CN114355956A - Underwater robot path planning method based on three-dimensional point cloud - Google Patents
Underwater robot path planning method based on three-dimensional point cloud Download PDFInfo
- Publication number
- CN114355956A CN114355956A CN202111610657.6A CN202111610657A CN114355956A CN 114355956 A CN114355956 A CN 114355956A CN 202111610657 A CN202111610657 A CN 202111610657A CN 114355956 A CN114355956 A CN 114355956A
- Authority
- CN
- China
- Prior art keywords
- point
- path
- candidate
- dimensional
- euclidean distance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000005070 sampling Methods 0.000 claims abstract description 78
- 238000001514 detection method Methods 0.000 claims description 91
- 230000005484 gravity Effects 0.000 claims description 11
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 230000009191 jumping Effects 0.000 claims description 6
- 238000005381 potential energy Methods 0.000 claims description 6
- 239000004576 sand Substances 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 3
- 238000010845 search algorithm Methods 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 abstract description 22
- 238000002474 experimental method Methods 0.000 abstract description 6
- 238000007781 pre-processing Methods 0.000 abstract description 2
- 230000007547 defect Effects 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 3
- 230000010354 integration Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000011664 nicotinic acid Substances 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
Images
Abstract
The invention provides an underwater robot path planning method based on three-dimensional point cloud. The method comprises the steps of preprocessing point cloud data, denoising and downsampling the point cloud data, and reducing memory burden by reducing huge data volume. During path planning, the defects of random sampling and long path of an RRT algorithm are overcome, and an improved method based on the RRT is provided. The method provided by the invention realizes the rapid generation of the obstacle avoidance path directly in the three-dimensional point cloud picture, and the reliability and the effectiveness of the method are proved through experiments.
Description
Technical Field
The invention belongs to the technical field of underwater robot path planning, and particularly relates to an underwater robot path planning method based on three-dimensional point cloud.
Background
With the development of underwater robot technology, underwater robots have been generally applied to the fields of underwater inspection, monitoring, exploration, search and salvage, scientific investigation and the like, and gradually become the main force in underwater search and rescue equipment by virtue of the remarkable advantages of strong operation capability, convenience in man-machine real-time interaction and the like. However, when a task is executed in a complex underwater environment, the underwater robot is often required to quickly reach a target location, the effectiveness and timeliness of path planning are particularly important, and how to quickly find an obstacle avoidance path in an unknown environment becomes a research hotspot in related fields.
The path planning algorithm is the core of path planning, and from the traditional algorithm to the combination with the bionic algorithm, the path planning intelligent algorithm has been developed greatly. For offline path planning, it is necessary to know surrounding environment information, such as model parameters, topology, etc., in advance, and the environment information is mostly static information, so as to deal with the problem, a common algorithm is as follows: dijkstra algorithm, a algorithm, etc. The online path planning is to construct an environment model in real time, and can be re-planned even in case of emergency, and meanwhile, the online path planning has good real-time performance, and the adopted algorithm is generally as follows: artificial potential field method (APF), fast search random tree (RRT), probabilistic map method (PRM), etc. The RRT algorithm can perform collision detection on the sampling points in an unknown high-dimensional environment, so that an obstacle avoidance path is effectively generated, but the generated path is not optimal.
The three-dimensional point cloud is obtained through some special scanning devices, such as laser radars, Kinect sensors, sonar devices and the like, the three-dimensional point cloud data is composed of discrete points representing three-dimensional space coordinates, and can also carry information such as RGB (red, green, blue), depth and the like, so that the environment information can be really and specifically restored, and the establishment of the environment by using the three-dimensional point cloud is widely researched at present. Then, due to the huge amount of point cloud data and the technical limitation, navigation and path planning based on three-dimensional point cloud are problems to be solved urgently.
Disclosure of Invention
The invention provides a path planning method of an underwater robot based on three-dimensional point cloud, aiming at overcoming the problem of path planning in an underwater three-dimensional environment and enabling the underwater robot to quickly and accurately reach a target location.
The invention provides an underwater robot path planning method based on three-dimensional point cloud, which comprises the following specific steps:
step 1: denoising the original three-dimensional point cloud to obtain a denoised three-dimensional point cloud, and downsampling the denoised three-dimensional point cloud to obtain a downsampled three-dimensional point cloud;
step 2: setting a three-dimensional space range according to the down-sampled three-dimensional point cloud obtained in the step 1;
and step 3: setting a starting point of a path and a target point of the path in the three-dimensional space range in the step 2, and simultaneously setting a step length, a bias sampling probability, a detection radius and a search radius;
and 4, step 4: constructing a current path point set, and taking a starting point of a path as a first path point of the current path point set;
and 5: and (3) in the three-dimensional space range in the step (2), carrying out random sampling according to the offset sampling probability to obtain sampling points.
Step 6: traversing each path point in the current path point set, calculating the Euclidean distance between each path point in the current path point set and the sampling point, selecting the path point with the minimum Euclidean distance from the sampling point in the current path point set, constructing a gravitation function of the target point of the path in the step 3 to the path point with the minimum Euclidean distance from the sampling point in the current path point set, and determining the position of the candidate path point by combining the sampling point and the step length in the step 3;
and 7: performing collision detection on the candidate path point in the step 6 and the down-sampled three-dimensional point cloud in the step 2, if the detection is passed, setting a path point which is selected from the current path point set and has the minimum Euclidean distance with the candidate path point as a father node of the candidate path point, and jumping to the step 8; if the detection is not passed, re-determining the position of the candidate path point according to the sampling point in the step 5, performing the collision detection on the candidate path point again, if the detection is passed, selecting the path point with the minimum Euclidean distance from the candidate path point in the current path point set as a father node of the candidate path point, and performing the step 8, otherwise, jumping to the step 5;
and 8: and traversing the path point set, calculating the Euclidean distance between each path point in the path point set and the candidate path point, selecting the point with the distance from the candidate path point smaller than the search radius in the step 3, and adding the point to the adjacent point set.
And step 9: traversing the adjacent point set obtained in the step 8, sequentially calculating the sum of the Euclidean distance from the adjacent point in the adjacent point set to the starting point of the path in the step 3 and the Euclidean distance from the adjacent point to the candidate path point, if the sum of the distances is smaller than the distance from the candidate path point to the starting point, performing collision detection on the connecting line between the adjacent point with the minimum sum of the distances and the candidate path point, if the detection is passed, continuously performing collision detection on the connecting line between the parent node of the adjacent point with the minimum sum of the distances and the candidate path point according to the triangle inequality principle, if the detection is passed, setting the parent node of the adjacent point with the minimum sum of the distances as the parent node of the candidate path point, if the detection is not passed, setting the adjacent point with the minimum sum of the distances as the parent node of the candidate path point, and adding the candidate path point in the step 6 into the path point set in the step 4;
step 10: traversing the adjacent point set obtained in the step 8, sequentially calculating the sum of the Euclidean distance from the adjacent point in the adjacent point set to the candidate route point and the Euclidean distance from the candidate route point to the starting point, if the sum of the distances is smaller than the Euclidean distance from the adjacent point to the starting point in the step 3, performing collision detection on a connecting line between the adjacent point and the candidate route point, if the detection is passed, continuously performing collision detection on a connecting line between the adjacent point and a father node of the candidate route point according to a triangle inequality principle, if the detection is passed, setting the father node of the candidate route point as a father node of the adjacent point, and if the detection is not passed, taking the candidate route point as the father node of the adjacent point.
Step 11: and repeating the processes from the step 5 to the step 10 until the Euclidean distance between the candidate path point and the target point in the step 3 is smaller than the step length in the step 9, stopping sampling, taking the candidate path point as a parent node of the target point, and adding the target point into the path point set in the step 4. And (4) searching a father node from the target point and connecting the father node until the path starting point in the step (3) is retrieved.
Preferably, the three-dimensional point cloud after down-sampling in step 1 is:
Cloud＝{(xi，yi，zi)|i∈(1，N)}
wherein the content of the first and second substances,n is the number of the points in the down-sampled three-dimensional point cloud, (x)i，yi，zi) Is the three-dimensional coordinate, x, of the ith point in the down-sampled three-dimensional point cloudiIs the X-axis coordinate, y of the ith point in the down-sampled three-dimensional point cloudiIs the Y-axis coordinate, z, of the ith point in the down-sampled three-dimensional point cloudiThe Z-axis coordinate of the ith point in the three-dimensional point cloud after down sampling;
preferably, the three-dimensional space range in step 2 is:
Space＝{(Xmin，Xmax)|(Ymin，Ymax)|(Zmin，Zmax)}
wherein, XminIs { (x)i，yi，zi) I ∈ (1, N) } minimum value of X-axis coordinate in XmaxIs { (x)i，yi，zi) I ∈ (1, N) } maximum value of X-axis coordinate, YminIs { (x)i，yi，zi) I ∈ (1, N) } minimum value of Y-axis coordinate in Y-axis coordinate, YmaxIs { (x)i，yi，zi) I ∈ (1, N) } maximum value of Y-axis coordinate, ZminIs { (x)i，yi，zi) I ∈ (1, N) } maximum value of Z-axis coordinate, ZmaxIs { (x)i，yi，zi) I belongs to the maximum value of the Z-axis coordinate in (1, N) };
preferably, the coordinates of the starting point of the path in step 3 are: pstart(ustart，vstart，wstart)
Wherein u isstartX-axis coordinate of starting point, vstartY-axis coordinate as starting point, wstartA Z-axis coordinate of a starting point, and satisfies:
and 3, the coordinates of the target point of the path are as follows: pgoal(ugoal，vgoal，wgoal)
Wherein u isgoalAs the X-axis coordinate of the target point, vgoalAs the Y-axis coordinate of the target point, wgoalIs the Z-axis coordinate of the target point and satisfies:
step 3, the step length is gamma;
step 3, the bias sampling probability is p, and p belongs to (0, 1);
step 3, the collision detection radius is R;
step 3, the search radius is r;
preferably, the current path point set in step 4 is:
Path＝{Pi(ui，vi，wi)|i∈(1，K)}
wherein K is the number of path points in the current path point set, (u)i，vi，zi) Is the three-dimensional coordinate of the ith point in the set of path points, uiIs the X-axis coordinate, v, of the ith point in the set of path pointsiIs the Y-axis coordinate of the ith point in the set of path points, wiThe Z-axis coordinate of the ith point in the path point set is obtained;
preferably, in step 5, the random sampling is performed according to the offset sampling probability to obtain sampling points, which is as follows:
generating a random number a belonged to (0, 1), and if a is greater than the bias probability P, generating a random three-dimensional coordinate point PrandAs sampling point, otherwise, selecting target point P of pathgoalAs sampling points in step 5:
preferably, in step 6, the path point with the minimum euclidean distance to the sampling point is:
Pn(un，vn，wn)
wherein u isnIs the X-axis coordinate of the point, vnIs the Y-axis coordinate of the point, wnIs the Z-axis coordinate of the point
Step 6, constructing a gravity function of the target point of the path in step 3 to the path point with the minimum euclidean distance with the candidate path point in the current path point set, specifically as follows:
step 3 target point P of the pathgoalThe gravitational potential energy generated for the path point with the minimum Euclidean distance from the sampling point in the current path point set is as follows:
wherein k is an adaptive gravity coefficient represented by PgoalAnd PnD is PgoalAnd PnEuclidean distance between:
d＝||Pgoal-Pn||
wherein, | | | is modular operation, calculate the Euclidean distance between two points;
obtaining P by gradient of gravitational potential energygoalTo PnThe generated attraction force is as follows:
F＝k×||Pgoal-Pn||
thereby constructing a target point PgoalTo PnThe resulting gravitational function is:
wherein, Pgoal-PnThe vector operation specifically comprises the following steps:
Pgoal-Pn＝(ugoal-un，vgoal-vn，wgoal-wn)
step 6 theCandidate waypoints are Pnew(unew，vnew，wnew)
Wherein u isnewAs X-axis coordinates of candidate waypoints, vnewAs Y-axis coordinates of candidate waypoints, wnewIs the Z-axis coordinate of the candidate waypoint.
And 6, combining the sampling point with the step length in the step 3, determining the coordinates of the candidate path points:
wherein k is the adaptive gravity coefficient, gamma is the step length in step 3, and PsFor the sample points, P, of step 5nFor the path point with the minimum Euclidean distance from the sampling point, P, described in step 6goalFor the target point of step 3, | | Ps-PnI is PsAnd PnEuclidean distance of, | | Pgoal-PnI is PgoalAnd PnThe euclidean distance of (c).
Preferably, the collision detection method in step 7 is as follows:
finding out points in the down-sampled three-dimensional point cloud, wherein the Euclidean distance between the point in the down-sampled three-dimensional point cloud and the candidate path point in the step 6 is smaller than the detection radius R in the step 3 by adopting an octree radius neighborhood search algorithm, and if the number of the found points is 0, the detection is passed; if the number of the found points is more than 0, the detection is not passed.
Step 7, the method for re-determining the positions of the candidate path points according to the sampling points comprises the following steps:
wherein gamma is the step length of step 3, PsFor the sample points, P, of step 5nIs the path point with the minimum Euclidean distance from the sampling point, | | P in the step 6s-PnI is PsAnd PnThe Euclidean distance of;
as a preference, the first and second liquid crystal compositions are,step 8, the adjacent points are set as Unear＝{Qi(ui，vi，wi)|i∈(1，n)}
Wherein n is the number of path points in the current path point set, (u)i，vi，zi) Is the three-dimensional coordinate of the ith point in the adjacent point set, uiIs the X-axis coordinate, v, of the ith point in the set of adjacent pointsiIs the Y-axis coordinate of the ith point in the adjacent point set, wiThe Z-axis coordinate of the ith point in the path point set is satisfied with the following conditions:
wherein u isnewFor the X-axis coordinate, v, of the candidate waypoint of step 6newFor the Y-axis coordinate, w, of the candidate waypoint of step 6newAnd (4) the Z-axis coordinate of the candidate path point in the step (6), and r is the search radius in the step (3).
Preferably, the euclidean distance between the neighboring points in the neighboring point set in step 9 and the starting point of the path in step 3 is calculated as:
step 9.1, if the adjacent point has a father node, calculating the Euclidean distance between the adjacent point and the father node;
step 9.2, if the father node of the adjacent point still has the father node, calculating the Euclidean distance between the father node and the father node, and adding the Euclidean distance with the last Euclidean distance;
and 9.3, repeating the step 9.2 until the father node is the starting point of the path in the step 3, and finally obtaining the Euclidean distance from the adjacent point to the starting point of the path in the step 3.
Step 9, the method for performing collision detection on the connecting line between the adjacent point with the minimum sum of distances and the candidate path point includes:
and (3) taking a detection point every step length gamma in the step (3) on the connecting line between the two points, carrying out the collision detection on the detection points in the step (7), if all the detection points pass, the collision detection of the connecting line passes, otherwise, the detection does not pass.
According to the principle of triangle inequality:
||Pnew-Pnear||+||Pnear-PnearParent||＞||Pnew-PnearParent||
wherein, PnewFor the candidate waypoints, P, of step 9nearFor the adjacent point of minimum distance, P, of step 9nearParentThe parent node of the neighboring point with the minimum distance in the step 9.
Preferably, the step 10 of continuing performing collision detection on the connection line between the neighboring point and the parent node of the candidate waypoint according to the triangle inequality principle specifically includes:
||PnewParent-Pnew||+||Pnew-Pnear||＞||PnewParent-Pnear||
wherein, PnewParentIs a parent node, P, of the candidate waypoint of step 9newFor the candidate waypoints, P, of step 9nearThe neighboring points are step 10.
Overall, the present invention enables the following gains to be achieved compared to the prior art:
the method is realized by using C + + based on visual Stidio 2015 under Windows, the PCL point cloud library is used for completing the whole method, cross-platform and cross-software are not needed, path planning can be directly completed in a three-dimensional point cloud map, and integration of point cloud processing and path planning is realized.
The point cloud preprocessing provided by the invention comprises denoising and down sampling, so that the overall data volume of the point cloud can be greatly reduced, and the memory burden is greatly reduced; in the collision detection, the combined octree is provided, all points in the point cloud do not need to be subjected to collision detection, and the radius field search is used, so that the collision detection speed can be increased, and the calculation burden is greatly reduced.
The invention provides an underwater robot path planning method based on three-dimensional point cloud, and provides an improved algorithm based on RRT (remote distance transform). A gravitational component is added to enable a sampling point to reach a target point more quickly, and by utilizing a triangle inequality principle, redundant points in a path are skipped, and the path is directly connected to a higher father node to enable a generated path to be shorter. Compared with the RRT algorithm, the method has the advantages that the generated path is shorter and smoother, the generated path is suitable for the movement control of the underwater robot, and the reliability and the effectiveness of the improved algorithm are proved through experiments.
Drawings
FIG. 1: and (4) three-dimensional point cloud before denoising.
FIG. 2: and denoising the three-dimensional point cloud.
FIG. 3: and (4) down-sampling the three-dimensional point cloud.
FIG. 4: and introducing a gravity function to determine a schematic diagram of the candidate path point.
FIG. 5: and re-determining the schematic diagram of the candidate path point.
FIG. 6: and selecting a father node schematic diagram for the candidate path points according to the triangle inequality.
FIG. 7: and selecting a father node schematic diagram for the adjacent points according to the triangle inequality.
FIG. 8: the invention provides a flow chart of an underwater robot path planning method based on three-dimensional point cloud.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail and completely with reference to the accompanying drawings and examples, it being understood that the specific embodiments described herein are only for the purpose of illustrating the present invention and are not to be construed as limiting the present invention.
The original three-dimensional point cloud used in the embodiment of the invention is data collected in a real underwater scene by adopting BlueView 5000. BlueView5000 is an underwater three-dimensional panoramic imaging sonar which can generate high-resolution images of underwater topography, structures and targets. The sonar adopts a compact low-weight design, is convenient to mount on a tripod or an ROV, and can generate three-dimensional point cloud of underwater scenes only by touching a button through three-dimensional scanning. The scanning sonar head and the integrated pan-tilt can generate sector scanning and spherical scanning data.
The language used by the invention is C + + language, visual Stidio 2015 under Windows is realized by relying on a PCL library, and the integration of point cloud pretreatment and underwater robot path planning is realized. The method comprises the steps of reducing the data volume of point clouds through denoising and downsampling, visualizing the point clouds and octree a data structure, so that the memory and the calculation burden of a system are saved. And then, an improved RRT algorithm is adopted to realize path planning in the three-dimensional point cloud picture, and experiments prove that the path generated by the method is faster and shorter and is more suitable for the movement control of the underwater robot.
The invention provides an underwater robot path planning method based on three-dimensional point cloud, which comprises the following steps:
step 1: denoising the original three-dimensional point cloud to obtain a denoised three-dimensional point cloud, and downsampling the denoised three-dimensional point cloud to obtain a downsampled three-dimensional point cloud;
in the embodiment of the invention, point cloud is denoised, a statistical analysis is carried out on the neighborhood of each point by utilizing an SOR algorithm, namely a statistical filter, and the points which do not accord with a certain standard are trimmed. Sparse outlier removal methods are based on the calculation of the distribution of distances from points to nearby points in the input data. For each point, its average distance to all its nearby points is calculated. Assuming that the result is a gaussian distribution whose shape is determined by the mean and standard deviation, points whose mean distance is outside the standard range (defined by the global distance mean and variance) can be defined as outliers and removed from the data set.
The point cloud is down-sampled, a VoexlGrid filter, namely a voxel filter, is utilized to create a three-dimensional voxel grid (the voxel grid can be imagined as a set of tiny space three-dimensional cubes) through input point cloud data, then in each voxel, the barycenters of all points in the voxel are used for approximately displaying other points in the voxel, so that all points in the voxel are finally represented by one barycenter point, and the filtered point cloud is obtained after all voxels are processed.
In the embodiment of the invention, fig. 1 is an original three-dimensional point cloud, fig. 2 is a denoised three-dimensional point cloud, and fig. 3 is a down-sampled three-dimensional point cloud. By contrast, the noise point of the denoised point cloud is obviously reduced, the data volume of the point cloud subjected to down-sampling is obviously reduced, and the data volume is shown in the following table:
| amount of data before processing | Post-processing data volume | |
| De-noising | 2089731 | 1848043 |
| Down sampling | 1848043 | 304850 |
Step 1, the down-sampled three-dimensional point cloud is:
Cloud＝{(xi，yi，zi)|i∈(1，N)}
in the embodiment of the invention, N-304850 is the number of the midpoints of the down-sampled three-dimensional point cloud, (x)i，yi，zi) Is the three-dimensional coordinate, x, of the ith point in the down-sampled three-dimensional point cloudiIs the X-axis coordinate, y of the ith point in the down-sampled three-dimensional point cloudiIs the Y-axis coordinate, z, of the ith point in the down-sampled three-dimensional point cloudiThe Z-axis coordinate of the ith point in the three-dimensional point cloud after down sampling;
step 2: setting a three-dimensional space range according to the down-sampled three-dimensional point cloud obtained in the step 1;
step 2, the three-dimensional space range is as follows:
Space＝{(Xmin，Xmax)|(Ymin，Ymax)|(Zmin，Zmax)}
in the examples of the present invention, XminIs { (x) { (4.0 {)i，yi，zi) I ∈ (1, N) } minimum value of X-axis coordinate in Xmax2.5 is { (x)i，yi，zi) I ∈ (1, N) } maximum value of X-axis coordinate, Ymin-18.0 is { (x)i，yi，zi) I ∈ (1, N) } minimum value of Y-axis coordinate in Y-axis coordinate, Ymax-8.0 is { (x)i，yi，zi) I ∈ (1, N) } maximum value of Y-axis coordinate, ZminIs { (x) { (2.0 {)i，yi，zi) I ∈ (1, N) } maximum value of Z-axis coordinate, Zmax2.5 is { (x)i，yi，zi) I belongs to the maximum value of the Z-axis coordinate in (1, N) };
and step 3: setting a starting point of a path and a target point of the path in the three-dimensional space range in the step 2, and simultaneously setting a step length, a bias sampling probability, a detection radius and a search radius;
and 3, the coordinates of the starting point of the path are as follows: pstart(ustart，vstart，wstart)
In the examples of the present invention, ustartX-axis coordinate with 1.5 as starting point, vstartY-axis coordinate with-13 as starting point, wstart-1.5, and satisfies:
and 3, the coordinates of the target point of the path are as follows: pgoal(ugoal，vgoal，wgoal)
In the examples of the present invention, ugoalX-axis coordinate of target point, v ═ 3.5goal3.5 is the Y-axis coordinate of the target point, wgoal-1.0 is the Z-axis coordinate of the target point and satisfies:
in the embodiment of the present invention, the step size γ is 0.3;
the bias sampling probability is p is 0.05;
the collision detection radius is R ═ 0.3;
the search radius is r-0.5;
and 4, step 4: constructing a current path point set, and taking a starting point of a path as a first path point of the current path point set;
step 4, the current path point set is as follows:
Path＝{Pi(ui，vi，wi)|i∈(1，K)}
wherein K is the number of path points in the current path point set, (u)i，vi，zi) Is the three-dimensional coordinate of the ith point in the set of path points, uiIs the X-axis coordinate, v, of the ith point in the set of path pointsiIs the Y-axis coordinate of the ith point in the set of path points, wiThe Z-axis coordinate of the ith point in the path point set is obtained;
and 5: and (3) in the three-dimensional space range in the step (2), carrying out random sampling according to the offset sampling probability to obtain sampling points.
And 5, randomly sampling according to the offset sampling probability to obtain sampling points, which specifically comprise the following steps:
generating a random number a belonged to (0, 1), and if a is greater than the bias probability P, generating a random three-dimensional coordinate point PrandAs sampling point, otherwise, selecting target point P of pathgoalAs sampling points in step 5:
step 6: traversing each path point in the current path point set, calculating the Euclidean distance between each path point in the current path point set and the sampling point, selecting the path point with the minimum Euclidean distance from the sampling point in the current path point set, constructing a gravitation function of the target point of the path in the step 3 to the path point with the minimum Euclidean distance from the sampling point in the current path point set, and determining the position of the candidate path point by combining the sampling point and the step length in the step 3;
and 6, the path point with the minimum Euclidean distance from the sampling point is as follows:
Pn(un，vn，wn)
wherein u isnIs the X-axis coordinate of the point, vnIs the Y-axis coordinate of the point, wnIs the Z-axis coordinate of the point
Step 6, constructing a gravity function of the target point of the path in step 3 to the path point with the minimum euclidean distance with the candidate path point in the current path point set, specifically as follows:
step 3 target point P of the pathgoalThe gravitational potential energy generated for the path point with the minimum Euclidean distance from the sampling point in the current path point set is as follows:
wherein k is an adaptive gravity coefficient represented by PgoalAnd PnD is PgoalAnd PnEuclidean distance between:
d＝||Pgoal-Pn||
wherein, | | | is modular operation, calculate the Euclidean distance between two points;
obtaining P by gradient of gravitational potential energygoalTo PnThe generated attraction force is as follows:
F＝k×||Pgoal-Pn||
thereby constructing a target point PgoalTo PnThe resulting gravitational function is:
wherein, Pgoal-PnThe vector operation specifically comprises the following steps:
Pgoal-Pn＝(ugoal-un，vgoal-vn，wgoal-wn)
step 6, the candidate path point is Pnew(unew，vnew，wnew)
Wherein u isnewAs X-axis coordinates of candidate waypoints, vnewAs Y-axis coordinates of candidate waypoints, wnewIs the Z-axis coordinate of the candidate waypoint.
And 6, combining the sampling point with the step length in the step 3, determining the coordinates of the candidate path points:
wherein k is the adaptive gravity coefficient, gamma is the step length in step 3, and PsFor the sample points, P, of step 5nFor the path point with the minimum Euclidean distance from the sampling point, P, described in step 6goalFor the target point of step 3, | | Ps-PnI is PsAnd PnEuclidean distance of, | | Pgoal-PnI is PgoalAnd PnThe euclidean distance of (c).
FIG. 4 is a schematic diagram of determining candidate waypoints by introducing a gravity function.
And 7: performing collision detection on the candidate path point in the step 6 and the down-sampled three-dimensional point cloud in the step 2, if the detection is passed, setting a path point which is selected from the current path point set and has the minimum Euclidean distance with the candidate path point as a father node of the candidate path point, and jumping to the step 8; if the detection is not passed, re-determining the position of the candidate path point according to the sampling point in the step 5, performing the collision detection on the candidate path point again, if the detection is passed, selecting the path point with the minimum Euclidean distance from the candidate path point in the current path point set as a father node of the candidate path point, and performing the step 8, otherwise, jumping to the step 5;
step 7, the collision detection method comprises the following steps:
finding out points in the down-sampled three-dimensional point cloud, wherein the Euclidean distance between the point in the down-sampled three-dimensional point cloud and the candidate path point in the step 6 is smaller than the detection radius R in the step 3 by adopting an octree radius neighborhood search algorithm, and if the number of the found points is 0, the detection is passed; if the number of the found points is more than 0, the detection is not passed.
In the embodiment of the invention, the octree is established for the point cloud data, so that huge data of the point cloud can be effectively compressed and expressed, the collision detection burden of path planning is greatly reduced, and the time cost is greatly saved. The data volume before and after octree establishment is shown in the following table:
| front of octree | After the octree | |
| Data volume | 304850 | 18720 |
Step 7, the method for re-determining the positions of the candidate path points according to the sampling points comprises the following steps:
wherein gamma is the step length of step 3, PsIs composed of a main body and a lower bodyStep 5 the sample point, PnIs the path point with the minimum Euclidean distance from the sampling point, | | P in the step 6s-PnI is PsAnd PnThe Euclidean distance of;
FIG. 5 is a diagram illustrating re-determining candidate waypoints.
And 8: and traversing the path point set, calculating the Euclidean distance between each path point in the path point set and the candidate path point, selecting the point with the distance from the candidate path point smaller than the search radius in the step 3, and adding the point to the adjacent point set.
Step 8, the adjacent points are set as Unear＝{Qi(ui，vi，wi)|i∈(1，n)}
Wherein n is the number of path points in the current path point set, (u)i，vi，zi) Is the three-dimensional coordinate of the ith point in the adjacent point set, uiIs the X-axis coordinate, v, of the ith point in the set of adjacent pointsiIs the Y-axis coordinate of the ith point in the adjacent point set, wiThe Z-axis coordinate of the ith point in the path point set is satisfied with the following conditions:
wherein u isnewFor the X-axis coordinate, v, of the candidate waypoint of step 6newFor the Y-axis coordinate, w, of the candidate waypoint of step 6newAnd (4) the Z-axis coordinate of the candidate path point in the step (6), and r is the search radius in the step (3).
And step 9: traversing the adjacent point set obtained in the step 8, sequentially calculating the sum of the Euclidean distance from the adjacent point in the adjacent point set to the starting point of the path in the step 3 and the Euclidean distance from the adjacent point to the candidate path point, if the sum of the distances is smaller than the distance from the candidate path point to the starting point, performing collision detection on the connecting line between the adjacent point with the minimum sum of the distances and the candidate path point, if the detection is passed, continuously performing collision detection on the connecting line between the parent node of the adjacent point with the minimum sum of the distances and the candidate path point according to the triangle inequality principle, if the detection is passed, setting the parent node of the adjacent point with the minimum sum of the distances as the parent node of the candidate path point, if the detection is not passed, setting the adjacent point with the minimum sum of the distances as the parent node of the candidate path point, and adding the candidate path point in the step 6 into the path point set in the step 4;
step 9, calculating the euclidean distance from the neighboring points in the neighboring point set to the starting point of the path in step 3 as follows:
step 9.1, if the adjacent point has a father node, calculating the Euclidean distance between the adjacent point and the father node;
step 9.2, if the father node of the adjacent point still has the father node, calculating the Euclidean distance between the father node and the father node, and adding the Euclidean distance with the last Euclidean distance;
and 9.3, repeating the step 9.2 until the father node is the starting point of the path in the step 3, and finally obtaining the Euclidean distance from the adjacent point to the starting point of the path in the step 3.
Step 9, the method for performing collision detection on the connecting line between the adjacent point with the minimum sum of distances and the candidate path point includes:
and (3) taking a detection point every step length gamma in the step (3) on the connecting line between the two points, carrying out the collision detection on the detection points in the step (7), if all the detection points pass, the collision detection of the connecting line passes, otherwise, the detection does not pass.
According to the principle of triangle inequality:
||Pnew-Pnear||+||Pnear-PnearParent||＞||Pnew-PnearParent||
wherein, PnewFor the candidate waypoints, P, of step 9nearFor the adjacent point of minimum distance, P, of step 9nearParentThe parent node of the neighboring point with the minimum distance in the step 9.
FIG. 6 is a schematic diagram illustrating selection of parent nodes for candidate waypoints according to the triangle inequality.
Step 10: traversing the adjacent point set obtained in the step 8, sequentially calculating the sum of the Euclidean distance from the adjacent point in the adjacent point set to the candidate route point and the Euclidean distance from the candidate route point to the starting point, if the sum of the distances is smaller than the Euclidean distance from the adjacent point to the starting point in the step 3, performing collision detection on a connecting line between the adjacent point and the candidate route point, if the detection is passed, continuously performing collision detection on a connecting line between the adjacent point and a father node of the candidate route point according to a triangle inequality principle, if the detection is passed, setting the father node of the candidate route point as a father node of the adjacent point, and if the detection is not passed, taking the candidate route point as the father node of the adjacent point.
According to the principle of triangle inequality:
||PnewParent-Pnew||+||Pnew-Pnear||＞||PnewParent-Pnear||
wherein, PnewParentIs a parent node, P, of the candidate waypoint of step 9newFor the candidate waypoints, P, of step 9nearThe neighboring points are step 10.
FIG. 7 is a diagram illustrating selection of parent nodes for neighbors according to the triangle inequality.
Step 11: and repeating the processes from the step 5 to the step 10 until the Euclidean distance between the candidate path point and the target point in the step 3 is smaller than the step length in the step 9, stopping sampling, taking the candidate path point as a parent node of the target point, and adding the target point into the path point set in the step 4. Searching a father node from the target point and connecting the father node until the path starting point in the step 3 is found back;
fig. 8 is a flowchart of an underwater robot path planning method based on three-dimensional point cloud according to the present invention.
In order to verify the reliability and effectiveness of the proposed underwater robot path planning method based on the three-dimensional point cloud, 100 experiments are carried out on the existing RRT algorithm, RRT algorithm and the algorithm of the invention in the same three-dimensional point cloud, and the average value is calculated, and the result is shown in the following table:
| run time | Mean path | Shortest path | |
| RRT algorithm | 0.78726s | 24.936m | 23.7907m |
| RRT algorithm | 1.37103s | 24.236m | 22.6923m |
| Algorithm of the invention | 1.00957s | 20.7532m | 20.3976m |
Experiments prove that the method provided by the invention is superior to the RRT algorithm in both path length and running time.
In general, the method performs well on the embodiment, is not only suitable for the embodiment, and has universality on any three-dimensional point cloud data. The underwater robot path planning system and method based on the three-dimensional point cloud provided by the invention are proved to be reliable and effective through experiments.
It should be understood that parts of the specification not set forth in detail are prior art;
it should be understood that the above description of the preferred embodiments is given for clarity and not for any purpose of limitation, and that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. An underwater robot path planning method based on three-dimensional point cloud is characterized by comprising the following steps:
step 1: denoising the original three-dimensional point cloud to obtain a denoised three-dimensional point cloud, and downsampling the denoised three-dimensional point cloud to obtain a downsampled three-dimensional point cloud;
step 2: setting a three-dimensional space range according to the down-sampled three-dimensional point cloud obtained in the step 1;
and step 3: setting a starting point of a path and a target point of the path in the three-dimensional space range in the step 2, and simultaneously setting a step length, a bias sampling probability, a detection radius and a search radius;
and 4, step 4: constructing a current path point set, and taking a starting point of a path as a first path point of the current path point set;
and 5: in the three-dimensional space range in the step 2, random sampling is carried out according to the offset sampling probability to obtain sampling points;
step 6: traversing each path point in the current path point set, calculating the Euclidean distance between each path point in the current path point set and the sampling point, selecting the path point with the minimum Euclidean distance from the sampling point in the current path point set, constructing a gravitation function of the target point of the path in the step 3 to the path point with the minimum Euclidean distance from the sampling point in the current path point set, and determining the position of the candidate path point by combining the sampling point and the step length in the step 3;
and 7: performing collision detection on the candidate path point in the step 6 and the down-sampled three-dimensional point cloud in the step 2, if the detection is passed, setting a path point which is selected from the current path point set and has the minimum Euclidean distance with the candidate path point as a father node of the candidate path point, and jumping to the step 8; if the detection is not passed, re-determining the position of the candidate path point according to the sampling point in the step 5, performing the collision detection on the candidate path point again, if the detection is passed, selecting the path point with the minimum Euclidean distance from the candidate path point in the current path point set as a father node of the candidate path point, and performing the step 8, otherwise, jumping to the step 5;
and 8: traversing the path point set, calculating the Euclidean distance between each path point in the path point set and the candidate path point, selecting the point with the distance from the candidate path point smaller than the search radius in the step 3, and adding the point to the adjacent point set;
and step 9: traversing the adjacent point set obtained in the step 8, sequentially calculating the sum of the Euclidean distance from the adjacent point in the adjacent point set to the starting point of the path in the step 3 and the Euclidean distance from the adjacent point to the candidate path point, if the sum of the distances is smaller than the distance from the candidate path point to the starting point, performing collision detection on the connecting line between the adjacent point with the minimum sum of the distances and the candidate path point, if the detection is passed, continuously performing collision detection on the connecting line between the parent node of the adjacent point with the minimum sum of the distances and the candidate path point according to the triangle inequality principle, if the detection is passed, setting the parent node of the adjacent point with the minimum sum of the distances as the parent node of the candidate path point, if the detection is not passed, setting the adjacent point with the minimum sum of the distances as the parent node of the candidate path point, and adding the candidate path point in the step 6 into the path point set in the step 4;
step 10: traversing the adjacent point set obtained in the step 8, sequentially calculating the sum of the Euclidean distance from the adjacent point in the adjacent point set to the candidate route point and the Euclidean distance from the candidate route point to the starting point, if the sum of the distances is less than the Euclidean distance from the adjacent point to the starting point in the step 3, performing collision detection on a connecting line between the adjacent point and the candidate route point, if the detection is passed, continuously performing collision detection on a connecting line between the adjacent point and a father node of the candidate route point according to a triangle inequality principle, if the detection is passed, setting the father node of the candidate route point as the father node of the adjacent point, and if the detection is not passed, taking the candidate route point as the father node of the adjacent point;
step 11: repeating the processes from the step 5 to the step 10 until the Euclidean distance between the candidate path point and the target point in the step 3 is smaller than the step length in the step 9, stopping sampling, taking the candidate path point as a parent node of the target point, and adding the target point into the path point set in the step 4; and (4) searching a father node from the target point and connecting the father node until the path starting point in the step (3) is retrieved.
2. The underwater robot path planning method based on the three-dimensional point cloud of claim 1, wherein the three-dimensional point cloud after down-sampling in the step 1 is:
Cloud＝{(xi,yi,zi)|i∈(1,N)}
wherein, N is the number of the midpoints of the three-dimensional point clouds after down sampling, (x)i,yi,zi) Is the three-dimensional coordinate, x, of the ith point in the down-sampled three-dimensional point cloudiIs the X-axis coordinate, y of the ith point in the down-sampled three-dimensional point cloudiIs the Y-axis coordinate, z, of the ith point in the down-sampled three-dimensional point cloudiThe Z-axis coordinate of the ith point in the down-sampled three-dimensional point cloud is obtained.
3. The underwater robot path planning method based on the three-dimensional point cloud of claim 1, wherein the three-dimensional space range of step 2 is as follows:
Space＝{(Xmin,Xmax)|(Ymin,Ymax)|(Zmin,Zmax)}
wherein, XminIs { (x)i,yi,zi) I ∈ (1, N) } minimum value of X-axis coordinate in XmaxIs { (x)i,yi，zi) I ∈ (1, N) } maximum value of X-axis coordinate, YminIs { (x)i,yi，zi) I ∈ (1, N) } minimum value of Y-axis coordinate in Y-axis coordinate, YmaxIs { (x)i，yi，zi) I ∈ (1, N) } maximum value of Y-axis coordinate, ZminIs { (x)i，yi，zi) I ∈ (1, N) } maximum value of Z-axis coordinate, ZmaxIs { (x)i，yi，zi) I ∈ (1, N) } maximum value of Z-axis coordinate.
4. The underwater robot path planning method based on the three-dimensional point cloud of claim 1, wherein the coordinates of the starting point of the path in the step 3 are as follows: pstart(ustart，vstart，wstart)
Wherein u isstartX-axis coordinate of starting point, vstartY-axis coordinate as starting point, wstartA Z-axis coordinate of a starting point, and satisfies:
and 3, the coordinates of the target point of the path are as follows: pgoal(ugoal，vgoal，wgoal)
Wherein u isgoalAs the X-axis coordinate of the target point, vgoalAs the Y-axis coordinate of the target point, wgoalIs the Z-axis coordinate of the target point and satisfies:
step 3, the step length is gamma;
step 3, the bias sampling probability is p, and p belongs to (0, 1);
step 3, the collision detection radius is R;
and 3, the search radius is r.
5. The underwater robot path planning method based on the three-dimensional point cloud of claim 1, wherein the current path point set of step 4 is:
Path＝{Pi(ui，vi，wi)|i∈(1，K)}
wherein K is the current pathNumber of path points in the set of points, (u)i，vi，zi) Is the three-dimensional coordinate of the ith point in the set of path points, uiIs the X-axis coordinate, v, of the ith point in the set of path pointsiIs the Y-axis coordinate of the ith point in the set of path points, wiIs the Z-axis coordinate of the ith point in the path point set.
6. The underwater robot path planning method based on the three-dimensional point cloud of claim 1, wherein the sampling points are obtained by randomly sampling according to the offset sampling probability in step 5, and specifically the following steps are performed:
generating a random number a belonged to (0, 1), and if a is greater than the bias probability P, generating a random three-dimensional coordinate point PrandAs sampling point, otherwise, selecting target point P of pathgoalAs sampling points in step 5:
7. the underwater robot path planning method based on the three-dimensional point cloud of claim 1, wherein the path point with the minimum Euclidean distance from the sampling point in the step 6 is as follows:
Pn(un，vn，wn)
wherein u isnIs the X-axis coordinate of the point, vnIs the Y-axis coordinate of the point, wnIs the Z-axis coordinate of the point
Step 6, constructing a gravity function of the target point of the path in step 3 to the path point with the minimum euclidean distance with the candidate path point in the current path point set, specifically as follows:
step 3 target point P of the pathgoalThe gravitational potential energy generated for the path point with the minimum Euclidean distance from the sampling point in the current path point set is as follows:
wherein k is an adaptive gravity coefficient represented by PgoalAnd PnD is PgoalAnd PnEuclidean distance between:
d＝||Pgoal―Pn||
wherein, | | | is modular operation, calculate the Euclidean distance between two points;
obtaining P by gradient of gravitational potential energygoalTo PnThe generated attraction force is as follows:
F＝k×||Pgoal―Pn||
thereby constructing a target point PgoalTo PnThe resulting gravitational function is:
wherein, Pgoal―PnThe vector operation specifically comprises the following steps:
Pgoal―Pn＝(ugoal―un，vgoal―vn，wgoal―wn)
step 6, the candidate path point is Pnew(unew，vnew，wnew)
Wherein u isnewAs X-axis coordinates of candidate waypoints, vnewAs Y-axis coordinates of candidate waypoints, wnewThe Z-axis coordinate of the candidate path point is obtained;
and 6, combining the sampling point with the step length in the step 3, determining the coordinates of the candidate path points:
wherein k is the adaptive gravity coefficient, gamma is the step length in step 3, and PsFor the sample points, P, of step 5nFor the path point with the minimum Euclidean distance from the sampling point, P, described in step 6goalFor the target point of step 3, | | Ps―PnI is PsAnd PnEuclidean distance of, | | Pgoal―PnI is PgoalAnd PnThe euclidean distance of (c).
8. The underwater robot path planning method based on the three-dimensional point cloud of claim 1, wherein the collision detection method in step 7 is as follows:
finding out points in the down-sampled three-dimensional point cloud, wherein the Euclidean distance between the point in the down-sampled three-dimensional point cloud and the candidate path point in the step 6 is smaller than the detection radius R in the step 3 by adopting an octree radius neighborhood search algorithm, and if the number of the found points is 0, the detection is passed; if the number of the found points is more than 0, the detection is not passed;
step 7, the method for re-determining the positions of the candidate path points according to the sampling points comprises the following steps:
wherein gamma is the step length of step 3, PsFor the sample points, P, of step 5nIs the path point with the minimum Euclidean distance from the sampling point, | | P in the step 6s―PnI is PsAnd PnThe euclidean distance of (c).
9. The method for planning the path of the underwater robot based on the three-dimensional point cloud of claim 1, wherein the neighboring points are set to be U in step 8near＝{Qi(ui，vi，wi)|i∈(1,n)}
Wherein n is the number of path points in the current path point set, (u)i,vi,zi) Is the three-dimensional coordinate of the ith point in the adjacent point set, uiIs the X-axis coordinate, v, of the ith point in the set of adjacent pointsiIs the Y-axis coordinate of the ith point in the adjacent point set, wiThe Z-axis coordinate of the ith point in the path point set is satisfied with the following conditions:
wherein u isnewFor the X-axis coordinate, v, of the candidate waypoint of step 6newFor the Y-axis coordinate, w, of the candidate waypoint of step 6newAnd (4) the Z-axis coordinate of the candidate path point in the step (6), and r is the search radius in the step (3).
10. The method for planning the path of the underwater robot based on the three-dimensional point cloud of claim 1, wherein the euclidean distance between the neighboring points in the neighboring point set and the starting point of the path in the step 3 in the step 9 is calculated as follows:
step 9.1, if the adjacent point has a father node, calculating the Euclidean distance between the adjacent point and the father node;
step 9.2, if the father node of the adjacent point still has the father node, calculating the Euclidean distance between the father node and the father node, and adding the Euclidean distance with the last Euclidean distance;
step 9.3, repeating step 9.2 until the father node is the starting point of the path in step 3, and finally obtaining the Euclidean distance from the adjacent point to the starting point of the path in step 3;
step 9, the method for performing collision detection on the connecting line between the adjacent point with the minimum sum of distances and the candidate path point includes:
taking a detection point from the step length gamma in the step 3 every time on the connecting line between two points, carrying out the collision detection on the detection points in the step 7, if all the detection points pass, the collision detection of the connecting line passes, otherwise, the detection does not pass;
according to the principle of triangle inequality:
||Pnew―Pnear||+||Pnear―PnearParent||>||Pnew―PnearParent||
wherein, PnewFor the candidate waypoints, P, of step 9nearFor the adjacent point of minimum distance, P, of step 9nearParentA parent node of the adjacent point with the minimum distance in the step 9;
preferably, the step 10 of continuing performing collision detection on the connection line between the neighboring point and the parent node of the candidate waypoint according to the triangle inequality principle specifically includes:
||PnewParent―Pnew||+||Pnew―Pnear||>||PnewParent―Pnear||
wherein, PnewParentIs a parent node, P, of the candidate waypoint of step 9newFor the candidate waypoints, P, of step 9nearThe neighboring points are step 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111610657.6A CN114355956A (en) | 2021-12-27 | 2021-12-27 | Underwater robot path planning method based on three-dimensional point cloud |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111610657.6A CN114355956A (en) | 2021-12-27 | 2021-12-27 | Underwater robot path planning method based on three-dimensional point cloud |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114355956A true CN114355956A (en) | 2022-04-15 |
Family
ID=81100727
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111610657.6A Pending CN114355956A (en) | 2021-12-27 | 2021-12-27 | Underwater robot path planning method based on three-dimensional point cloud |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114355956A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115774452A (en) * | 2023-02-13 | 2023-03-10 | 南京航空航天大学 | Three-dimensional model surface coverage path planning method based on shape constraint |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112650218A (en) * | 2020-12-04 | 2021-04-13 | 国网湖北省电力有限公司检修公司 | Transformer substation inspection route planning method and device based on collision detection |
| CN113325845A (en) * | 2021-05-31 | 2021-08-31 | 的卢技术有限公司 | Unmanned driving path planning method based on hybrid APF and RRT |
-
2021
- 2021-12-27 CN CN202111610657.6A patent/CN114355956A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112650218A (en) * | 2020-12-04 | 2021-04-13 | 国网湖北省电力有限公司检修公司 | Transformer substation inspection route planning method and device based on collision detection |
| CN113325845A (en) * | 2021-05-31 | 2021-08-31 | 的卢技术有限公司 | Unmanned driving path planning method based on hybrid APF and RRT |
Non-Patent Citations (6)
| Title |
|---|
| YANG WANG 等: "Research on stealth route planning based on improved RRT algorithm", 2020 5TH INTERNATIONAL CONFERENCE ON MECHANICAL, CONTROL AND COMPUTER ENGINEERING (ICMCCE), 13 May 2021 (2021-05-13), pages 1491 - 1495 * |
| 张伟民 等: "基于改进RRT*算法的移动机器人路径规划", 华中科技大学学报(自然科学版), vol. 49, no. 1, 31 January 2021 (2021-01-31), pages 31 - 36 * |
| 李智: "基于深度强化学习的无人车路径规划研究", 哈尔滨工业大学硕士论文, 10 December 2021 (2021-12-10), pages 36 - 37 * |
| 涂睿 等: "移动机器人实时采样路径重规划", 计算机工程与应用, vol. 57, no. 20, 15 October 2021 (2021-10-15), pages 157 - 163 * |
| 王先伟 等: "采茶机器人导航避障及路径规划研究", 农业装备与车辆工程, vol. 57, no. 12, 17 December 2019 (2019-12-17), pages 121 - 124 * |
| 胡兵: "基于改进RRT算法的移动机器人路径规划研究", 中国优秀硕士学位论文全文数据库 信息科技辑, 15 April 2020 (2020-04-15), pages 15 - 27 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115774452A (en) * | 2023-02-13 | 2023-03-10 | 南京航空航天大学 | Three-dimensional model surface coverage path planning method based on shape constraint |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11567502B2 (en) | Autonomous exploration framework for indoor mobile robotics using reduced approximated generalized Voronoi graph | |
| CN111429574B (en) | Mobile robot positioning method and system based on three-dimensional point cloud and vision fusion | |
| Li et al. | Deep sensor fusion between 2D laser scanner and IMU for mobile robot localization | |
| Parisotto et al. | Global pose estimation with an attention-based recurrent network | |
| CN108152831B (en) | Laser radar obstacle identification method and system | |
| CN108898676B (en) | Method and system for detecting collision and shielding between virtual and real objects | |
| WO2021082571A1 (en) | Robot tracking method, device and equipment and computer readable storage medium | |
| Rosinol et al. | Incremental visual-inertial 3d mesh generation with structural regularities | |
| He et al. | Vision-based UAV flight control and obstacle avoidance | |
| CN111968229A (en) | High-precision map making method and device | |
| Cui et al. | Real-time dense mapping for self-driving vehicles using fisheye cameras | |
| CN113592891B (en) | Unmanned vehicle passable domain analysis method and navigation grid map manufacturing method | |
| Song et al. | End-to-end learning for inter-vehicle distance and relative velocity estimation in adas with a monocular camera | |
| Liu et al. | An enhanced lidar inertial localization and mapping system for unmanned ground vehicles | |
| Agrawal et al. | PCE-SLAM: A real-time simultaneous localization and mapping using LiDAR data | |
| CN114066773B (en) | Dynamic object removal based on point cloud characteristics and Monte Carlo expansion method | |
| US10445611B1 (en) | Method for detecting pseudo-3D bounding box to be used for military purpose, smart phone or virtual driving based-on CNN capable of converting modes according to conditions of objects and device using the same | |
| CN114355956A (en) | Underwater robot path planning method based on three-dimensional point cloud | |
| Cigla et al. | Gaussian mixture models for temporal depth fusion | |
| Cigla et al. | Image-based visual perception and representation for collision avoidance | |
| Gautam et al. | An experimental comparison of visual SLAM systems | |
| Xue et al. | Real-time 3D grid map building for autonomous driving in dynamic environment | |
| Li et al. | Efficient laser-based 3D SLAM in real time for coal mine rescue robots | |
| Akremi et al. | Visual navigation of uavs in indoor corridor environments using deep learning | |
| Paton et al. | Eyes in the back of your head: Robust visual teach & repeat using multiple stereo cameras |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |