A method for evaluating laser positioning reliability, chip and robot

Abstract

The application discloses a laser positioning confidence evaluation method, a chip and a robot, S1: acquiring laser points participating in laser positioning, randomly selecting a laser point, and dividing a grid area based on the position of the laser point on a probability grid map; S2: calculating the probability value of the laser point on the probability grid map through the information of the grid area; S3: acquiring the detection distances of all the laser points participating in laser positioning acquired by a laser radar, and then screening the laser points to acquire the number value of the screened laser points; S4: acquiring a probability weighted average value through the probability values of all the laser points participating in laser positioning on the probability grid map and the detection distances; and S5: obtaining the confidence of this positioning based on the probability weighted average value, the number value of all the laser points participating in laser positioning, the number value of the screened laser points and the number value of the laser points participating in laser positioning.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of intelligent robots, and relates to a laser positioning confidence evaluation method, a chip and a robot. BACKGROUND

[0002] When a laser radar robot is running, laser frames are used for positioning. Laser frames obtained at different moments and in different positions have different confidence after positioning. For example, when a small range is moved, the obtained laser frames change little, and the confidence should gradually increase with time; when a new scene is explored, the robot continuously enters a new space, and the laser frames change greatly, so the confidence of positioning should be reduced. Meanwhile, some sensors (such as a gyroscope) in the robot need laser positioning information for error correction, map information labeling and the like. If the confidence of positioning is not evaluated, the positioning will be inaccurate. SUMMARY

[0003] To solve the above technical defects, the application discloses a laser positioning confidence evaluation method, a chip and a robot. The application evaluates and calculates the confidence by combining laser point distribution, grid hitting, laser point quantity and the like, and provides multiple parameters for targeted and flexible optimization, so that the evaluation accuracy of the confidence is improved. The specific technical scheme is as follows.

[0004] A laser positioning confidence evaluation method, the method comprising the following steps: S1: obtaining laser points participating in laser positioning, randomly selecting a laser point, and dividing a grid region based on the position of the laser point on a probability grid map; S2: calculating the probability value of the laser point on the probability grid map through the information of the grid region, repeating steps S1 and S2 until the probability values of all laser points participating in laser positioning on the probability grid map are obtained; S3: obtaining the detection distances of all laser points participating in laser positioning obtained by a laser radar, and then screening the laser points to obtain the number value of the screened laser points; S4: obtaining a probability weighted average value through the probability values of all laser points participating in laser positioning on the probability grid map and the detection distances; and S5: obtaining the confidence of this positioning based on the probability weighted average value, the number value of all laser points participating in laser positioning, the number value of the screened laser points and the number value of the set laser points participating in laser positioning.

[0005] Further, in step S1, the grid region is divided based on the position of the laser point on the probability grid map, comprising the following steps: obtaining the position of the laser point on the probability grid map, and then finding the grid intersection point on the probability grid map closest to the position; then taking the grid intersection point as the center to divide the grid region of n*n grids on the probability grid map; wherein n is a positive integer.

[0006] Further, after the grid area is divided, the probability median is used to fill the grid area located outside the probability grid map.

[0007] Further, the probability value of the laser point on the probability grid map is calculated by the information of the grid area, and a bicubic interpolation method is used, including the following steps: obtaining the distance between each grid in the grid area and the laser point, and then obtaining the corresponding coefficients of the rows and columns in the grid area according to the distance; obtaining the corresponding weight value of each grid through the corresponding coefficients of the rows and columns, and then obtaining the pixel value of the laser point through the weight value by using the summation formula, and then obtaining the probability value corresponding to the pixel value.

[0008] Further, in step S3, the number of screened laser points is obtained by screening the laser points.

[0009] Further, in step S4, the probability weighted average value is obtained by the probability values and the detection distances of all laser points participating in the laser positioning on the probability grid map, including the following formula: A=(S1×T L1 + S2×T L2 +...+ S N ×T LN ) / N;wherein S is the probability value of the laser point, T is the set distance setting base, L is the detection distance of the laser point obtained by the laser radar, and N is the number of laser points participating in the laser positioning.

[0010] Further, in step S5, the confidence of this positioning is obtained based on the probability weighted average value, the number of laser points participating in the laser positioning, the number of screened laser points, and the set number of laser points participating in the laser positioning, including the following formula: C=A×R (N / M-1) ×K (F / N-1) ;wherein R is the laser base weight, K is the hit point number weight, M is the set number of laser points participating in the laser positioning, and F is the number of screened laser points, wherein N is less than or equal to M.

[0011] Further, the robot performs positioning and confidence evaluation by laser data each time.

[0012] A chip for storing a program configured to perform the above-mentioned laser positioning confidence evaluation method.

[0013] A robot equipped with a master control chip, wherein the master control chip is the above-mentioned chip, and the robot is provided with a laser radar for obtaining laser point data.

[0014] Compared with the prior art, the technical scheme of the application calculates the confidence evaluation by combining laser point distribution, grid hit, laser point quantity and the like, and provides multiple parameters for targeted and flexible tuning, improves the evaluation accuracy of the confidence, and also makes the related processing more reasonable and accurate. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a flowchart of a laser positioning confidence evaluation method according to an embodiment of the application;

[0016] Figure 2 is a schematic diagram of a probability grid map according to an embodiment of the application. DETAILED DESCRIPTION

[0017] The specific embodiments of the application will be further described below with reference to the accompanying drawings. It should be pointed out that the following detailed description is exemplary and is intended to provide further description of the application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as generally understood by those skilled in the art to which the application belongs.

[0018] As shown in Figure 1 A laser positioning confidence evaluation method, a laser radar robot uses laser frames for positioning when running. Different numbers of laser frames obtained at different times and different positions should have different confidence after positioning. For example, when moving in a small range, the change in the obtained laser frames is small, and the confidence should gradually increase over time. When exploring a new scene, the robot continuously enters new space, and the change in the laser frames will be large, and the positioning confidence should be reduced. Therefore, it is necessary to evaluate the confidence of laser positioning. The method includes the following steps:

[0019] Step S1: the laser radar robot obtains the laser points participating in laser positioning, randomly selects a laser point, and divides the grid area based on the position of the laser point on the probability grid map. The robot generally performs positioning through a frame of laser data, records the number of laser points after positioning is completed after positioning is completed using laser data, then randomly selects a laser point, and divides the grid area based on the position of the laser point on the probability grid map, including the following steps: obtaining the position of the laser point on the probability grid map, and then finding the grid intersection point on the probability grid map closest to the position; then taking the grid intersection point as the center, dividing the grid area of n*n grids on the probability grid map; wherein n is a positive integer. After the grid area is divided, the grids of the grid area located outside the probability grid map are filled with a probability median, which means that the value corresponding to 50% probability is an obstacle, and 50% probability is empty. For example, if the grid is represented by a 0-1 linear, the corresponding area is the probability of an obstacle, and the probability median is 0.5.

[0020] Step S2: the laser radar robot calculates the probability value of the laser point on the probability grid map through the information of the grid area, repeats steps S1 and S2, and obtains the probability value of all laser points participating in laser positioning on the probability grid map. In the calculation of the probability value of the laser point on the probability grid map through the information of the grid area, a bicubic interpolation method is mainly used. Bicubic interpolation is a complex interpolation method that can create smoother image edges than bilinear interpolation. It is also called bicubic interpolation, which is a method for "interpolating" or increasing the number / density of "pixels" in an image. Generally, interpolation techniques are used to increase graphic data so that the printed area and / or resolution can be increased when it is printed or other forms of output. Including the following steps: obtaining the distance between each grid in the grid area and the laser point, then obtaining the corresponding coefficients of the rows and columns in the grid area according to the distance; obtaining the corresponding weight value of each grid through the corresponding coefficients of the rows and columns, then obtaining the pixel value of the laser point through the weight value using the summation formula, and then obtaining the probability value corresponding to the pixel value. The probability values of the laser points participating in laser positioning on the probability grid map are obtained one by one until the probability values of the laser points participating in laser positioning on the probability grid map are obtained, and the next step is entered.

[0021] Step S3: the laser radar robot acquires the detection distance of all laser points participating in laser positioning acquired by the laser radar, and then screens the laser points to acquire the number value of the screened laser points. In step S3, the robot first acquires the detection distance of the laser points obtained after the laser radar detects the laser points during laser point acquisition, and then screens the laser points to acquire the number value of the screened laser points, acquires the number value of the laser points with a probability value greater than the probability median, and is used for subsequent calculation.

[0022] Step S4: acquire a probability weighted average value through the probability value and detection distance of all laser points participating in laser positioning on the probability grid map. The calculation is mainly performed according to the following formula: A=(S1×T L1 + S2×T L2 +...+ S N ×T LN ) / N; wherein S is the probability value of the laser point, T is the set distance setting weight, L is the detection distance of the laser point acquired by the laser radar, and N is the number value of the laser point participating in laser positioning. The set distance setting weight T is set according to the actual situation and is not limited.

[0023] S5: based on the probability weighted average value, the number value of all laser points participating in laser positioning, the number value of the screened laser points, and the set number value of the laser points participating in laser positioning, the confidence of this positioning is obtained. In step S5, based on the probability weighted average value, the number value of all laser points participating in laser positioning, the number value of the screened laser points, and the set number value of the laser points participating in laser positioning, the confidence of this positioning is obtained, including the following formula: C=A×R (N / M-1) ×K (F / N-1) ; wherein R is the laser point number weight, K is the hit point number weight, M is the set number value of the laser points participating in laser positioning, and F is the number value of the screened laser points, wherein N is less than or equal to M. The robot performs positioning confidence evaluation through laser point data participating in laser positioning each time positioning is performed. According to the confidence, the accuracy of each positioning is judged to select that positioning as the standard for cleaning, mapping, etc.

[0024] A chip for storing a program, the program being configured to perform the above-mentioned laser positioning confidence evaluation method.

[0025] A robot equipped with a master control chip, the master control chip being the above-mentioned chip, and the robot being provided with a laser radar for acquiring laser point data.

[0026] Compared with the prior art, the technical scheme of the application calculates the confidence evaluation by combining laser point distribution, grid hit, and laser point quantity, and provides multiple parameters for targeted and flexible tuning, improves the evaluation accuracy of the confidence, and also makes the related processing more reasonable and accurate.

[0027] As shown in Figure 2 , a laser point is randomly selected from the laser points participating in laser positioning, and then the position point Pi of the laser point on the probability grid map is obtained. Then the grid intersection point Ci closest to the point Pi on the probability grid map is found. A 4*4 grid around the grid intersection point Ci is selected to form a grid area. If the grid area has a grid outside the probability grid map, the probability median or the initial value of the probability grid map is used to fill the grid outside the probability grid map. Then, bicubic interpolation is performed on the 4x4 grid to obtain the probability value Si corresponding to the point Pi on the probability grid map. This method is a conventional technique and will not be described here. Then, the detection distance L of Pi and the laser radar observation point (generally the center of the laser radar) is recorded, which is the detection distance of the laser radar when obtaining the laser point data. The number value F of the power supply whose probability value Si is greater than the probability median is recorded. The number value M of the laser points participating in laser positioning is set, M is the number of laser points participating in positioning in an ideal case, and is set according to the performance of the laser head, the performance of the operator, and the algorithm required, and M≥N. The probability weighted average value A of all laser points participating in laser positioning is calculated as A=(S1×T L1 + S2×T L2 +...+ S N ×T LN ) / N. When calculating, the probability value S of the laser point is multiplied by the distance setting weight T, which is a base number and the detection distance L is an exponential number, and then the average value of all laser point values is obtained to obtain the probability weighted average value A. Then, the confidence C is calculated by the formula C=A×R (N / M-1) ×K (F / N-1) , the confidence C is equal to, the probability weighted average value A is multiplied by, the laser point number weight R is a base number and (N / M-1) is an exponential number, and then multiplied by the hit point number weight K, which is a base number and (F / N-1) is an exponential number. Among them, the distance setting weight T, the laser point number weight R and the hit point number weight K, wherein the number of laser points is the number of laser points participating in laser positioning, the number of hit points is the number of laser points obtained in a frame, and is a set value, which can be flexibly tuned by T, R and K.

[0028] The above merely describes preferred embodiments of the present application, but is not intended to limit the present application in other forms, and any skilled person in the art can modify or change the above disclosed technical contents into equivalent embodiments with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiments without departing from the technical solution of the present application and according to the technical essence of the present application still falls within the protection scope of the present application.

Claims

1. A method of evaluating laser position fix reliability, characterized by, The method comprises the following steps: S1: acquiring laser points participating in laser positioning, randomly selecting a laser point, and dividing a grid area based on the position of the laser point on the probability grid map; S2: calculating the probability value of the laser point on the probability grid map through the information of the grid area, repeating steps S1 and S2 until the probability values of all laser points participating in laser positioning on the probability grid map are acquired; S3: acquiring the detection distances of all laser points participating in laser positioning acquired by the laser radar, and then screening the laser points to acquire the number value of the screened laser points; S4: acquiring a probability weighted average value through the probability values of all laser points participating in laser positioning on the probability grid map and the detection distances; S5: obtaining the confidence of this positioning based on the probability weighted average value, the number value of all laser points participating in laser positioning, the number value of the screened laser points, and the number value of the set laser points participating in laser positioning; In step S3, the number value of the screened laser points is acquired by screening the laser points to acquire the number value of the screened laser points. In step S4, the probability weighted average value is acquired through the probability values of all laser points participating in laser positioning on the probability grid map and the detection distances, including the following formula: A = (S1 x T L1 + S2 x T L2 +... + S N x T LN ) / N; Wherein, S is the probability value of the laser point, T is the distance set base, L is the detection distance of the laser point acquired by the laser radar, and N is the number value of the laser points participating in laser positioning. In step S5, the confidence of this positioning is obtained based on the probability weighted average value, the number value of all laser points participating in laser positioning, the number value of the screened laser points, and the number value of the set laser points participating in laser positioning, including the following formula: C = A x R (N / M-1) x K (F / N-1) ; Wherein, R is the laser base weight, K is the hit point number weight, M is the number value of the set laser points participating in laser positioning, and F is the number value of the screened laser points, wherein N is less than or equal to M.

2. The method of claim 1, wherein, In step S1, the grid area is divided based on the position of the laser point on the probability grid map, including the following steps: The position of the laser point on the probability grid map is acquired, and the nearest grid intersection point on the probability grid map is found; Then, the grid area of n*n grids is divided on the probability grid map with the grid intersection point as the center; wherein n is a positive integer.

3. The method of claim 2, wherein, After the grid area is divided, the grid outside the probability grid map is filled with the median probability value.

4. The method of claim 1, wherein, In the calculation of the probability value of the laser point on the probability grid map through the information of the grid area, a bicubic interpolation method is adopted, including the following steps: The distance between each grid in the grid area and the laser point is acquired, and then the corresponding coefficients of the rows and columns in the grid area are acquired according to the distance; The corresponding weight value of each grid is acquired through the corresponding coefficients of the rows and columns, and then the pixel value of the laser point is acquired through the weight value by using the summation formula, and then the probability value corresponding to the pixel value is acquired.

5. The method of claim 1, wherein, The robot performs positioning through a frame of laser data, and confidence assessment is performed through the laser points in the frame.

6. A chip for storing a program, characterized by The program is configured to perform the laser positioning reliability evaluation method of any one of claims 1-5.

7. A robot equipped with a master control chip, characterized by, The master chip is the chip of claim 6, and the robot is provided with a laser radar for acquiring laser point data.

Images