A method for online packing and storage of extraterrestrial object samples

Abstract

This invention relates to an online binning and placement method for extraterrestrial object sampling. The sampled rock samples are simplified into regular cuboids. The internal space of the bin is reconstructed using a depth camera. Feasible placement locations are initially screened using farthest-point sampling. An evaluation index function for sample placement is designed, including indicators such as distance, actual space occupancy within the bin, space occupancy of the sample itself, size similarity, and center deviation. The placement location with the highest evaluation is selected, achieving high-volume-ratio binning and placement. This invention uses a dynamic sampling rate and farthest-point sampling to initially screen all feasible placement points, improving the efficiency of finding the optimal placement.

Description

Technical Field

[0001] This invention relates to an online packing and placement method for extraterrestrial object sampling, belonging to the field of extraterrestrial object detection technology. Background Technology

[0002] Currently, sampling devices for extraterrestrial object exploration missions primarily rely on drilling, suitable for sampling deep soil layers in the regolith of planetary surfaces. However, these devices are relatively complex and highly susceptible to the characteristics of the planetary surface soil. While sampling devices and techniques for surface rocks on extraterrestrial objects are relatively simple, the varying sizes of the sampled rocks necessitate efficient packing and storage methods to improve packing efficiency and sample volume. Summary of the Invention

[0003] The technical problem solved by this invention is to overcome the shortcomings of the prior art and propose an online binning and placement method for extraterrestrial celestial body sampling. The method simplifies the sampled rock samples into regular cuboids, reconstructs the internal space of the bin using a depth camera, uses sampling at the farthest point to initially screen feasible placement locations and reduce the amount of computation, selects the placement location with the highest evaluation, and achieves high volumetric binning and placement.

[0004] The solution of the present invention is:

[0005] A method for online packing and placement of extraterrestrial object samples simplifies the sampled rock samples into regular cuboids, reconstructs the internal space of the box using a depth camera, uses sampling at the farthest point to initially screen feasible placement locations, designs a sample placement evaluation index function, and the indexes include distance index, actual space occupancy index, placement space occupancy index, size similarity index, and center deviation index. The placement location with the highest evaluation is selected to achieve high volumetric efficiency packing and placement.

[0006] Furthermore, the interior space of the box is reconstructed using a depth camera. The specific steps are as follows:

[0007] 2.1 The continuous internal space of the box is discretized into a discrete space of unit length d, and this space is described by a three-dimensional matrix B. The X, Y, and Z directions of matrix B represent the length, width, and height of the box, respectively. The three-dimensional dimensions of matrix B are n. x ×n y ×n z The value in matrix B indicates whether the space inside the box corresponding to that position is occupied, with 1 indicating occupied and 0 indicating unoccupied.

[0008] 2.2 Use a depth camera to observe the box from a vertical direction and obtain the height map M of the area inside the box. h , height map M h The space occupancy matrix B mapped to the box.

[0009] Furthermore, using farthest point sampling to initially screen feasible placement locations and reduce computational load, the specific implementation steps are as follows:

[0010] 3.1 The sampled rock samples are simplified into regular cuboids and discretized to obtain three-dimensional dimensions l. x ×l y ×l z A rectangular prism;

[0011] 3.2 There are 6 possible placement orientations for a cuboid in the box [l] x ,l y ,l z ]、[l x ,l z ,l y ]、[l y ,l x ,l z ]、[l y ,l z ,l x ]、[l z ,l x ,l y ]、[l z ,l y ,l x For each placement orientation, calculate the placement matrix M0, where the size of matrix M0 is n. x ×n y When the remaining space in B(i,j,:) is greater than l z M0(i,j) is 1 when the condition is met, and 0 otherwise.

[0012] 3.3 Regarding the placement direction in step 3.2 [l] x ,l y ,l z ], using size l x ×l y A two-dimensional convolution kernel k, with values ​​all equal to 1, is used to perform a convolution operation M0*k on matrix M0, resulting in a matrix M of the same size as M0. k For M k Each element M in k (i,j), if M k The value of (i,j) is less than l. x ·l y (l x and l y If the product of the two is 0, then set it to 0; otherwise set it to 1.

[0013] 3.4 Take out M k The positions p:{(i,j)} corresponding to all elements with a value of 1 in (i,j) are used to filter ε·n from p using the farthest point sampling method. pSample p ε ε is the sampling rate, n p Let p be the number of elements, where the sampling rate ε = min(ε base ·(1+v r ) λε ,ε max ), where ε base Based on the sampling rate, ε max For the maximum sampling rate, v r λ represents the current occupancy rate of the box's internal space. ε To extend the power coefficients, using this dynamic sampling rate can improve the efficiency of finding the optimal placement.

[0014] 3.5 Repeat steps 3.2, 3.3, and 3.4 until all six placement orientations have been calculated, obtaining all sampled placement cases P:{(l i ,l j ,l k ,i,j)},l i l j and l k This refers to the placement orientation in step 3.2.

[0015] Furthermore, a sample placement evaluation index function is designed, and the placement position with the highest evaluation is selected to achieve high-volume packing and organization. The specific implementation steps are as follows:

[0016] Step 1: For each of the placement scenarios P, calculate the space occupancy matrix B′ of the boxes after placement according to this scenario, and calculate the centroid position g of the occupied space in matrix B′: [g x ,g y ,g z The distance index c1 to the origin o of the box space coordinate system is given by: c1 = |g·([n x ,n y ,n z ] T ) λ1 | λ2 Where λ1 and λ2 are the power coefficients of the distance index, respectively;

[0017] Step 2, calculate the actual space occupied in the box after the sample is placed, index c2, c2: the specific expression is c2=(∑(B′-B) / (l x ·l y ·l z )) λ3 , where λ3 is the power coefficient of the actual space occupancy index inside the box;

[0018] Step 3: Calculate the space occupancy index c3 within the box after sample placement using a 3D convolution kernel k3 with all element values ​​of 1, and a size m equal to the estimated minimum sampled rock sample volume. x ×m y ×m z Perform convolution operations B*k3 and B′*k3 respectively to obtain a matrix B of the same size as B. k3 and B′ k3 For B k3 and B′ k3 Each element B in k3 (i,j,k) and B′ k3 (i,j,k), if the value is greater than 0, it is set to 0, otherwise it is set to 1. Then the specific expression of c3 is c3=(∑(B k3 -B′ k3 ) / (l x ·l y ·l z )) λ4 λ4 is the power coefficient of the space occupied inside the box;

[0019] Step 4, calculate the similarity index c4 between the box size and the sample placement orientation. c4: The specific expression is c4 = (2 - ([n x ,n y ,n z ]·[l x ,l y ,l z ] T ) / (|[n x ,n y ,n z ]|·|[l x ,l y ,l z ]|)) λ5 , where λ5 is the power coefficient of the size similarity index;

[0020] Step 5: Calculate the deviation index c5 between the center position of the sample and the center position of the contact surface of the placement area. The specific expression for c5 is c5 = (1 + (|s p -s s |)·(1-v r ) / |s p |) λ6 , where s p These are the coordinates of the sample center in the sample volume coordinate system, s s λ6 is the coordinate of the center position of the contact surface of the area where the sample is placed in the volumetric coordinate system, and λ6 is the power coefficient of the center deviation index.

[0021] Step 6: Calculate the comprehensive evaluation index C:{c1·c2·c3·c4·c5} for all placement scenarios. This index comprehensively evaluates the occupancy of the space inside the box after placement. The placement scenario with the minimum value in C is the highest-rated placement scenario, and the sample is placed in the corresponding position.

[0022] The advantages of this invention compared to the prior art are:

[0023] (1) The present invention uses dynamic sampling rate to perform preliminary screening of all feasible placement points by sampling at the farthest point, which can improve the efficiency of finding the best placement situation;

[0024] (2) The present invention uses five indicators to comprehensively evaluate the feasible placement of each sample, including distance index, actual space occupied in the box index, space occupied in the box, size similarity index and center deviation index, which can improve the packing and placement volume ratio. Attached Figure Description

[0025] Figure 1 This shows the placement of samples 1 through 8.

[0026] Figure 2 This shows the placement of samples 21 through 28. Detailed Implementation

[0027] The present invention will be further described below in conjunction with the embodiments.

[0028] This embodiment of an online binning and placement method for extraterrestrial object sampling simplifies the sampled rock samples into regular cuboids, reconstructs the internal space of the bin using a depth camera, uses the farthest point sampling to initially screen feasible placement locations and reduce computational load, designs a sample placement evaluation index function that considers factors such as sample placement sparsity, placement direction, placement location, placement center of gravity, actual space occupancy within the bin, and space occupancy within the bin, selects the placement location with the highest evaluation, and achieves high volumetric binning and placement.

[0029] like Figure 1 , 2 As shown, the specific implementation steps of the online packing and placement method for extraterrestrial body sampling are as follows:

[0030] Step 1: Reconstruct the interior space of the box using a depth camera. The specific steps are as follows:

[0031] Step 1: Discretize the continuous internal space of the box into discrete spaces of unit length d, and describe this space using a three-dimensional matrix B. The X, Y, and Z directions of matrix B represent the length, width, and height of the box, respectively, and the three-dimensional dimensions of matrix B are n. x ×n y ×n z=50×30×24, the value in matrix B indicates whether the space inside the box corresponding to that position is occupied, 1 for occupied and 0 for unoccupied;

[0032] Step 2: Use a depth camera to observe the box from a vertical direction and obtain the height map M of the area inside the box. h , height map M h The space occupancy matrix B mapped to the box.

[0033] Step 2: Use the farthest point sampling to initially screen feasible placement locations. The specific implementation steps are as follows:

[0034] Step 1: Simplify the sampled rock sample into a regular cuboid and discretize it to obtain a three-dimensional dimension of l. x ×l y ×l z A cuboid with three dimensions of random values ​​ranging from 6 to 15;

[0035] Step 2, the cuboid has 6 possible placement orientations in the box [l x ,l y ,l z ]、[l x ,l z ,l y ]、[l y ,l x ,l z ]、[l y ,l z ,l x ]、[l z ,l x ,l y ]、[l z ,l y ,l x For each placement orientation, such as [l] x ,l y ,l z ] Calculate the placeable graph matrix M0, where the size of matrix M0 is n. x ×n y When the remaining space in B(i,j,:) is greater than l z M0(i,j) is 1 when the condition is met, and 0 otherwise.

[0036] Step 3, regarding the placement direction in Step 2 [l x ,l y ,l z ], using size l x ×l y A two-dimensional convolution kernel k, with values ​​all equal to 1, is used to perform a convolution operation M0*k on matrix M0, resulting in a matrix M of the same size as M0. k For Mk Each element M in k (i,j), if M k The value of (i,j) is less than l. x ·l y (l x and l y If the product of the two is 0, then set it to 0; otherwise set it to 1.

[0037] Step 4, remove M k The positions p:{(i,j)} corresponding to all elements with a value of 1 in (i,j) are used to filter ε·n from p using the farthest point sampling method. p Sample p ε (ε is the sampling rate, n) p (where p is the number of elements), where the sampling rate ε = min(ε base ·(1+v r ) λε ,ε max ), where ε base Take 0.15, ε max Take 0.8, λ ε Choosing 3, using this dynamic sampling rate can improve the efficiency of finding the optimal placement.

[0038] Step 5: Repeat steps 2, 3, and 4 until all 6 placement orientations have been calculated, obtaining all sampled placement cases P:{(l i ,l j ,l k ,i,j)}(l i l j and l k (This refers to the placement orientation in step 2).

[0039] Step 3: Calculate the placement evaluation index for each placement scenario and select the placement scenario with the highest evaluation. The specific implementation steps are as follows:

[0040] Step 1: For each of the placement scenarios P, calculate the space occupancy matrix B′ of the boxes after placement according to this scenario, and calculate the centroid position g of the occupied space in matrix B′: [g x ,g y ,g z The distance index c1 to the origin o of the box space coordinate system is given by c1 = |g·([n x ,n y ,n z ] T ) λ1 | λ2 , where λ1 and λ2 are 0.29 and 0.4 respectively;

[0041] Step 2, calculate the actual space occupancy index c2 after the sample is placed. The specific expression for c2 is c2=(∑(B′-B) / (l x ·l y ·l z )) λ3 , where λ3 is 0.2;

[0042] Step 3: Calculate the space occupancy index c3 within the box after sample placement using a 3D convolution kernel k3 with all element values ​​of 1, and a size m equal to the estimated minimum sampled rock sample volume. x ×m y ×m z Perform convolution operations B*k3 and B′*k3 respectively to obtain a matrix B of the same size as B. k3 and B′ k3 For B k3 and B′ k3 Each element B in k3 (i,j,k) and B′ k3 (i,j,k), if the value is greater than 0, it is set to 0, otherwise it is set to 1. Then the specific expression of c3 is c3=(∑(B k3 -B′ k3 ) / (l x ·l y ·l z )) λ4 , where λ4 is 0.29;

[0043] Step 4: Calculate the similarity index c4 between the box size and the sample placement orientation. The specific expression for c4 is c4 = (2 - ([n x ,n y ,n z ]·[l x ,l y ,l z ] T ) / (|[n x ,n y ,n z ]|·|[l x ,l y ,l z ]|)) λ5 , where λ5 is 0.3;

[0044] Step 5: Calculate the deviation index c5 between the center position of the sample and the center position of the contact surface of the placement area. The specific expression for c5 is c5 = (1 + (|s p -s s |)·(1-v r ) / |s p |) λ6 , where s pThese are the coordinates of the sample center in the sample volume coordinate system, s s It is the coordinate of the center position of the contact surface of the area where the sample is placed in the volumetric coordinate system, and λ6 is taken as 0.2;

[0045] Step 6: Calculate the comprehensive evaluation index C: {c1·c2·c3·c4·c5} for all placement scenarios. This index comprehensively evaluates the space occupancy within the box after placement. The placement scenario corresponding to the minimum value in C is the highest-rated placement scenario, and the sample is placed in the corresponding position. For example... Figure 1 The image shows the placement of samples 1 through 8 inside the box.

[0046] Step four: Repeat steps one, two, and three until the collected new rock samples cannot fit into the box. Figure 2 The diagram shows the placement of samples 21 to 28 within the bin. This example placed a total of 28 samples. On a computer configured with Windows 10 and an i7-9750H CPU, the average computation time for each placement was 9.2 seconds, and the bin space utilization rate was 82.7%.

[0047] The technical problem this invention aims to solve is to design an online binning and placement method for extraterrestrial object sampling, enabling online device placement with only the current sample size known, thereby improving binning efficiency. This invention discloses an online binning and placement method for extraterrestrial object sampling that simplifies the sampled rock sample into a regular cuboid, reconstructs the internal space of the bin using a depth camera, uses farthest point sampling to initially screen feasible placement locations and reduce computational load, and designs a sample placement evaluation index function that considers factors such as the current sample placement sparsity, placement direction, placement position, placement center of gravity, actual space occupancy within the bin, and space occupied by the sample within the bin. The method selects the placement location with the highest evaluation, achieving high-volume-rate binning and placement.

[0048] 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 possible changes and modifications to the technical solutions of the present invention by utilizing the methods and techniques disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

Claims

1. A method for online packing and storage of extraterrestrial object samples, characterized in that, The sampled rock samples were simplified into regular cuboids. The space inside the box was reconstructed using a depth camera. Feasible placement locations were initially screened using sampling at the farthest point. An evaluation index function for sample placement was designed, including indicators such as distance, actual space occupancy inside the box, space occupancy for placement inside the box, size similarity, and center deviation. The placement location with the highest evaluation was selected to achieve high volumetric efficiency in box packing and placement. Design a sample placement evaluation index function, select the placement position with the highest evaluation, and achieve high-volume packing and organization. The specific implementation steps are as follows: Step 1: For each of the placement scenarios P, calculate the space occupancy matrix Bσ of the boxes after placement according to this scenario, and calculate the centroid position g of the occupied space in matrix Bσ: [g x ,g y ,g z The distance index c1 to the origin o of the box space coordinate system; Step 2: Calculate the actual space occupied in the box after the sample is placed, index c2; Step 3: Calculate the space occupancy index c3 within the box after sample placement using a 3D convolution kernel k3 with all element values ​​of 1, and a size m equal to the estimated minimum sampled rock sample volume. x ×m y ×m z Perform convolution operations B*k3 and Bσ*k3 respectively to obtain a matrix B of the same size as B. k3 and Bσ k3 For B k3 and B′ k3 Each element B in k3 (i,j,k) and B′ k3 (i,j,k), if the value is greater than 0, set it to 0, otherwise set it to 1; Step 4: Calculate the similarity index c4 between the box size and the sample placement orientation; Step 5: Calculate the deviation index c5 between the center position of the sample and the center position of the contact surface of the placement area; Step 6: Calculate the comprehensive evaluation index C:{c1·c2·c3·c4·c5} for all placement scenarios. This index comprehensively evaluates the occupancy of the space inside the box after placement. The placement scenario with the minimum value in C is the highest-rated placement scenario, and the sample is placed in the corresponding position.

2. The method for online packing and placement of extraterrestrial object samples according to claim 1, characterized in that, The specific steps for reconstructing the internal space of the box using a depth camera are as follows: 2.1 The continuous internal space of the box is discretized into a discrete space of unit length d, and this space is described by a three-dimensional matrix B. The X, Y, and Z directions of matrix B represent the length, width, and height of the box, respectively. The three-dimensional dimensions of matrix B are n. x ×n y ×n z The value in matrix B indicates whether the space inside the box corresponding to that position is occupied, with 1 indicating occupied and 0 indicating unoccupied. 2.2 Use a depth camera to observe the box from a vertical direction and obtain the height map M of the area inside the box. h , height map M h The space occupancy matrix B mapped to the box.

3. The method for online packing and placement of extraterrestrial object samples according to claim 1, characterized in that, Using the farthest point sampling to initially screen feasible placement locations and reduce computational load, the specific implementation steps are as follows: 3.1 The sampled rock samples are simplified into regular cuboids and discretized to obtain three-dimensional dimensions l. x ×l y ×l z A rectangular prism; 3.2 There are 6 possible placement orientations for a cuboid in the box [l] x ,l y ,l z ]、[l x ,l z ,l y ]、[l y ,l x ,l z ]、[l y ,l z ,l x ]、[l z ,l x ,l y ]、[l z ,l y ,l x For each placement orientation, calculate the placement matrix M0, where the size of matrix M0 is n. x ×n y When the remaining space in B(i,j,:) is greater than l z M0(i,j) is 1 when the condition is met, and 0 otherwise. 3.3 Regarding the placement direction in step 3.2 [l] x ,l y ,l z ], using size l x ×l y A two-dimensional convolution kernel k, with values ​​all equal to 1, is used to perform a convolution operation M0*k on matrix M0, resulting in a matrix M of the same size as M0. k For M k Each element M in k (i,j), if M k The value of (i,j) is less than l. x ·l y (l x and l y If the product of the two is 0, then set it to 0; otherwise set it to 1. 3.4 Take out M k The positions p:{(i,j)} corresponding to all elements with a value of 1 in (i,j) are used to filter ε·n from p using the farthest point sampling method. p Sample p ε ε is the sampling rate, n p Let p be the number of elements; 3.5 Repeat steps 3.2, 3.3, and 3.4 until all six placement orientations have been calculated, obtaining all sampled placement cases P:{(l i ,l j ,l k ,i,j)},l i l j and l k This refers to the placement orientation in step 3.

2.

4. The method for online packing and placement of extraterrestrial object samples according to claim 1, characterized in that, The specific expression for c1 is: c1 = |g·([n x ,n y ,n z ] T ) λ1 | λ2 , where λ1 and λ2 are the power coefficients of the distance index, respectively.

5. The method for online packing and placement of extraterrestrial object samples according to claim 1, characterized in that, The specific expression for c2 is c2=(∑(B′-B) / (l x ·l y ·l z )) λ3 , where λ3 is the power coefficient of the actual space occupancy index inside the box.

6. The method for online packing and storage of extraterrestrial body samples according to claim 1, characterized in that, The expression for c3 is c3=(∑(B k3 -B′ k3 ) / (l x ·l y ·l z )) λ4 , where λ4 is the power coefficient of the space occupied inside the box.

7. The method for online packing and placement of extraterrestrial object samples according to claim 1, characterized in that, The specific expression for c4 is c4 = (2 - ([n]). x ,n y ,n z ]·[l x ,l y ,l z ] T ) / (|[n x ,n y ,n z ]|·|[l x ,l y ,l z ]|)) λ5 , where λ5 is the power coefficient of the size similarity index.

8. The method for online packing and placement of extraterrestrial object samples according to claim 1, characterized in that, The expression for c5 is c5 = (1 + (|s p -s s |)·(1-v r ) / |s p |) λ6 , where s p These are the coordinates of the sample center in the sample volume coordinate system, s s λ6 is the coordinate of the center position of the contact surface of the area where the sample is placed in the volumetric coordinate system, and λ6 is the power coefficient of the center deviation index.

9. The method for online packing and placement of extraterrestrial object samples according to claim 3, characterized in that, In step 3.4, the sampling rate ε = min(ε base ·(1+v r ) λε ,ε max ), where ε base Based on the sampling rate, ε max For the maximum sampling rate, v r λ represents the current occupancy rate of the box's internal space. ε To extend the power coefficients, using this dynamic sampling rate ε can improve the efficiency of finding the optimal placement.

Images