A General Assessment Method for Autonomous Navigation Capabilities Based on Similarity and Hierarchical Analysis

Abstract

This invention discloses a method for evaluating the general capabilities of autonomous navigation systems based on similarity and hierarchical analysis. To achieve an accurate and comprehensive evaluation of the general capabilities of autonomous navigation systems and provide a basis and direction for research and improvement of autonomous navigation technology, the invention includes the following steps: 1. Obtaining target values ​​for evaluating the general capabilities of autonomous navigation systems; 2. Comparing the target values ​​with actual values ​​and calculating the similarity; 3. Evaluating the similarity of the indicators using the hierarchical analysis method. This invention summarizes and generalizes evaluation indicators for the general capabilities of autonomous navigation systems, establishes a method for comparing the similarity between target and actual values, and uses the hierarchical analysis method to fuse the similarity of the indicators. This invention is applicable to the comprehensive evaluation of general capabilities under different scenarios and tasks. The selected indicators can comprehensively describe the navigation performance of autonomous navigation systems in time and space, and can quantitatively provide evaluation results. This invention belongs to the field of autonomous navigation system evaluation technology.

Description

Technical Field

[0001] This invention relates to a general method for evaluating the autonomous navigation capabilities based on similarity and hierarchical analysis, and belongs to the field of autonomous navigation system evaluation technology. Background Technology

[0002] An ideal unmanned system should have the ability to move autonomously to a target area by sensing the external environment with the help of sensors without external intervention. Autonomous navigation technology is the core technology to achieve this goal. Intelligent agents with autonomous navigation capabilities can be collectively referred to as autonomous navigation systems, such as autonomous mobile robots and autonomous unmanned underwater vehicles.

[0003] Because autonomous navigation systems often operate in complex and varied environments and perform a wide range of tasks, such as mobile robots working in complex outdoor environments and underwater vehicles that can operate in different waters, high demands are placed on the versatility of autonomous navigation systems. Therefore, it is very important to conduct an accurate and reasonable performance evaluation of the versatility of autonomous navigation systems.

[0004] However, current research on the evaluation of autonomous navigation systems mainly suffers from the following problems:

[0005] 1. How to quantitatively calculate the various evaluation indicators that reflect the general capabilities of autonomous navigation;

[0006] 2. How to integrate the various evaluation indicators to form the final general competency assessment result;

[0007] Existing performance evaluation methods mostly evaluate static targets, calculating various evaluation indicators by measuring their physical information. However, autonomous navigation systems, as mobile targets, generate a large amount of data in time and space. To comprehensively evaluate their universality, it is necessary to make reasonable use of the path and time data generated during autonomous navigation, which puts forward new requirements for the calculation of evaluation indicators and the construction of evaluation systems. Summary of the Invention

[0008] In order to achieve an accurate and comprehensive assessment of the general capabilities of autonomous navigation systems and provide a basis and direction for the research and improvement of autonomous navigation technology, this invention proposes a method for assessing the general capabilities of autonomous navigation based on similarity and hierarchical analysis.

[0009] The technical solution adopted by the present invention to solve the above problems is as follows:

[0010] This invention includes the following steps:

[0011] Step 1: Obtain the target value for the general assessment of autonomous navigation capabilities;

[0012] Step 2: Compare the target value with the actual value and calculate the similarity.

[0013] Step 3: Evaluate the similarity of indicators using the analytic hierarchy process (AHP).

[0014] Furthermore, in step 1, the target value for the general capability assessment of the autonomous navigation system is the expected target value obtained by the global path planning algorithm under the specified scenario, including path point set, path length, path energy consumption, and path time consumption data.

[0015] Furthermore, in step 2, the index similarity includes consumption similarity, trajectory space similarity, and trajectory projection similarity.

[0016] Furthermore, the similarity calculation is as follows:

[0017] (1) Time consumption ratio

[0018] The ratio of the actual time taken to reach the target location to the ideal time taken to reach the target location is calculated as follows:

[0019]

[0020] Where T L T represents the actual time taken to reach the target. A This represents the ideal time to reach the destination; the closer the ratio is to 1, the less time is spent on path planning.

[0021] (2) Path consumption ratio

[0022] The ratio of the actual distance traveled to the target location to the ideal path distance to the target location is calculated as follows:

[0023]

[0024] Where S L S represents the length of the actual trajectory taken to reach the target. A This represents the length of the trajectory taken to reach the target under ideal conditions; the closer the ratio is to 1, the shorter the path taken by the path planning.

[0025] (3) Energy consumption ratio

[0026] The ratio of the energy consumed to actually reach the target location to the energy consumed to reach the target under ideal conditions is calculated as follows:

[0027]

[0028] Among them, E L E represents the energy consumed to actually reach the target location. AThis represents the energy consumed to reach the target under ideal conditions. The closer the ratio is to 1, the less energy is consumed by the path planning.

[0029] Furthermore, the similarity of the trajectory space is determined using the Framinger distance method.

[0030] Let the pair (S, d) be a metric space, where d is a metric function on S, and let A and B be two continuous curves on S, satisfying:

[0031] A:[0,1]→S,B:[0,1]→S (4)

[0032] The Framing distance between curve A and curve B is defined as shown in equation (5).

[0033]

[0034] In Equation 5, d is a metric function on S, α and β are two reparameterized functions for the unit interval, and F(A,B) is the Framing distance between curves A and B. Let t be a time point, then at that moment, the sampling point on curve A is A(α(t)) and the sampling point on curve B is B(β(t)). In each sampling, t discretely traverses the interval [0,1] to obtain the maximum distance under this sampling.

[0035]

[0036] The trajectory projection similarity is calculated as follows:

[0037] In target detection tasks, the intersection-union ratio (IUGR) is used as a metric to describe the overlap between two bounding boxes, and is used to compare the difference between the actual navigation trajectory and the optimal trajectory. Here, A and B are the areas enclosed by the actual navigation trajectory and the optimal trajectory projected onto the ground and the coordinate axes, respectively. The IUGR is the area of ​​the area where A and B intersect, divided by the area of ​​the area where A and B merge, as shown in Equation (6).

[0038]

[0039] Before solving the intersection-union ratio, we first solve for the areas enclosed by the actual navigation trajectory and the optimal trajectory projected onto the ground and the coordinate axes, respectively. Since the path planning first yields a series of discrete coordinate points, we solve for the areas enclosed by the lines connecting the two points and the coordinate axes, and then sum them. Equations (7) and (8) are the formulas for calculating the intersection area and the union area, respectively.

[0040]

[0041] In equations (7) and (8), A and B are two curves, x i and x i-1Let y be two adjacent points on the x-axis. Ai and y Ai-1 Let xi and xi-1 represent the y values ​​corresponding to curve A, respectively. Bi and y Bi-1 They represent x respectively i and x i-1 The corresponding y value on curve B; A∩B represents the intersection of the areas enclosed by curves A and B and the x-axis, and A∪B represents the union of the areas enclosed by curves A and B and the x-axis;

[0042] Project the images onto the XOY, YOZ, and XOZ planes respectively, calculate the intersection-union ratio (IU / R) of the projected areas for each plane, and then sum the IU / R values ​​of the three planes using a weighted average. This sum is the final IU / R value.

[0043] IOU = 0.7 × IOU XOY +0.15×IOU XOZ +0.15×IOU YOZ (9)

[0044] In equation (9), IOU represents the total intersection-union ratio, and IOU XOY IOU represents the intersection-union ratio of the curves in the XOY plane. XOZ The Intersection over Union (IOU) represents the crossover ratio of the curves in the XOZ plane. YOZ This represents the intersection-to-union ratio of the curves in the YOZ plane.

[0045] Furthermore, in step 4, the evaluation method of the analytic hierarchy process is as follows:

[0046] 4.1 Construct a hierarchical structure;

[0047] 4.2 Construct the comparison judgment matrix and calculate the weight values;

[0048] 4.3 Substitute the weight values ​​and similarity values ​​to calculate the evaluation results.

[0049] The beneficial effects of this invention are:

[0050] 1. This invention summarizes and generalizes the general capability evaluation indicators of autonomous navigation systems, establishes a method for comparing the similarity between the target value and the actual value of the indicators, and uses the analytic hierarchy process (AHP) to achieve the fusion of indicator similarity.

[0051] 2. The general capability evaluation method for autonomous navigation systems proposed in this invention is applicable to the comprehensive evaluation of general capabilities under different scenarios and tasks. The selected indicators can comprehensively describe the navigation performance of autonomous navigation systems in time and space, and can quantitatively provide evaluation results.

[0052] 3. This invention enables an accurate and comprehensive assessment of the general capabilities of autonomous navigation systems, thereby providing a basis and direction for the research and improvement of autonomous navigation technology. It also has certain reference value for performance evaluation research in other fields. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the overall process of general capability assessment for this invention;

[0054] Figure 2 This is a schematic diagram illustrating the acquisition of the target value for the general capability assessment of this invention;

[0055] Figure 3 This is a schematic diagram illustrating the similarity calculation of the indicators in this invention;

[0056] Figure 4 This is a diagram of the Ferramucci distance;

[0057] Figure 5 This is a schematic diagram illustrating the calculation of the area enclosed by the path and coordinate axes;

[0058] Figure 6 This is a flowchart of the Analytic Hierarchy Process (AHP).

[0059] Figure 7 This is a diagram of the hierarchical structure. Detailed Implementation

[0060] The general process of the autonomous navigation general capability assessment method based on similarity and hierarchical analysis described in this embodiment is as follows: Figure 1 As shown, the evaluation method is as follows:

[0061] 1. First, the expected target value is obtained through a global path planning algorithm under the specified scenario. The target value includes data such as path point set, path length, path energy consumption, and path time consumption.

[0062] 2. Then compare the results with the actual values ​​obtained by the autonomous navigation system in the same scenario, and calculate the similarity of the corresponding technical indicators;

[0063] 3. Then, the analytic hierarchy process (AHP) is used to fuse and score the similarity of the indicators to obtain the final evaluation result, thereby achieving the evaluation of the general capability of autonomous navigation.

[0064] In step 1, the target value of the autonomous navigation system's capability assessment is obtained by a global path planning algorithm under a specified scenario. This target value includes data such as the path point set, path length, path energy consumption, and path time consumption. Taking an autonomous vehicle as an example, since the A* algorithm can obtain the optimal path, it is used for path planning and to obtain a theoretical path point set. Then, based on the theoretical path point set, the target value data such as path length, path energy consumption, and path time consumption are calculated. The calculation method is as follows: Figure 2 As shown.

[0065] In step 2, the similarity of the indicators is calculated as follows:

[0066] Based on data characteristics such as path point set, path length, path energy consumption, and path time consumption of autonomous navigation systems, the similarity of indicators can be divided into three aspects: consumption similarity, trajectory space similarity, and trajectory projection similarity. Figure 3 As shown below, the similarity calculation process will be introduced.

[0067] 2.1 Consumption Similarity Calculation

[0068] (1) Time consumption ratio

[0069] The time consumption ratio refers to the ratio of the actual time taken to reach the target location to the ideal time taken to reach the target location.

[0070]

[0071] Where T L T represents the actual time taken to reach the target. A This represents the ideal time to reach the destination. The closer the ratio is to 1, the less time is spent on path planning.

[0072] (2) Path consumption ratio

[0073] Path cost ratio refers to the ratio of the actual distance traveled to reach the target location to the ideal path distance to reach the target location.

[0074]

[0075] Where S L S represents the length of the actual trajectory taken to reach the target. A This represents the length of the trajectory taken to reach the target under ideal conditions. The closer the ratio is to 1, the shorter the path taken by the path planning.

[0076] (3) Energy consumption ratio

[0077] Energy consumption ratio refers to the energy E actually consumed to reach the target location. L The energy E required to reach the target under ideal conditions A The ratio of the two values. The closer the ratio is to 1, the less energy is consumed in path planning.

[0078]

[0079] 2.2 Trajectory Space Similarity

[0080] Trajectory similarity assessment can be categorized into several methods; here, we use the Framinger distance method. The Framinger distance is a description of path space similarity, considering both the distance within the path space and its applicability to comparing the similarity of spatial paths. The Framinger distance calculates the maximum difference between two sequences without considering factors such as curve trend similarity. Let the pair (S, d) be a metric space, where d is a metric function on S. Let A and B be two continuous curves on S, satisfying:

[0081] A:[0,1]→S,B:[0,1]→S (4)

[0082] The Framing distance between curve A and curve B is defined as shown in equation (5), which will be explained below. d is a metric function on S, α and β are two reparameterized functions of the unit interval, and F(A,B) is the Framing distance between curve A and curve B. Let t be a time point, then at that time, the sampling point on curve A is A(α(t)) and the sampling point on curve B is B(β(t)). In each sampling, t discretely traverses the interval [0,1], and the maximum distance under this sampling can be obtained.

[0083]

[0084] The Framinger distance is the value obtained through sampling that minimizes the maximum distance. For example... Figure 4 As shown, by traversing each pair of points on sequences A and B, the maximum distance between them is calculated, and the difference between the ideal path and the actual path is evaluated. This method is also applicable to three-dimensional paths.

[0085] 2.3 Trajectory Projection Similarity

[0086] In target detection tasks, the Intersection of Union (IoU) is used as a metric to describe the degree of overlap between two bounding boxes. Here, we borrow the concept of IoU to compare the difference between the actual navigation trajectory and the optimal trajectory. Here, A and B are the areas enclosed by the actual navigation trajectory and the optimal trajectory projected onto the ground and the coordinate axes, respectively. The IoU is the area of ​​the area where A and B intersect, divided by the area of ​​the area where A and B merge, as shown in Equation (6).

[0087]

[0088] Before solving for the intersection-union ratio, we should first solve for the areas enclosed by the actual navigation trajectory and the optimal trajectory projected onto the ground and the coordinate axes, such as... Figure 5As shown, since the path planning first yields a series of discrete coordinate points, the area enclosed by the line connecting the two points and the coordinate axis is calculated separately and then summed. Equations (7) and (8) are the formulas for calculating the area of ​​the intersection region and the area of ​​the union region, respectively.

[0089] The following explains formulas (7) and (8), where A and B are two curves, x i and x i-1 Let y be two adjacent points on the x-axis. Ai and y Ai-1 Let xi and xi-1 represent the y values ​​corresponding to curve A, respectively. Bi and y Bi-1 They represent x respectively i and x i-1 The corresponding y-value on curve B. A∩B represents the intersection of the areas enclosed by curves A and B and the x-axis, and A∪B represents the union of the areas enclosed by curves A and B and the x-axis.

[0090]

[0091] Since we are studying three-dimensional terrain, its coordinate axes are divided into three axes: X, Y, and Z, and it is also divided into three coordinate planes. Therefore, we need to project onto the XOY, YOZ, and XOZ planes respectively, calculate the intersection-union ratio of the projected areas respectively, and then sum the intersection-union ratios of the three planes with weights to obtain the final intersection-union ratio value.

[0092] Equation (9) is explained below. IOU represents the total intersection-union ratio. XOY IOU represents the intersection-union ratio of the curves in the XOY plane. XOZ The Intersection over Union (IOU) represents the crossover ratio of the curves in the XOZ plane. YOZ This represents the intersection-to-union ratio of the curves in the YOZ plane.

[0093] IOU = 0.7 × IOU XOY +0.15×IOU XOZ +0.15×IOU YOZ (9)

[0094] 3. Evaluation using the Analytic Hierarchy Process (AHP)

[0095] The Analytic Hierarchy Process (AHP), proposed by American operations researcher Saaty in the 1970s, is a systems analysis method that combines qualitative and quantitative analysis. It is characterized by its clear approach and ability to model the analyst's knowledge and preferences, and is currently widely used in decision-making, forecasting, and evaluation. This method analyzes the factors involved in a complex problem and their interrelationships, categorizing different elements into different levels. It then uses a judgment matrix to compare the elements at each level, determining the weight of each element on the criteria for that level, and finally calculating the combined weight of each level's elements for the final goal.

[0096] The evaluation process of the analytic hierarchy process is as follows: Figure 6 As shown, a hierarchical structure is first constructed. Based on this, a comparison judgment matrix is ​​constructed using the TLSaaty1-9 scale scoring standard. The normalized eigenvector of the comparison judgment matrix is ​​then calculated using the square root method, satisfying the consistency test. The calculated normalized eigenvector represents the weight ranking of each factor. Finally, the weight values ​​and similarity values ​​are substituted to calculate the evaluation result. The specific steps are described below.

[0097] First, establish a hierarchical structure for the evaluation indicators, such as... Figure 7 As shown, the hierarchical structure is divided into three layers: the target layer, the criteria layer, and the scheme layer. The target layer represents the results of the autonomous navigation system's capability assessment; the criteria layer represents consumption similarity, trajectory space similarity, and trajectory projection similarity; and the scheme layer represents time consumption ratio, path consumption ratio, and energy consumption ratio.

[0098] After constructing the hierarchical structure, the TLSaaty 1-9 scale scoring standard is used to assign values ​​to the relative importance of the indicators, forming a judgment matrix. The selection of matrix element values ​​is achieved by comparing the importance of each element pairwise. Specifically, the importance of each indicator in the actual navigation task should be considered. For example, for more exploratory tasks, path quality is more important than energy consumption and time, while for more goal-oriented tasks, time and energy consumption are more important.

[0099] After constructing the judgment matrix, the normalized eigenvectors of the matrix are obtained using the root method. The obtained eigenvectors W are the weight sequences of each factor, and the elements W in the eigenvectors are... i This is the corresponding weight. Finally, the obtained weight W is... i Similarity calculation value S i Multiply and sum, where n is the number of scheme-level indicators, as shown in Equation (10), and finally calculate the evaluation result of the general capability of the autonomous navigation system.

[0100]

[0101] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent substitutions, and improvements made to the above embodiments without departing from the scope of the present invention, based on the technical essence of the present invention and within the spirit and principles of the present invention, shall still fall within the protection scope of the present invention.

Claims

1. A general capability evaluation method for autonomous navigation systems based on similarity and analytic hierarchy process, characterized in that, The method includes the following steps: Step 1: Obtain the target value for the general assessment of autonomous navigation capabilities; Step 2: Compare the target value with the actual value and calculate the similarity. Step 3: Evaluate the similarity of indicators using the analytic hierarchy process (AHP). In step 2, the similarity of indicators includes consumption similarity, trajectory space similarity, and trajectory projection similarity; The similarity calculation is as follows: (1) Time consumption ratio The ratio of the actual time taken to reach the target location to the ideal time taken to reach the target location is calculated as follows: (1) in T L This indicates the actual time taken to reach the destination. T A This represents the ideal time to reach the destination; the closer the ratio is to 1, the less time is spent on path planning. (2) Path consumption ratio The ratio of the actual distance traveled to the target location to the ideal path distance to the target location is calculated as follows: (2) in S L This represents the length of the actual trajectory taken to reach the target. S A This represents the length of the trajectory taken to reach the target under ideal conditions; the closer the ratio is to 1, the shorter the path taken by the path planning. (3) Energy consumption ratio The ratio of the energy consumed to actually reach the target location to the energy consumed to reach the target under ideal conditions is calculated as follows: (3) in, E L This represents the energy consumed to actually reach the target location; E A This represents the energy consumed to reach the target under ideal conditions; the closer the ratio is to 1, the less energy is consumed by the path planning. The similarity of trajectory spaces is determined using the Framincher distance method. Let the binary pair ( S , d ) is a metric space, where d yes S The metric function on, let A and B yes S Two continuous curves on the above satisfy: (4) Then the curve A With curve B The Franzing distance is defined as shown in equation (5). (5) In Equation 5, d yes S The metric function on, α and β These are two reparameterized functions on a unit interval. ，F ( A,B Let t be the Framing distance between curve A and curve B; let t be time point, then at that time, curve A... A The sampling points on are A ( α ( t )), curve B Upsampling points are B ( β ( t In each sampling t By discretely traversing the interval [0,1], the maximum distance under this sampling condition is obtained. ; The trajectory projection similarity is calculated as follows: In object detection tasks, the intersection-union ratio (IUGR) is used as a metric to describe the degree of overlap between two bounding boxes, and is used to compare the difference between the actual navigation trajectory and the optimal trajectory; among which... A and B Let be the areas enclosed by the actual navigation trajectory and the optimal trajectory projected onto the ground and bounded by the coordinate axes, respectively. A and B The area of ​​the intersecting region is greater than that of the previous region. A and B The area of ​​the merged region is the intersection-union ratio, as shown in equation (6). （6） Before solving the intersection-union ratio, we first solve for the areas enclosed by the actual navigation trajectory and the optimal trajectory projected onto the ground and the coordinate axes, respectively. Since the path planning first yields a series of discrete coordinate points, we solve for the areas enclosed by the lines connecting the two points and the coordinate axes, and then sum them. Equations (7) and (8) are the formulas for calculating the intersection area and the union area, respectively. （7） （8） In equations (7) and (8), A and B Two curves, x i and x i-1 for x Two adjacent points on the axis, y Ai and y Ai-1 They represent xi and xi-1 In the curve A The corresponding above y value, y Bi and y Bi-1 They represent x i and x i-1 In the curve B The corresponding above y value; A ∩ B Represents curve A With curve B and x The intersection of the areas enclosed by axes. A ∪ B Represents curve A With curve B and x The union of the areas enclosed by axes; To each XOY, YOZ, XOZ For each plane projection, calculate the intersection-union ratio of the projected areas. Then, sum the intersection-union ratios of the three planes using a weighted average to obtain the final intersection-union ratio value. （9） In equation (9), IOU Indicates the total intersection and union ratio. IOU XOY Indicates the curve at XOY The intersection-union ratio of a plane. IOU XOZ Indicates the curve at XOZ The intersection-union ratio of a plane. IOU YOZ Indicates the curve at YOZ The intersection-union ratio of planes.

2. The general capability evaluation method for autonomous navigation systems based on similarity and analytic hierarchy process according to claim 1, characterized in that, In step 1, the target value for the general capability assessment of the autonomous navigation system is the expected target value obtained by the global path planning algorithm under the specified scenario, including path point set, path length, path energy consumption, and path time consumption data.

3. The general capability evaluation method for autonomous navigation systems based on similarity and analytic hierarchy process according to claim 1, characterized in that, In step 4, the evaluation method of the analytic hierarchy process is as follows: 4.1 Construct a hierarchical structure; 4.2 Construct the comparison judgment matrix and calculate the weight values; 4.3 Substitute the weight values ​​and similarity values ​​to calculate the evaluation results.

Images