A method for predicting satellite catalog passability based on multi-sensor fusion

Abstract

The present invention provides a method for predicting the passability of a star chart based on multi-sensor fusion, including: determining the target location of a rover; obtaining a target ground image and star-soil geological structure data of the target location; semantically segmenting the target ground image and the star-soil geological structure to obtain a semantic segmentation result; predicting the passability of the target location; the rover moving to the target location; outputting the actual value of the passability of the target location based on the wheel slip rate; and correcting the predicted value of the passability. The method for predicting the passability of a star chart based on multi-sensor fusion provided by the present invention solves the problem that the traditional unmanned rover cannot predict the feedback of the current wheel-ground interaction when it touches the ground, resulting in untimely response, and breaks through the limitation that visual means cannot accurately distinguish the mechanical properties of the star-soil in a uniform environment of a star chart with similar colors and structures, improves the prediction ability of the carrying capacity of the star-soil, and ensures the safety of large-scale mobile detection by unmanned rover.

Description

Technical Field

[0001] The invention relates to a star catalog passability prediction method based on multi-sensor fusion. Background Art

[0002] When implementing planetary exploration missions, in order to ensure the safety of astronauts, it is usually necessary to launch an unmanned planetary rover to conduct a more detailed survey of the surface environment. In order to fully understand the geological distribution of the surface, the unmanned planetary rover needs to have the ability to move over a large range, and a higher detection priority should be set for some special terrains (such as craters). Affected by low gravity, the particle size and looseness of the planetary surface soil vary greatly. Even the internal mechanical characteristics of the ground with similar appearance may be very different. The overall environment is single but the mechanical properties are complex and changeable. The soft and low-cohesion soil cannot bear the weight of the unmanned planetary rover, which is prone to wheel sinking and even trapping the rover.

[0003] Under the current conventional means of exploring the planetary soil surface, the feedback of the current wheel-ground interaction can only be sensed when the wheels of an unmanned planetary rover touch the planetary soil. In this case, even if the adhesion of the planetary soil under the wheel is found to be low, since the planetary rover has moved to the top of the low-adhesion planetary soil surface, the remedial measures that can be taken are extremely limited. The United States' "Spirit" Mars rover completely lost its mobility function for this reason, which led to the failure of the entire exploration mission.

[0004] Most existing planetary rovers are equipped with visual sensors, so using visual means to classify the ground has become a research hotspot to solve this problem. However, due to the limitations of visual characteristics, this method can usually only detect clearly distinguishable foreign objects such as stones and craters, but cannot detect the specific mechanical properties of planetary soil. This leads to limited classification effect in the uniform environment of planets with similar colors and structures. Summary of the invention

[0005] The purpose of the present invention is to provide a star catalog passability prediction method based on multi-sensor fusion, which has the advantage of strong prediction ability.

[0006] To achieve the above object, the present invention provides a star catalog passability prediction method based on multi-sensor fusion, which comprises:

[0007] S10, determining the target location for the planetary rover to move to;

[0008] S20, the visual camera performs visual detection on the ground of the target location to obtain a target ground image of the target location, and the detection radar performs geological structure detection on the ground of the target location to obtain the star soil geological structure of the target location;

[0009] S30, performing semantic segmentation on the target ground image based on texture and the star-soil geological structure to obtain a semantic segmentation result;

[0010] S40, predicting the passability of the target location based on the semantic segmentation result to obtain a passability performance prediction value;

[0011] S50, the planetary vehicle moves to the target location;

[0012] S60, obtaining a wheel slip rate of the planetary vehicle at the target location, and outputting an actual value of the passability performance of the target location based on the wheel slip rate;

[0013] S70, correcting the passable performance prediction value with the passable performance actual value, and using the difference between the passable performance actual value and the passable performance prediction value as a correction parameter.

[0014] In this scheme, the target ground image and the star-soil geological structure are integrated to obtain the predicted value of the predicted surface's passability, and the actual value of the passability is obtained by the actually measured wheel slip rate. Correction parameters are obtained by comparing the actual value of the passability and the predicted value of the passability, which provides a basis for correcting the subsequent predicted value of the passability, thereby improving the accuracy of the prediction of the planetary surface's passability.

[0015] Preferably, when no correction parameters are stored in the planetary vehicle, step S10 includes:

[0016] S101, selecting a plurality of candidate locations near the planetary rover;

[0017] S102, the visual camera obtains ground images of the plurality of candidate locations respectively, and the detection radar obtains the star-soil geological structure of the plurality of candidate locations;

[0018] S103, predicting a passability prediction value of each candidate location according to the ground image of the candidate location and the star-soil geological structure;

[0019] S104: Select the candidate location with the best performance prediction value as the target location.

[0020] In this scheme, when the rover has just landed and no correction parameters have been generated, the hardest part of the surface near the rover is first selected as the target location based on past experience to minimize the probability of the rover's wheels sinking.

[0021] Preferably, when the planetary vehicle stores correction parameters, step S10 includes:

[0022] S101, selecting a plurality of candidate locations near the planetary rover;

[0023] S102, the visual camera obtains ground images of the plurality of candidate locations respectively, and the detection radar obtains the star-soil geological structure of the plurality of candidate locations;

[0024] S103, predicting a passability prediction value of each candidate location according to the ground image of the candidate location and the star-soil geological structure;

[0025] S104, using the correction parameter to correct the pass performance prediction value to obtain a pass performance correction value;

[0026] S105: Select the candidate location with the best pass performance correction value as the target location.

[0027] Preferably, the star soil geological structure acquired by the detection radar includes the dielectric constant of the target location.

[0028] Preferably, in S60, obtaining the wheel slip rate of the planetary vehicle at the target location includes:

[0029] S601, acquiring a wheel track image at the target location;

[0030] S602: Acquire the wheel slip rate based on the rut image.

[0031] In this solution, a rutting image is obtained through a camera, and the wheel slip rate is calculated based on the rutting image. The rutting image is easy to obtain and the slip rate calculation accuracy is high.

[0032] Preferably, between step S40 and step S50, the following steps are further included:

[0033] S41, determining whether the correction parameter has been stored in the memory; if so, using the correction parameter to correct the passable performance prediction value to obtain the corrected passable performance prediction value; if the correction parameter has not been stored, directly executing S50.

[0034] Preferably, after step S70, the following steps are further included:

[0035] S71. Determine whether the correction parameter has been stored in the memory. If the correction parameter has been stored, use the correction parameter obtained in S70 to correct the existing correction parameter; if the correction parameter has not been stored, store the correction parameter.

[0036] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the above-mentioned star catalog passability prediction method based on multi-sensor fusion is implemented.

[0037] The present application also provides an electronic device, including a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the above-mentioned star catalog passability prediction method based on multi-sensor fusion is implemented.

[0038] In summary, compared with the prior art, the star catalog passability prediction method based on multi-sensor fusion provided by the present invention has the following beneficial effects:

[0039] The star catalog passability prediction method based on multi-sensor fusion provided by the present invention introduces three perception modes: visual camera, detection radar, and wheel-ground direct contact feedback. It solves the problem that traditional unmanned planetary rovers cannot predict the current wheel-ground interaction feedback when they touch the ground, resulting in untimely response. It breaks through the limitation that visual means cannot accurately distinguish the mechanical properties of star soil in a uniform environment of star tables with similar colors and structures, improves the prediction ability of star soil carrying capacity, and ensures the safety of large-scale mobile detection of unmanned planetary rovers. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Flow chart of the star catalog passability prediction method based on multi-sensor fusion in this embodiment. DETAILED DESCRIPTION

[0041] The following will be combined with the attached embodiment of the present invention Figure 1 , the technical solutions, structural features, objectives achieved and effects in the embodiments of the present invention are described in detail.

[0042] It should be noted that the drawings are in a very simplified form and use non-precise proportions. They are only used to conveniently and clearly assist in explaining the embodiments of the present invention, and are not used to limit the conditions for the implementation of the present invention. Therefore, they have no substantive technical significance. Any structural modification, change in proportional relationship or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical content disclosed by the present invention.

[0043] It should be noted that, in the present invention, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only the elements explicitly listed, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0044] The present invention provides a star catalog passability prediction method based on multi-sensor fusion. A visual camera and a detection radar are provided on the planetary rover. The prediction method comprises the following steps:

[0045] S10, determine the target location of the planetary rover. In this embodiment, when the planetary rover has just landed on a planet, due to the different gravitational constants of each planet, the planetary rover's memory has not yet stored the data of the correction parameters of the landing planet. The planetary rover cannot use the correction parameters and prediction models established on the earth or other planets to make a relatively accurate prediction of the carrying capacity of the planetary soil. Therefore, when the planetary rover has just landed, the planetary rover does not store the correction parameters for the planet. At this time, step S10 includes:

[0046] S101. Select multiple candidate locations near the rover. The visual camera and detection radar on the rover explore and search the texture of the star soil in multiple directions near the rover as candidate locations for the forward target.

[0047] S102: The visual camera obtains the ground images of the plurality of candidate locations respectively, and the detection radar obtains the star-soil geological structure of the plurality of candidate locations. For each candidate location, the visual camera and the detection radar detect the ground image and star-soil geological structure of the candidate location respectively.

[0048] S103, predicting the passability prediction value of each candidate location based on the ground image and the terrestrial geological structure of the candidate location. The planetary rover first predicts the passability of each candidate location based on the prediction model established on the earth or other planets to obtain the prediction result of the passability of each candidate location.

[0049] S104, selecting the candidate location with the best predicted passability as the target location. Ranking the candidate locations based on the passability of each candidate location obtained in the previous step, and selecting the candidate location with the best passability as the target location to minimize the probability of sinking into the surface of the star soil and losing the ability to move.

[0050] S20, the visual camera performs visual detection on the ground of the target location to obtain the target ground image of the target location, and the detection radar performs geological structure detection on the ground of the target location to obtain the star soil geological structure of the target location. After determining the target location, the visual camera takes a picture of the target ground to obtain the ground image of the target ground, and the detection radar scans the target location to obtain the star soil geological structure of the target ground.

[0051] S30, semantic segmentation is performed on the target ground image based on texture and on the star soil geological structure to obtain semantic segmentation results. The target ground image is semantically segmented to obtain a semantic segmentation result based on image texture, and the star soil geological structure is semantically segmented to obtain a semantic segmentation result based on dielectric properties. Based on the ground image of the target location obtained in the previous step, the ground image is semantically segmented, and texture semantic segmentation is to segment the sand, rock and other areas appearing on the ground image. The different areas appearing in the ground image are segmented. Dielectric property semantic segmentation is to identify the star soil geological structure of the target location. In this embodiment, the star soil geological structure obtained by the detection radar includes star soil sedimentary structure, particle size and material composition. In the process of semantic segmentation, the star soil sedimentary structure, particle size, material composition and other characteristics are automatically identified. In addition, in this embodiment, the star soil geological structure obtained by the detection radar also includes the dielectric constant of the target location. Automatically identify abnormal mutations in geological characteristics based on dielectric properties and output corresponding segmentation results.

[0052] S40, predicting the passability of the target location through the semantic segmentation result and the label segmentation result to obtain a passability prediction value. After obtaining the semantic segmentation result of the target location photo and identifying the star soil geological structure, predicting the passability of the target location to obtain the passability prediction data of the target location.

[0053] In this embodiment, between step S40 and step S50, the following steps are further included:

[0054] S41, judging whether the memory has stored the correction parameter, if the memory has stored the correction parameter, then using the existing correction parameter to correct the pass performance prediction value to obtain the corrected pass performance prediction value. If the memory has not stored the correction parameter, then directly executing S50.

[0055] S50, the planetary car moves to the target location. After the prediction is completed, the planetary car moves to the selected target location.

[0056] S60, obtaining the wheel slip rate of the planetary vehicle at the target location, and outputting the actual value of the passability of the target location based on the wheel slip rate. In this embodiment, obtaining the wheel slip rate of the planetary vehicle at the target location includes: S601, obtaining the rutting image at the target location; after the planetary vehicle moves to the target location, the planetary vehicle is provided with a camera specifically used to obtain the rutting image, and the camera takes the rutting image. S602, obtaining the wheel slip rate based on the rutting image. The processor obtains the wheel slip rate based on the rutting image analysis. The wheel slip rate is calculated as the ratio between the distance traveled by the wheel and the displacement formed by the wheel rotation. When the ratio is 1, it indicates that the vehicle travel is completely caused by the rotation of the wheel, and the passability of this location is good; when the ratio is far from 1, it indicates that there is slippage or idling between the wheel and the ground. The greater the difference between the ratio and 1, the poorer the passability of this location. After obtaining the wheel slip rate as an indicator, the actual value of the passability at the target location is obtained based on the wheel slip rate.

[0057] S70, the passability prediction value is corrected with the actual passability value, and the difference between the actual passability value and the passability prediction value is a correction parameter. The passability prediction value obtained in S40 is compared with the actual passability value obtained in S60, and the difference between the two is represented by a correction parameter, and the correction parameters of various landforms of the planet where the planetary rover is located are continuously accumulated to continuously improve the accuracy of the passability prediction of other locations on the planet.

[0058] In this embodiment, after step S70, the following steps are further included:

[0059] S71, judging whether the memory has stored the correction parameters, if the memory has stored the correction parameters, then using the correction parameters obtained in S70 to correct the existing correction parameters; the correction parameters can be stored separately based on the semantic segmentation results of the visual camera, that is, different correction parameters are stored for different star-soil surface features to improve the accuracy of passability prediction for different star-soil surfaces. If the planetary rover does not store the correction parameters, then the correction parameters are stored for subsequent correction of the passability prediction value to improve the accuracy of the prediction.

[0060] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the star catalog passability prediction method based on multi-sensor fusion as described above is implemented.

[0061] The present application also provides an electronic device, including a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the star catalog passability prediction method based on multi-sensor fusion as described above is implemented.

[0062] Example 2

[0063] The same parts of this embodiment 2 as those of embodiment 1 will not be repeated, and only the different parts will be described. In embodiment 1, no correction parameters are stored in the planetary vehicle, while in embodiment 2, the correction parameters of the planet are already stored in the planetary vehicle.

[0064] When the planetary vehicle stores correction parameters, step S10 includes:

[0065] S101, selecting multiple candidate locations near the planetary rover. The visual camera and detection radar on the planetary rover explore and search the texture of the star soil in multiple directions near the planetary rover as candidate locations for the forward target.

[0066] S102, the visual camera obtains the ground images of multiple candidate locations respectively, and the detection radar obtains the star-soil geological structure of the multiple candidate locations; for each candidate location, the visual camera and the detection radar respectively detect the ground image and star-soil geological structure of the candidate location.

[0067] S103, predicting the passability prediction value of each candidate location based on the ground image and the planetary soil geological structure of the candidate location; the planetary rover first predicts the passability of each candidate location based on the prediction model established on the earth or other planets to obtain the prediction result of the passability of each candidate location.

[0068] S104, using correction parameters to correct the passable performance prediction value to obtain a passable performance correction value; for the passable performance prediction value obtained by calculation, using correction parameters to correct the passable performance prediction value to obtain a passable performance correction value.

[0069] S105, selecting the candidate location with the best passability correction value as the target location. Ranking the candidate locations based on the passability of each candidate location obtained in the previous step, and selecting the candidate location with the best passability as the target location to minimize the probability of sinking into the surface of the star soil and losing the ability to move.

[0070] In other embodiments, the target location of the rover may be directly determined according to mission requirements.

[0071] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

Claims

1. A star catalog passability prediction method based on multi-sensor fusion, characterized in that: The planetary rover is provided with a visual camera and a detection radar, and the prediction method includes: S10, determining the target location for the planetary rover to move to; S20, the visual camera performs visual detection on the ground of the target location to obtain a target ground image of the target location, and the detection radar performs geological structure detection on the ground of the target location to obtain star-soil geological structure data of the target location; S30, performing semantic segmentation on the target ground image based on texture and the star-soil geological structure to obtain a semantic segmentation result; S40, predicting the passability of the target location based on the semantic segmentation result to obtain a passability performance prediction value; S50, the planetary vehicle moves to the target location; S60, obtaining a wheel slip rate of the planetary vehicle at the target location, and outputting an actual value of the passability performance of the target location based on the wheel slip rate; S70, correcting the predicted value of passability with the actual value of passability, and using the difference between the actual value of passability and the predicted value of passability as a correction parameter; obtaining the predicted value of passability of the predicted surface by fusing the target ground image and the star-soil geological structure, and obtaining the actual value of passability by the actually measured wheel slip rate.

2. The star catalog passability prediction method based on multi-sensor fusion according to claim 1, characterized in that: When no correction parameters are stored in the planetary vehicle, step S10 includes: S101, selecting a plurality of candidate locations near the planetary rover; S102, the visual camera obtains ground images of the plurality of candidate locations respectively, and the detection radar obtains the star-soil geological structure of the plurality of candidate locations; S103, predicting a passability prediction value of each candidate location according to the ground image of the candidate location and the star-soil geological structure; S104: Select the candidate location with the best performance prediction value as the target location.

3. The star catalog passability prediction method based on multi-sensor fusion as claimed in claim 1, characterized in that: When the planetary vehicle stores correction parameters, step S10 includes: S101, selecting a plurality of candidate locations near the planetary rover; S102, the visual camera obtains ground images of the plurality of candidate locations respectively, and the detection radar obtains the star-soil geological structure of the plurality of candidate locations; S103, predicting a passability prediction value of each candidate location according to the ground image of the candidate location and the star-soil geological structure; S104, using the correction parameter to correct the pass performance prediction value to obtain a pass performance correction value; S105: Select the candidate location with the best pass performance correction value as the target location.

4. The star catalog passability prediction method based on multi-sensor fusion according to claim 1, characterized in that: The star soil geological structure acquired by the detection radar includes the dielectric constant of the target location.

5. The star catalog passability prediction method based on multi-sensor fusion according to claim 1, characterized in that: At S60, obtaining the wheel slip rate of the planetary vehicle at the target location includes: S601, acquiring a wheel track image at the target location; S602: Acquire the wheel slip rate based on the rut image.

6. The star catalog passability prediction method based on multi-sensor fusion according to claim 1, characterized in that: Between step S40 and step S50, the following steps are also included: S41, determining whether the correction parameter has been stored in the memory; if so, using the correction parameter to correct the passable performance prediction value to obtain the corrected passable performance prediction value; if the correction parameter has not been stored, directly executing S50.

7. The star catalog passability prediction method based on multi-sensor fusion according to claim 6, characterized in that: After step S70, the following steps are also included: S71. Determine whether the correction parameter has been stored in the memory. If the correction parameter has been stored, use the correction parameter obtained in S70 to correct the existing correction parameter; if the correction parameter has not been stored, store the correction parameter.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the star catalog passability prediction method based on multi-sensor fusion as described in any one of claims 1 to 7 is implemented.

9. An electronic device, characterized in that: The invention comprises a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the star catalog passability prediction method based on multi-sensor fusion as described in any one of claims 1 to 7 is implemented.

Images