A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle

Abstract

The present invention discloses a method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle, including: 1. Calculating the initial running speed of the unmanned vehicle; 2. Calculating the acceleration of the unmanned vehicle's movement; 3. Inserting n sub-nodes between two adjacent nodes; 4. Calculating the corresponding running speed of the unmanned vehicle at each sub-node and assigning a value to the speed variable applied to the model; 5. Accumulating the distances between all sub-nodes according to the speed of each sub-node and the model update period between two adjacent nodes, and comparing the accumulated result with the actual distance between the two nodes to obtain the difference value; 6. Compensating for the error; 7. Repeating step 3 to realize the 3D numerical simulation of the running state of the unmanned vehicle. The present invention replaces the traditional average speed simulation with a variable-speed motion simulation with higher accuracy between two nodes; eliminates the poor visual effects of sudden jumps or jams in the running speed that may occur in the "constant speed per step" processing method, and has higher simulation accuracy.

Description

Technical Field

[0001] The present invention belongs to the fields of 3D modeling and digital twin, and particularly relates to a method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle. Background Art

[0002] With the rapid advancement of the digital transformation in various industries, the construction of digital factories based on 3D digital twin has been increasingly emphasized, rapidly popularized and applied.

[0003] Currently, during the development of 3D digital twin systems, for the simulation of the motion state of moving objects, the accuracy (i.e., the degree of approximation of the model to the actual object's motion state in terms of speed change) is not yet high enough. In specific application scenarios, the requirements for simulation accuracy are not high either, mostly remaining at the basic perception of people's vision, that is, as long as there is no obvious abnormality. The specific technical method is that in the application, for each segment of the model's movement from the current position (node) to the next position (node), the coordinates of the next position are assigned to the model object, along with the running speed from the current position coordinates to the next position coordinates. Between two adjacent position nodes, although the running speed of the actual object may change, the running speed of the model is a constant average speed, resulting in a deviation in the motion state between the model and the actual object. This error is not only manifested in the degree of approximation to the actual situation, but also the virtual object will have a poor visual effect of small jumps or freezes.

[0004] The above-mentioned error can be reduced by shortening the position node spacing. Theoretically speaking, if the position node spacing can be divided small enough, the above-mentioned error will also become very small accordingly. However, in the real world, the size of the position node spacing is related to the detection method of the actual object's position information, the data transmission method, and the data reception and storage method, etc. Under the condition that the running speed of the unmanned vehicle is the same, the position node spacing is completely determined by the time node spacing, that is, for the model application, the "time granularity" of the data (i.e., the data acquisition frequency) determines the size of the model simulation accuracy. The smaller the time granularity, the higher the simulation accuracy.

[0005] Due to the comprehensive influence of the above-mentioned various factors, the time granularity of the data is limited. Currently, the smallest time granularity reported in the industry is about 200 milliseconds. And due to different technical means and application scenarios adopted, in many similar applications, the smallest time granularity may be 400 milliseconds, 600 milliseconds, 1000 milliseconds, etc. In the examples of this article, the actual time granularity encountered is as long as 3000 milliseconds and 4000 milliseconds.

[0006] Under the limitation that the data granularity cannot be too small, the processing method of taking the average value between two adjacent nodes at each node for the moving speed of the virtual object (model) to perform uniform motion, that is, using the average speed to approximately simulate variable-speed motion, has a large error. For project demonstrations with low requirements, it can meet the requirements. The problem is that the motion state shown by its model is quite different from the actual motion state, which is likely to cause illusions to users. For applications that need to verify the motion state with the model, the simulation accuracy under this processing method cannot meet the application requirements. Summary of the Invention

[0007] Object of the Invention: The object of the present invention is to provide a method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle. Between two nodes (each step of the model movement), a variable-speed motion simulation with higher accuracy is used to replace the traditional average speed simulation with lower accuracy, eliminating the poor visual effects of sudden jumps or freezes in the running speed that may occur in the "constant speed per step" processing method, and having higher simulation accuracy.

[0008] Technical Solution: A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle according to the present invention, based on the data of the initial node of the target unmanned vehicle, completes the full-time real-time 3D numerical simulation of the running state of the unmanned vehicle by performing the following steps, including the following steps:

[0009] Step 1: Through the data of the initial node, approximately calculate the initial running speed of the unmanned vehicle.

[0010] Step 2: Based on the initial running speed of the unmanned vehicle and the distance to the next node, approximately calculate to obtain the acceleration of the unmanned vehicle's movement.

[0011] Step 3: Insert n sub-nodes between two adjacent nodes. The simulation accuracy is proportional to the number of sub-nodes, but the number of inserted sub-nodes n is limited by the system update frequency setting, the size of the 3D digital twin application, and computer performance factors.

[0012] Step 4: At each sub-node, calculate the corresponding running speed of the unmanned vehicle based on the approximately calculated acceleration of the unmanned vehicle's movement, and assign a speed variable to the model application.

[0013] Step 5: Accumulate the speed of each sub-node and the model update period DeltaUpdate between two adjacent nodes to obtain the total distance between all sub-nodes, and compare the accumulated result with the actual distance between the two nodes to obtain the difference, which is the intermediate temporary position error of the simulation algorithm.

[0014] Step 6: Incorporate the intermediate temporary position error of the simulation algorithm into the acceleration calculation of the next step, thereby compensating for the above intermediate temporary position error by adjusting the magnitude of the running acceleration of the next step.

[0015] Step 7: Loop back to Step 3. If the 3D digital system is not intentionally shut down, this loop will continue indefinitely, thereby achieving real-time 3D numerical simulation of the unmanned vehicle operation state at all times.

[0016] Furthermore, Step 1 is specifically as follows: The acquisition and transmission of the real-time position data of the unmanned vehicle are carried out at equal time intervals, denoted as DeltaDA , and first calculate the relative displacement between the current position of the unmanned vehicle and its DeltaDA previous position , in meters; then calculate the average speed of the unmanned vehicle at the point (i - 1, 0) , in m / s:

[0017] .

[0018] Furthermore, Step 2 is specifically as follows: Set the acceleration during the movement between two adjacent large nodes to be constant, and calculate the acceleration a i of the unmanned vehicle, in m / s 2 :

[0019] ;

[0020] ;

[0021] Among them, represents the X-axis coordinate of the next node that the unmanned vehicle is to reach during operation, represents the current node that the unmanned vehicle has reached during operation, represents the acceleration of the unmanned vehicle at the current point, represents the time variable.

[0022] Furthermore, Step 4 is specifically as follows:

[0023] Calculate the speed V i,0 of the unmanned vehicle at the point (i, 0), in m / s:

[0024] V i,0 = V i-1,0 + a i * t = V i-1,0 + a i * DeltaDA

[0025] Calculate the speed at the intermediate node between two adjacent large nodes , in m / s:

[0026] ;

[0027] Among them, CountUpdate represents the count of the number of 3D model updates within each time period of DeltaDA.

[0028] Further, step 5 is specifically as follows: For the speed of each intermediate node corresponding to step 5, calculate successively:

[0029] X i,j = X i,j-1 + V i,j * DeltaUpdate, j = 1, 2, 3,…, n

[0030] where X i,j is the position coordinate mark of the vehicle, the subscript variable i represents the main node number, j represents the sub-node number, and X i,j-1 represents the position coordinate mark of the vehicle at the previous sub-node;

[0031] Assign values to the parameters defining the next running state of the 3D model:

[0032] Substitute the speed at the intermediate node between two adjacent major nodes and the position coordinate mark X i,j of the vehicle into the corresponding parameters of the 3D digital twin application, and the definition of the motion state of the model at the next time step DeltaUpdate is completed, and the motion state of the model changes immediately accordingly;

[0033] Calculation of the intermediate error d:

[0034] d = X i+1,0 - X i,n+1

[0035] In the formula, X i+1,0 is the position coordinate of the driverless vehicle obtained by data acquisition, while X i,n+1 is the position coordinate calculated according to the above formula; if this value d is greater than zero, it means that the gap between the position of the model and the actual position is d, that is, the model still needs to move a distance represented by d to reach the actual position; if it is negative, it indicates that the 3D model has moved too far. In both cases, corresponding compensation needs to be made in the model motion algorithm.

[0036] Further, step 6 is specifically as follows: After the intermediate node speed and coordinate data of the current round are calculated, incorporate the intermediate error d into the calculation of the next round of acceleration a i+1 i.e.:

[0037] ; ;

[0038] The acceleration a i+1 calculated from the above formula is substituted into the calculation of step 3, and then the parameters V i+1,j and X defining the next motion state of the 3D driverless vehicle model are calculated accordinglyi+1,j , and cycle downwards like this, thus completing the 3D digital twin of the entire unmanned vehicle's motion state.

[0039] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: The present invention has a detection mechanism for quickly detecting the initial motion state (initial velocity, acceleration) through initial data; between two nodes (each step of the model motion), a more accurate variable-speed motion simulation is used to replace the traditional lower-precision average velocity simulation; the simulation position error is compensated in a timely manner at each step of the motion, avoiding the accumulation of errors; the bad visual effects of sudden jumps or lags in the running speed that may occur in the "constant speed per step" processing method are eliminated, and the simulation accuracy is higher; the present invention creates a new theoretical method for simulating the motion of objects in the field of 3D digital twins and a system integration method for applying this theoretical method to an actual 3D digital twin system; due to the significantly improved simulation accuracy of the motion speed of this invention, the digital twin system established based on this technology can be applied to the technical verification of the functions and performances of intelligent devices (such as intelligent vehicles, etc.); by applying the technology of the present invention, the deficiencies in data acquisition speed (data granularity) and data transmission speed (network speed) can be made up in different digital twin application scenarios; the technical idea of this invention can be used for reference and extended to 3D simulations of circular motion or other curvilinear motions, thereby exerting its greater invention value. Description of the Drawings

[0040] Figure 1 It is a schematic diagram of the model of the present invention;

[0041] Figure 2 It is the actual running coordinate trajectory diagram of the vehicle of the present invention;

[0042] Figure 3 It is the actual running coordinate trajectory diagram of the vehicle of the present invention (for a small time period);

[0043] Figure 4 It is the speed change diagram of the unfitted model of the present invention;

[0044] Figure 5 It is the speed change diagram of the model of the present invention after fitting;

[0045] Figure 6 It is the running coordinate trajectory diagram of the unmanned vehicle after the model of the present invention is simulated;

[0046] Figure 7 It is the 3D digital twin diagram of the unmanned vehicle. Detailed Embodiments

[0047] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

[0048] A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle includes the following steps:

[0049] Step 1: Based on the data of the initial node, preliminarily approximate and calculate the initial running speed of the unmanned vehicle;

[0050] Step 2: Approximate and calculate based on the initial running speed of the unmanned vehicle and the distance to the next node to obtain the acceleration of the unmanned vehicle's movement;

[0051] Step 3: Insert n sub-nodes between two adjacent nodes. The simulation accuracy is proportional to the number of sub-nodes, but the number of inserted sub-nodes n is limited by the system update frequency setting, the size of the 3D digital twin application, and the computer performance factor;

[0052] Step 4: Calculate the corresponding running speed of the unmanned vehicle according to the acceleration of the unmanned vehicle's movement obtained by approximation at each sub-node, and assign a value to the speed variable of the model application;

[0053] Step 5: Accumulate the speed of each sub-node and the model update period DeltaUpdate between two adjacent nodes to obtain the total distance of all sub-nodes, and compare the accumulated result with the actual distance between the two nodes to obtain the difference, which is the intermediate temporary position error of the simulation algorithm;

[0054] Step 6: Incorporate the intermediate temporary position error of the simulation algorithm into the next acceleration calculation, so as to compensate for the above intermediate position temporary position error by adjusting the magnitude of the next running acceleration;

[0055] Step 7: Loop back to Step 3. Unless the system application is closed, this loop will continue until the real-time 3D numerical simulation of the running state of the unmanned vehicle is achieved throughout the whole time period.

[0056] I. Notation Explanation:

[0057] For the convenience of describing this algorithm, the following notation explanations are made:

[0058] 1) The acquisition and transmission of the real-time position data of the unmanned vehicle are carried out at equal time intervals. For example, it is acquired and transmitted once every 200 milliseconds. Notated as DeltaDA;

[0059] 2) The action of the digital twin model always lags behind the actual action by DeltaDA. For example, it lags behind the time of one data point, which is 200 milliseconds;

[0060] 3) The update time interval of the digital twin system is DeltaUpdate, for example, 20 milliseconds (which can be set);

[0061] 4) Count the number of 3D model updates, CountUpdate, within each time period of DeltaDA;

[0062] 5) Use the horizontal X-axis as the coordinate system to describe the position of the vehicle during operation, as shown in the appendix; Figure 1 as shown;

[0063] 6) Select the movement in the same direction for algorithm description, that is, in the current coordinate system, X i+1,0 > X i,0 . The discussion of the opposite direction is completely similar;

[0064] 7) The method of marking the position coordinates of the vehicle, X i,j , where the subscript variable i represents the main node number and j represents the sub-node number, refer to the appendix; Figure 1 ;

[0065] 8) Time variable, t;

[0066] 9) Node coordinate position error, d

[0067] 10) The number of intermediate position nodes, n. Based on the above node division method, the following relationship exists: X i,n+1 = X i+1,0

[0068] 11) Timestamp of the unmanned vehicle position data acquisition, T i

[0069] The position point (i, j) is the current point of the model. At this point, the 3D model parameters are to be assigned the motion parameters (target coordinates and speed)

[0070] II. Algorithm Description

[0071] The following is a step-by-step description of the algorithm content.

[0072] Step 1: Calculate the relative displacement, m, of the unmanned vehicle's current position and the position before time DeltaDA:

[0073] X i – X i-1

[0074] Step 2: Calculate the average speed of the unmanned vehicle at point (i - 1, 0), m / s:

[0075]

[0076] Step 3: Assume that the acceleration during the movement between two adjacent large nodes (points with filling in the appendix) is constant, and calculate the acceleration a Figure 1 of the unmanned vehicle's operation, m / s i , m / s2 :

[0077]

[0078]

[0079] Step 4: Calculate the speed of the driverless vehicle at point (i, 0), m / s:

[0080] V i,0 = V i-1,0 + a i * t = V i-1,0 + a i * DeltaDA

[0081] Step 5: Calculate the speed at the intermediate node between two adjacent major nodes, m / s:

[0082] , j = 1, 2, 3, …, n.

[0083] Step 6: For the speed of each intermediate node corresponding to Step 5, calculate successively:

[0084] X i,j = X i,j-1 + V i,j * DeltaUpdate, j = 1, 2, 3, …, n

[0085] Step 7: Assign values to the parameters defining the next running state of the 3D model:

[0086] Substitute the V i,j and X i,j values calculated in Steps 5 and 6 into the corresponding parameters of the 3D digital twin application, thus completing the definition of the motion state of the model at the next time step (DeltaUpdate). The motion state of the model changes immediately accordingly.

[0087] Step 8: Calculate the intermediate error:

[0088] d = X i+1,0 - X i,n+1

[0089] where X i+1,0 is the position coordinate of the driverless vehicle obtained by data acquisition, and X i,n+1 is the position coordinate calculated according to the formula in Step 6. Therefore, if this value (d) is greater than zero, it means that the difference between the position of the model and the actual position is d, that is, the model still needs to move a distance represented by d to reach the actual position. If it is negative, it indicates that the 3D model has moved too far. In both cases, corresponding compensation needs to be made in the model motion algorithm.

[0090] Step 9, Error Compensation:

[0091] After the current round of intermediate node data (speed and coordinates) is calculated, the error d obtained from Step 8 will be incorporated into the calculation of the next round of acceleration a, i.e.: i+1 That is:

[0092]

[0093]

[0094] The acceleration a calculated from the above formula i+1 , substituting into the calculation of Step 4, then the corresponding motion state parameters V i+1,j and X i+1,j of the defined 3D model (driverless vehicle) for the next step are calculated. By looping downwards in this way, the 3D digital twin of the entire motion state of the driverless vehicle is completed.

[0095] As Figure 1 shown, Figure 1 it shows the coordinate schematic diagram of the driverless vehicle at each moment during one-way movement. The dark filled points correspond to each data point collected, and the order is represented by the subscript variable i; the unfilled points represent the intermediate interpolation points, and their order is represented by the subscript variable j.

[0096] Figure 2 It shows the change of the actual position of the actual driverless vehicle running over time. The abscissa represents time (s), and the ordinate represents the coordinate position of the vehicle.

[0097] Figure 3 It is the change of the coordinate position of the driverless vehicle over time for a small time period intercepted for the description of the model method.

[0098] Figure 4 It shows the change of the running speed of the driverless vehicle shown by the model if it is not processed according to the model fitting calculation of this application. The biggest feature and also the biggest drawback of this method is that the running speed of the model changes in a stepped manner, so there will be a jump mutation in the running speed of the model at each node, giving people an illusion of jumping or jamming visually.

[0099] Figure 5 It shows the change of the running speed of the model after being fitted by the algorithm of this model. The black points are the inserted sub-nodes. Therefore, the "jumping" phenomenon of the running speed is completely eliminated after being processed by the model, and the change of the speed over time is also closer to the change of the actual running speed of the vehicle.

[0100] Figure 6 It shows the change of the coordinate position of the driverless vehicle running over time after being fitted by the algorithm of this model, compared withFigure 3 In comparison, its running trajectory becomes smoother and more realistic.

[0101] Figure 7 The figure shows a 3D digital twin system of an unmanned vehicle created by applying the algorithm of this patent. The unmanned vehicle on the twin system runs smoothly, achieving very good simulation accuracy and visual display effects.

Claims

1. A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle, characterized in that, Based on the data of the initial node of the target driverless vehicle, the full-time real-time 3D numerical simulation of the running state of the driverless vehicle is completed by performing the following steps, including the following steps: Step 1: Based on the data of the initial node of the driverless vehicle, the initial running speed of the driverless vehicle is preliminarily approximated and calculated; Step 2: Based on the initial running speed of the driverless vehicle and the distance to the next node, an approximation calculation is performed to obtain the acceleration of the driverless vehicle's movement; Step 3: Insert n sub-nodes between two adjacent nodes. The simulation accuracy is proportional to the number of sub-nodes. The number of inserted sub-nodes n is limited by the system update frequency setting, the size of the 3D digital twin application, and the computer performance factor; Step 4: At each sub-node, based on the approximated acceleration of the driverless vehicle's movement obtained, the corresponding running speed of the driverless vehicle is calculated, and the speed variable is assigned to the model application; Step 5: Accumulate the speed of each sub-node and the model update period DeltaUpdate between two adjacent nodes to obtain the total distance between all sub-nodes, and compare the accumulated result with the actual distance between the two nodes to obtain the difference, which is the intermediate temporary position error of the simulation algorithm; Step 6: Incorporate the intermediate temporary position error of the simulation algorithm into the next acceleration calculation, so as to compensate for the intermediate temporary position error by adjusting the magnitude of the next running acceleration; Step 7: Repeat the loop to step 3 to achieve the full-time real-time 3D numerical simulation of the running state of the driverless vehicle.

2. A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle according to claim 1, characterized in that, Step 1 is specifically as follows: The acquisition and transmission of the real-time position data of the driverless vehicle are carried out at equal time intervals, denoted as DeltaDA. First, calculate the relative displacement X of the current position of the driverless vehicle and the position before time DeltaDA. i -X i-1 , with the unit of m; then calculate the average speed V of the driverless vehicle at point (i - 1, 0). i-1,0 , m / s:

3. A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle, according to claim 2, characterized in that, Step 2 specifically is: Set the acceleration during movement between two adjacent large nodes to be constant, and calculate the acceleration a of the driverless vehicle i , m / s 2 : Among them, X i+1,0 represents the X-axis coordinate of the next node that the driverless vehicle is going to reach, and X i,0 represents the current node that the driverless vehicle has reached. a i represents the acceleration of the driverless vehicle at the current point, and t represents the time variable.

4. A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle according to claim 3, characterized in that, Step 4 is specifically: Calculate the speed V of the driverless vehicle at point (i, 0) i,0 , m / s: V i,0 = V i-1,0 + a i * t = V i-1,0 + a i * DeltaDA Calculate the velocity V at the intermediate node between two adjacent major nodes i,j , m / s: Among them, CountUpdate represents the 3D model update times count within each time period of DeltaDA.

5. A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle according to claim 4, characterized in that, Step 5 is specifically: For the speed of each intermediate node corresponding to step 5, calculate in sequence: X i,j = X i,j-1 + V i,j *DeltaUpdate, j = 1, 2, 3, …, n Among them, X i,j is the position coordinate mark of the vehicle. The subscript variable i represents the main node number, and j represents the branch node number. X i,j-1 represents the position coordinate mark of the vehicle at the previous branch node; Assign values to the parameters defining the next running state of the 3D model: The speed V at the intermediate node between two adjacent major nodes i,j and the position coordinate mark X of the vehicle i,j are substituted into the corresponding parameters of the 3D digital twin application, thus completing the definition of the motion state of the model at the next time step DeltaUpdate, and the motion state of the model immediately changes accordingly; Calculation of the intermediate temporary position error d: d = X i+1,0 -X i,n+1 Where X i+1,0 is the position coordinate of the unmanned vehicle obtained by data acquisition, while X i,n+1 is the position coordinate calculated according to the formula; if the intermediate temporary position error d is greater than zero, it means that the gap between the position of the model and the actual position is d, that is, the model still needs to move the distance represented by d to reach the actual position; if it is negative, it indicates that the 3D model has moved too far. In both cases, corresponding compensation needs to be carried out in the model motion algorithm.

6. A method for improving the control accuracy of the running speed of a 3D digital twin unmanned vehicle, according to claim 5, characterized in that, Step 6 specifically is: After the speed and coordinate data of the intermediate nodes in the current round are calculated, incorporate the intermediate temporary position error d into the calculation of the acceleration a in the next round, that is: i+1 Namely: The acceleration a calculated from the above formula i+1 , substituting into the calculation in step 3, the corresponding motion state parameters V of the defined 3D driverless vehicle model for the next step are calculated i+1,j and X i+1,j , and so on in a loop, thus completing the 3D digital twin of the entire motion state of the driverless vehicle

Images