. This process is then used to stitch together the animation segments between all time points.

3. The method according to claim 2, characterized in that, The preset time is 1 second to 1 minute. After a preset number of preset time intervals, at least part of the simulation animation will be deleted. This saves space.

4. The method according to claim 3, characterized in that, The preset number is 2-5. The deletion refers to deleting all of them, or deleting 2-4 simulation animations arranged in the order of their generation time.

5. The method according to claim 3, characterized in that, The methods for predicting the information that needs to be given to the vehicle include: P1 calculates the position of the personal device of vulnerable road users within a radius of 200-300m around the vehicle in the 3D map at the corresponding time of the two previous time nodes based on the current position, and determines whether the vulnerable road user is a pedestrian or a non-motorized vehicle driver based on the position change of the personal device. P2 predicts the location of the personal devices of vulnerable road users and / or the status of roadside facilities within 50-200m around the vehicle when the vehicle reaches the corresponding time point at the current vehicle speed. When the distance between the location of the vehicle at the next time node and the location of the personal device of a vulnerable road user within 50-200m around the vehicle is within the threshold range, P3 will give the first prompt message. When the roadside facility requires the vehicle to slow down or pay attention, and / or the current driving lane is inconsistent with the lane that the vehicle should be driving in, P3 will give the second prompt message.

6. The method according to claim 4 or 5, characterized in that, The threshold range is 50-70m for pedestrians and 50-80m for non-motorized vehicles; the prediction method when roadside facilities require vehicles to slow down or pay attention is: T1 continuously obtains the distance between the location of roadside facilities and the current vehicle location until the distance is less than 100-150m, at which point T2 is executed. T2 calculates whether the roadside facility is in a state that allows the vehicle to leave at the current speed. If so, no deceleration is required; otherwise, the vehicle is required to decelerate. If the roadside facilities themselves remain in a fixed state and always require vehicles to slow down or pay attention, then the prediction will be that vehicles are required to slow down or pay attention.

7. The method according to claim 6, characterized in that, The S4 vehicle system determines whether a reminder needs to be issued based on the simulation animation and received prompts, including the following: The S4-1 cloud server will send the location of the vulnerable road user's personal device to the vehicle system at the next corresponding time point. When the S4-2 vehicle system plays the animation to the next time point, it calculates the distance between the vehicle's position in the animation and the position of the vulnerable road user's personal device in S4-1. When a prompt message is received, if the distance is within the threshold range, the first prompt message is executed; otherwise, the first prompt message is not executed. If no prompt message is received, the first prompt message is executed if the distance is within the threshold range; otherwise, the first prompt message is not executed. The second prompt message is executed if and only if the vehicle system receives the second prompt message.

8. The method according to claim 7, characterized in that, The parameters include the precise time when the traffic light is activated, the status of the traffic light when it is activated, the time when the street light is turned on and off, the range of the collision avoidance facility, and traffic signs. For traffic lights, leaving the roadside facility means that the entire vehicle has crossed the stop line and continues to drive. For traffic signs that require deceleration or caution, step T3 is executed.

Images