WO2022152263A1 - 斜坡感知方法、装置、机器人和存储介质 - Google Patents
斜坡感知方法、装置、机器人和存储介质 Download PDFInfo
- Publication number
- WO2022152263A1 WO2022152263A1 PCT/CN2022/072121 CN2022072121W WO2022152263A1 WO 2022152263 A1 WO2022152263 A1 WO 2022152263A1 CN 2022072121 W CN2022072121 W CN 2022072121W WO 2022152263 A1 WO2022152263 A1 WO 2022152263A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- slope
- point cloud
- radar points
- radar
- seed
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000012216 screening Methods 0.000 claims abstract description 9
- 238000012790 confirmation Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000008447 perception Effects 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 208000035657 Abasia Diseases 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 208000034423 Delivery Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000009194 climbing Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 235000012054 meals Nutrition 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/481—Constructional features, e.g. arrangements of optical elements
Definitions
- the present application relates to the field of robotics, and in particular, to a slope sensing method, device, robot and computer-readable storage medium.
- the cost pressure of logistics and distribution makes distribution robots gradually applied to various short-distance distribution scenarios such as delivery and meal delivery.
- This unmanned distribution method can not only save labor costs, but also improve the user's interesting experience, and in To a certain extent, the privacy of users' shopping is guaranteed, so the application of delivery robots in short-distance delivery is bound to become more and more extensive.
- the robot may encounter obstacles on the way of delivery, and usually detects obstacles through lidar technology. For safety reasons, once an obstacle is detected, it will stop moving forward. However, in reality, there may also be up and down slopes on the delivery path. As shown in Figure 1, when the slope is large, the laser rays of the lidar will hit the slope, causing the robot to mistake the slope as an obstacle, causing the robot to not Walk or execute preset obstacle avoidance processing. In this case, the robot's execution of obstacle avoidance will cause the robot to be unable to walk up the slope and deviate from the pre-determined delivery route, affecting normal delivery.
- a slope sensing method, apparatus, robot, and computer-readable storage medium are provided.
- a slope sensing method is implemented based on lidar technology, and includes the following steps: performing point cloud clustering on all radar points scanned in the vicinity of the slope to obtain several point cloud clusters; The suspected slope point cloud cluster is screened out from the point cloud clustering; the point cloud cluster belonging to the linear type in the screened out suspected slope point cloud cluster is confirmed as the slope point cloud cluster, and the slope point cloud cluster is clustered. The corresponding lidar scan area of the class is identified as a slope.
- a slope perception device comprising: a clustering unit for performing point cloud clustering on all radar points scanned in the area adjacent to the slope to obtain several point cloud clusters; a screening unit for clustering from the several point clouds The suspected slope point cloud cluster is screened out from the class; the slope confirmation unit is used to confirm the point cloud cluster belonging to the linear type in the screened out suspected slope point cloud cluster as the slope point cloud cluster, and the slope The lidar scan area corresponding to the point cloud clustering is identified as a slope.
- a robot comprising: a memory and a processor; the memory stores executable program codes; the processor coupled with the memory calls the executable program codes stored in the memory to execute the above the slope perception method.
- a computer-readable storage medium having a computer program stored thereon, the computer program, when executed by a processor, implements the slope sensing method as described above.
- Fig. 1 is the light path diagram of the laser ray when the robot provided by the prior art climbs a slope
- FIG. 2 is a schematic flowchart of a slope sensing method provided by the first embodiment of the present application
- FIG. 3 is a clustering effect diagram provided by the first embodiment of the present application.
- FIG. 4 is a schematic flowchart of a slope sensing method provided by a second embodiment of the present application.
- FIG. 5 is a schematic flowchart of a slope sensing method provided by a third embodiment of the present application.
- FIG. 6 is a structural diagram of a slope sensing device provided by a fourth embodiment of the present application.
- FIG. 7 is a structural diagram of a clustering unit provided by the fourth embodiment of the present application.
- FIG. 8 is a structural diagram of a slope confirmation unit provided by a fourth embodiment of the present application.
- FIG. 9 is a schematic structural diagram of a robot provided by a fifth embodiment of the present application.
- the embodiments of the present application are implemented based on the lidar radar technology. First, all the radar points scanned in the vicinity of the slope are subjected to point cloud clustering, and then the suspected slope point cloud cluster is finally screened out, and then the suspected slope point cloud is further selected from the suspected slope point cloud.
- the clustering obtains a cluster of slope point clouds, and the laser radar scanning area corresponding to the cluster of slope point clouds is confirmed as a slope.
- FIG. 2 is a schematic flowchart of a slope sensing method provided by the first embodiment of the present application.
- This slope perception method is based on lidar technology and is mainly applicable to delivery robots in hotels, restaurants and other places, as well as delivery robots and even autonomous vehicles running in fixed places such as logistics parks.
- the method specifically includes:
- the area near the slope refers to the area that is within the preset range near the slope and includes the slope.
- the slope sensing is performed only when the smart device travels to the area near the slope, and the slope does not need to be sensed when traveling outside the area near the slope. The normal travel helps to greatly reduce the amount of calculation and avoid misperception due to misjudgment of the device in the non-slope adjacent area.
- the area near the slope can be determined in various ways, such as using RGBD cameras to identify whether to travel to the area near the slope, or using 3D lidar/multi-line lidar to identify whether to travel to the area near the slope, or Mark the area adjacent to the slope in advance to obtain the location of the area adjacent to the slope.
- the installer can record the position details of all slopes on the travel route in the world coordinate system in advance on the intelligent device such as the robot when installing the intelligent device such as the robot ⁇ (x1, y1), (x2, y2), (x3, y3), .
- the intelligent device such as the robot
- the intelligent device such as the robot
- the intelligent device such as the robot
- the intelligent device such as the robot
- the intelligent device such as the robot
- the intelligent device such as the robot
- the intelligent device such as the robot
- the intelligent device such as the robot
- the intelligent device such as the robot ⁇ (x1, y1), (x2, y2), (x3, y3)
- the installer can record the position details of all slopes on the travel route in the world coordinate system in advance on the intelligent device such as the robot when installing the intelligent device such as the robot ⁇ (x1, y1), (x2, y2), (x3, y3), .
- take the recorded slope position coordinates as the center of the circle
- This method of marking the adjacent area of the slope is relatively simple and direct. Of course, other methods can also be used to mark the adjacent area of the slope. For example, when the coordinates of the marked slope position are about to be traveled, the coordinates face a similar semicircle area between the travel directions. It can be used as a slope adjacent area, because the robot only needs to consider this half of the situation when climbing or descending the slope.
- the robot will not only face the slope, but may also face obstacles such as people and objects. It needs to be distinguished to take corresponding strategies. For example, when it is confirmed that the front is a slope, it can go straight, and when it is confirmed that there is an obstacle ahead, it can Stop moving forward and handle the alarm, or execute the preset obstacle avoidance plan.
- the radar points scanned in the vicinity of the slope are clustered and screened to determine the slope.
- Clustering is a machine learning technique that involves grouping data points. For a given set of data points, a clustering algorithm can be used to divide each data point into a specific group. In theory, data points in the same group should have similar properties and/or characteristics, while data points in different groups should have highly different properties and/or characteristics.
- each point cloud cluster represents a collection of multiple radar points, while the point cloud cluster of the slope is usually linear, and the point cloud cluster of obstacles will show various irregular shapes, which can be The clustering effect shown in Figure 3 is obtained.
- the point in each circle in Figure 3 is represented as a point cloud cluster. Some point cloud clusters are linear, and some point cloud clusters are polylines or arcs. , when clustering, you can also mark different clusters with different colors for easy distinction.
- the specific clustering algorithm used is not limited.
- the region growing algorithm can be used.
- the idea of this algorithm is relatively simple, and it can usually segment connected regions with the same characteristics, and can provide good boundary information and segmentation results. The best performance can be achieved when no prior knowledge is available, and it can be used to segment more complex images, such as natural scenes, coins, medical images, etc.
- Density-Based clustering with noise can also be used.
- Spatial Clustering of Applications with Noise, DBSCAN DBSCAN
- this algorithm can cluster dense data sets of any shape, and find outliers while clustering, and the clustering results are rarely biased.
- the point cloud clusters obtained in the above step S101 include both slope point cloud clusters and obstacle point cloud clusters, and the scope needs to be further narrowed.
- the number of radar points in the slope point cloud cluster is often large.
- a number threshold can be preset, and when the number of radar points in the point cloud cluster is found to be greater than the preset number threshold, the point cloud cluster is confirmed as a suspected slope point cloud cluster.
- Step S103 confirming the point cloud cluster belonging to the linear type in the filtered suspected slope point cloud cluster as the slope point cloud cluster, and confirming the lidar scanning area corresponding to the slope point cloud cluster as the slope.
- the point cloud clusters screened according to the number of radar points in the above steps are only suspected slope point cloud clusters, and it is still impossible to accurately determine whether they are real slope point cloud clusters.
- only the linear point cloud clusters of the filtered suspected slope point cloud clusters belong to the slope point cloud clusters, and the other polyline or arc point cloud clusters are considered to be obstacle point clouds.
- the specific method for judging whether the point cloud clustering belongs to the linear type is not limited.
- the average distance of the straight line can be used to judge whether it is a straight line. It can also be judged according to how many of the distances from each radar point to the proposed straight line are greater than the preset distance value. For example, when the distance to the proposed straight line is greater than the preset distance value When the number of radar points in the distance is less than 5, it is determined that the suspected slope point cloud cluster is a slope point cloud cluster.
- the radar points in the confirmed slope point cloud cluster can also be filtered out from all the radar points scanned in the vicinity of the slope, so that all the remaining lidar point sets are obstacles, and the robot Corresponding strategies can be better taken accordingly, such as avoiding obstacles or stopping and calling the police.
- the slope sensing method provided by the first embodiment of the present application is based on the lidar technology and the method of pre-marking the area near the slope, and performs point cloud clustering on all the radar points scanned in the area near the slope, and finally filters out the suspected Slope point cloud clustering, further obtain the slope point cloud cluster from the suspected slope point cloud clustering, and confirm the lidar scanning area corresponding to the slope point cloud cluster as the slope, so that the slope on the robot distribution path can be compared with the slope point cloud.
- the obstacles are distinguished to ensure the normal distribution of the robot, and the perception is only performed when the robot is in the vicinity of the pre-marked slope. When the robot is not in the vicinity of the slope, it does not need to perceive, which greatly reduces the amount of calculation. False detection of non-slope areas is also avoided.
- FIG. 4 shows the flow of the slope sensing method provided by the second embodiment of the present application, wherein steps S102 and S103 are the same as the first embodiment and will not be repeated.
- Step S101 further includes:
- Step S1011 randomly select one radar point from all the radar points scanned in the area adjacent to the slope as a seed point for area growth.
- Step S1012 grouping the first-level non-seed radar points and the seed radar points searched within a preset distance range of the seed points into a point cloud cluster.
- Step S1013 starting with the first-level non-seed radar points, searching for radar points within the preset distance range step by step, until there are no radar points within the preset distance range, and then searching all levels of non-seed radar points. Radar points are also grouped into the same point cloud cluster.
- Step S1014 if there are unclassified radar points, randomly select one of the unclassified radar points as a new seed point, and then repeat the above search steps until all radar points complete the point cloud clustering.
- a region growing algorithm is used to cluster point clouds.
- the basic idea of the region growing algorithm is to combine pixels with similar properties together. For each area, a seed point must be designated as the starting point of growth, and then the pixels in the area around the seed point are compared with the seed point, and the points with similar properties are combined to continue to grow outward until no pixels that meet the conditions are selected. Until it is included, the growth of such an area is complete.
- Similarity property is defined as a radar point within a preset distance range from the seed point or the non-seed point at each level.
- first-level non-seed radar points within a certain distance from the seed point are grouped with the seed points, and then repeat the previous step for such first-level non-seed radar points to find other radar points, which are also classified into this category.
- search for the secondary non-seed radar points within a preset distance from the primary non-seed radar point and also classify them into this category, and then search for the third-level non-seed radar points within the preset distance range from the secondary non-seed radar point.
- Class non-seed radar points also fall into this category. This kind of search has been continued until such radar points cannot find any adjacent points, and this category is completed.
- FIG. 5 shows the flow of the slope sensing method provided by the third embodiment of the present application, wherein steps S101 and S102 are the same as those in the first and second embodiments and will not be repeated.
- Step S103 further includes:
- Step S1031 fitting each suspected slope point cloud cluster into a straight line.
- Step S1032 Calculate the average difference of the distances between the radar points inside each suspected slope point cloud cluster and the respective corresponding straight lines.
- step S1033 the suspected slope point cloud clusters whose distance average difference is lower than the preset average difference threshold are confirmed as slope point cloud clusters.
- the least squares method is used to fit a straight line, and then the average distance error between the radar points in the point cloud cluster and the straight line is calculated.
- the average distance error is lower than a certain threshold, the point cloud cluster is considered to be close to a straight line. , which is considered to be the slope scanned by the radar.
- the average distance error is obtained by calculating the average value of the distances from all radar points to the straight line, then calculating the error between the distances from all radar points to the straight line and the average value, and then averaging all the errors. This reflects whether the point cloud cluster is close enough to a straight line.
- FIG. 6 shows the structure of the slope sensing device provided by the fourth embodiment of the present application. For the convenience of description, only the parts related to the present application are shown.
- the device includes a clustering unit 61, a screening unit 62 and a slope confirmation unit 63.
- the clustering unit 61 is used to perform point cloud clustering on all radar points scanned in the vicinity of the slope to obtain several point cloud clusters
- the screening unit 62 is used to screen out the suspected slope point cloud clusters from the several point cloud clusters;
- the slope confirmation unit 63 is used to filter out the point clouds belonging to the linear type in the screened out described suspected slope point cloud clusters.
- the class is identified as a slope point cloud cluster, and the lidar scanning area corresponding to the slope point cloud cluster is identified as a slope.
- the apparatus further includes: a filtering unit, configured to filter out the radar points in the slope point cloud cluster from all the radar points scanned in the vicinity of the slope.
- a filtering unit configured to filter out the radar points in the slope point cloud cluster from all the radar points scanned in the vicinity of the slope.
- the device further includes: a slope adjacent area determination unit for recording the position coordinates of all slopes on the travel route, and then for each slope, taking the recorded position coordinates as the center of the circle and creating a circular area with a preset radius As slope adjacent areas, all slope adjacent areas are pre-marked in the created map. Determining the area adjacent to the slope in advance will help to greatly reduce the amount of calculation and avoid false perception due to misjudgment of the device in the non-slope adjacent area.
- the clustering unit 61 also includes a seed point confirmation module 611, a first-level non-seed radar point classification module 612, and a second-level non-seed radar point classification module 613, wherein the seed point confirmation module 611 is used to randomly select one radar point from all the radar points scanned in the vicinity of the slope and the unclassified radar points as a seed point for regional growth; the first-level non-seed radar point classification module 612 is used to classify the The first-level non-seed radar points and the seed radar points searched within the preset distance range of the seed points are classified into a point cloud cluster; the non-seed radar point classification module 613 below the second level is used to Start with non-seed radar points, search for radar points within the preset distance range step by step until there are no radar points within the preset distance range, and then group the searched non-seed radar points at all levels to the same point Cloud clustering.
- the seed point confirmation module 611 is used to randomly select one radar point from all the radar points scanned in the vicinity of
- the screening unit 62 is specifically configured to identify a point cloud cluster whose number of radar points is greater than a preset number threshold in the several point cloud clusters as a suspected slope point cloud cluster.
- the slope confirmation unit 63 includes a fitting module 631 , a mean difference calculation module 632 , and a slope confirmation module 633 , wherein the fitting module 631 is used to fit each suspected slope point cloud cluster into a Straight line, the mean difference calculation module 632 is used to calculate the distance mean difference between the radar points inside each suspected slope point cloud cluster and the corresponding straight line; the slope confirmation module 633 is used to make the distance mean difference lower than the preset mean difference threshold The suspected slope point cloud cluster is confirmed as the slope point cloud cluster.
- a fifth embodiment of the present application further provides a robot, including a memory 100 and a processor 200 , and the processor 200 may be the slope sensing device in the above embodiments.
- the memory 100 may be, for example, hard drive memory, non-volatile memory (such as flash memory or other electronically programmable limit erasure memory used to form solid state drives, etc.), volatile memory (such as static or dynamic random access memory, etc.), etc. , the embodiments of the present application are not limited.
- the memory 100 stores executable program codes; the processor 200 coupled with the memory 100 invokes the executable program codes stored in the memory to execute the slope sensing method as described above.
- the sixth embodiment of the present application also provides a computer-readable storage medium, and the computer-readable storage medium may be provided in the robot in each of the foregoing embodiments, and the computer-readable storage medium may be the aforementioned FIG. 9 Memory 100 in the illustrated embodiment.
- a computer program is stored on the computer-readable storage medium, and when the program is executed by the processor, implements the slope sensing method described in the embodiments shown in FIG. 2 to FIG. 5 .
- the computer-storable medium may also be a U disk, a removable hard disk, a read-only memory (ROM, Read-Only Memory), a RAM, a magnetic disk, or an optical disk, and other mediums that can store program codes.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Optical Radar Systems And Details Thereof (AREA)
Abstract
一种斜坡感知方法、装置、机器人和计算机可读存储介质。斜坡感知方法基于激光雷达技术实现，包括以下步骤：将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类（S101）；从若干点云聚类中筛选出疑似斜坡点云聚类（S102）；将筛选出的疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将斜坡点云聚类对应的激光雷达扫描区域确认为斜坡（S103）。
Description
本申请要求于2021年1月18日提交中国国家知识产权局专利局、申请号为202110063124.4、申请名称为“斜坡感知方法、装置、机器人和存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
本申请涉及机器人技术领域，尤其涉及一种斜坡感知方法、装置、机器人和计算机可读存储介质。
物流配送的成本压力使得配送机器人正逐渐应用于送货、送餐等各种短距离配送场景，这种无人化的配送方式不仅可以节约人力成本，还可提高用户的趣味性体验，并在一定程度上保障用户购物的私密性，所以配送机器人在短距离配送中的应用必定会越来越广泛。
机器人在配送途中可能会遇到障碍物，通常会通过激光雷达技术进行障碍物检测，出于安全考虑，一旦检测到障碍物就会停止前行。但是现实中配送路径上也可能会存在上下坡的情况，如图1所示，当坡度较大时，激光雷达的激光射线会打到斜坡上，使机器人误认为斜坡是障碍物，导致机器人不走或者执行预置的避障处理，这种情况下机器人执行避障又会导致无法走上斜坡而偏离事先确定的配送路线，影响正常配送。同理，当下坡时，由于激光雷达的激光射线会达到斜坡前方的平整路面上，也有可能认为是遇到障碍物而停止前行或进行避障。这种误检测严重影响了机器人的正常配送，亟待改进。
在此处键入技术问题描述段落。
根据本申请的各种实施例，提供一种斜坡感知方法、装置、机器人和计算机可读存储介质。
一种斜坡感知方法，所述斜坡感知方法基于激光雷达技术实现，包括下述步骤：将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类；从所述若干点云聚类中筛选出疑似斜坡点云聚类；将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。
一种斜坡感知装置，包括：聚类单元，用于将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类；筛选单元，用于从所述若干点云聚类中筛选出疑似斜坡点云聚类；斜坡确认单元，用于将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。
一种机器人，包括：存储器和处理器；所述存储器存储有可执行程序代码；与所述存储器耦合的所述处理器，调用所述存储器中存储的所述可执行程序代码，执行如上所述的斜坡感知方法。
一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时，实现如上所述的斜坡感知方法。
本申请的一个或多个实施例的细节在下面的附图和描述中提出。本申请的其它特征和优点将从说明书、附图以及权利要求书变得明显。
在此处键入有益效果描述段落。
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他实施例的附图。
图1为现有技术提供的机器人爬坡时的激光射线的光路图；
图2为本申请第一实施例提供的斜坡感知方法的流程示意图；
图3为本申请第一实施例提供的聚类效果图；
图4为本申请第二实施例提供的斜坡感知方法的流程示意图；
图5为本申请第三实施例提供的斜坡感知方法的流程示意图；
图6为本申请第四实施例提供的斜坡感知装置的结构图；
图7为本申请第四实施例提供的聚类单元的结构图；
图8为本申请第四实施例提供的斜坡确认单元的结构图；
图9为本申请第五实施例提供的机器人的结构示意图。
为了便于理解本申请，下面将参照相关附图对本申请进行更全面的描述。附图中给出了本申请的较佳实施例。但是，本申请可以以许多不同的形式来实现，并不限于本文所描述的实施例。相反地，提供这些实施例的目的是使对本申请的公开内容的理解更加透彻全面。
除非另有定义，本文所使用的所有的技术和科学术语与属于发明的技术领域的技术人员通常理解的含义相同。本文中在发明的说明书中所使用的术语只是为了描述具体的实施例的目的，不是旨在限制本申请。本文所使用的术语“和／或”包括一个或多个相关的所列项目的任意的和所有的组合。
本申请实施例基于激光雷达雷达技术实现，首先将在所述斜坡临近区域扫描到的所有雷达点进行点云聚类，然后最终从中筛选出疑似斜坡点云聚类，再进一步从疑似斜坡点云聚类得到斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。
图2为本申请第一实施例提供的斜坡感知方法的流程示意图。此斜坡感知方法基于激光雷达技术实现，主要适用于酒店、餐厅等场所内的配送机器人，还可适用于在物流园区等固定场所内运行的送货机器人甚至自动驾驶汽车。如图2所示，该方法具体包括：
S101、将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类。
斜坡临近区域是指在斜坡附近预置范围内且包含斜坡的区域，本实施例中，只在智能设备行进至斜坡临近区域时才进行斜坡感知，在斜坡临近区域之外行进时不需要感知斜坡而正常行进，有助于大大减少计算量，避免在非斜坡临近区域由于器件的误判而出现误感知。
斜坡临近区域可以通过多种方式确定，如在智能设备行进过程中采用RGBD相机识别是否行进至斜坡临近区域，又如采用三维激光雷达/多线激光雷达识别是否行进至斜坡临近区域，还可以是预先对斜坡临近区域进行标记，得到斜坡临近区域的位置。
以预先标记斜坡临近区域的方式为例，可以是安装人员在安装机器人等智能设备时，预先在机器人等智能设备上记录行进路线上所有斜坡在世界坐标系的位置明细{(x1,y1), (x2, y2),
(x3, y3), ..., (xn, yn)}，然后将距离各斜坡位置坐标一定范围内的区域视作斜坡临近区域，最终得到n个斜坡临近区域。具体地，对于每个斜坡，以记录的斜坡位置坐标为圆心、以预置的半径创建圆形区域作为斜坡临近区域，再将所有的斜坡临近区域预先标记在所建的地图中，其中，地图可用于导航、建模、探测、定位等方面。
这种斜坡邻近区域的标记方式较为简单直接，当然也可以采用其他的方式来标记斜坡临近区域，例如，当将要行进至标记的斜坡位置坐标时，该坐标面对行进方向之间的类似半圆区域可以作为斜坡临近区域，因为机器人爬坡或下坡时只需考虑这半边的情况。
在斜坡临近区域内，机器人不仅会面临斜坡，也可能会面临人、物等障碍，需要区分出来以采取对应的策略，例如当确认前方是斜坡时，可以直行，当确认前方存在障碍时，可以停止前行并做报警处理，或者执行预置的避障方案。
本实施例中，对在斜坡临近区域扫描到的雷达点采用聚类并筛选的方式来确定斜坡。所谓聚类，是一种涉及数据点分组的机器学习技术，对于给定的一组数据点，可以使用聚类算法将每个数据点划分为一个特定的组。理论上，同一组中的数据点应该具有相似的属性和/或特征，而不同组中的数据点应该具有高度不同的属性和/或特征。具体到本案，每个点云聚类表示多个雷达点的集合，而斜坡的点云聚类通常会呈直线型，障碍物的点云聚类则会表现出各种不规则的形状，可得到如图3所示的聚类效果，图3中每个圆圈中的点表示为一个点云聚类，有的点云聚类呈直线型，有的点云聚类呈折线或弧线型，聚类时，还可以将不同的聚类分别标识以不同的颜色方便区分。
所采用的具体聚类算法不限，例如，可以采用区域生长算法，这种算法的思想相对简单，通常能将具有相同特征的联通区域分割出来，并能提供很好的边界信息和分割结果。在没有先验知识可以利用时，可以取得最佳的性能，可以用来分割比较复杂的图像，如自然景物、硬币、医学图像等。也可以采用具有噪声的基于密度聚类（Density-Based
Spatial Clustering of Applications with Noise，DBSCAN）算法，这种算法能够对任意形状的稠密数据集进行聚类，并且在聚类的同时发现异常点，聚类结果鲜有发生偏倚等优点。
S102、从所述若干点云聚类中筛选出疑似斜坡点云聚类。
上述步骤S101得到的点云聚类既包含斜坡点云聚类，也包含障碍物的点云聚类，需要进一步缩小范围。
根据实验发现，斜坡点云聚类的雷达点数往往较多，当一个点云聚类中的雷达点数超过一定阈值时，才能认为有可能是斜坡，否则则认为是人腿等障碍物。因此可以预置一个数量阈值，当发现有点云聚类中雷达点数量大于预置的数量阈值时，将该点云聚类确认为疑似斜坡点云聚类。
步骤S103，将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。
上述步骤根据雷达点数筛选出的点云聚类只是疑似斜坡点云聚类，还无法准确判断出是否是真实的斜坡点云聚类。从形状上看，筛选出的疑似斜坡点云聚类只有直线型的点云聚类才属于斜坡点云聚类，其他折线或弧线型的点云聚类则认为是障碍物点云。
点云聚类是否属于直线型的具体判断方法不限，例如可以是基于最小二乘法将疑似斜坡点云聚类拟合出直线，再根据疑似斜坡点云聚类中的各雷达点与所拟出直线的平均距离来判断是否属于直线型，也可以根据各雷达点到所拟出直线的距离中有几个大于预设距离值来判断，例如，当到所拟出直线的距离大于预设距离的雷达点数低于5个时，则判断该疑似斜坡点云聚类为斜坡点云聚类。
进一步地，还可以将确认出的斜坡点云聚类中的雷达点从在所述斜坡临近区域扫描到的所有雷达点中过滤掉，这样剩下的激光雷达点集合中全部是障碍物，机器人可以更好地据此采取对应的策略，如避障或停止前行并报警。
本申请第一实施例所提供的斜坡感知方法基于激光雷达技术和预先标记斜坡临近区域的方式，将在所述斜坡临近区域扫描到的所有雷达点进行点云聚类，然后最终从中筛选出疑似斜坡点云聚类，再进一步从疑似斜坡点云聚类得到斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡，从而可以将机器人配送路径上的斜坡与障碍物区分开来，保证了机器人的正常配送，并且仅在行进至预先标记好的斜坡临近区域时才进行感知，而当机器人不处于斜坡临近区域时则不需要感知，大大减少了计算量，也避免了非斜坡区域的误检测。
在第一实施例的基础上，图4示出了本申请第二实施例提供的斜坡感知方法的流程，其中，步骤S102与S103与第一实施例相同不再赘述。步骤S101进一步包括：
步骤S1011，从在所述斜坡临近区域扫描到的所有雷达点中随机选择一个雷达点作为区域生长的种子点。
步骤S1012，将所述种子点预置距离范围内搜寻到的一级非种子雷达点与所述种子雷达点归至一个点云聚类。
步骤S1013，以所述一级非种子雷达点为开始，以搜寻预置距离范围内雷达点的方式进行逐级搜寻，直至在预置距离范围内无雷达点，然后将搜寻的各级非种子雷达点也归至同一个点云聚类。
步骤S1014，若存在未归类的雷达点，则从未归类的雷达点中随机选择一个作为新的种子点，然后重复上述搜寻的步骤，直至所有的雷达点全部完成点云聚类。
本实施例采用区域生长算法进行点云聚类，区域生长算法的基本思想是将有相似性质的像素点合并到一起。对每一个区域要先指定一个种子点作为生长的起点，然后将种子点周围领域的像素点和种子点进行对比，将具有相似性质的点合并起来继续向外生长，直到没有满足条件的像素被包括进来为止，这样一个区域的生长就算完成了。
具体到本实施例，需要随机确定一个雷达点为种子点，该种子点作为区域生长的起点，然后以此起点一级一级的寻找“具有相似性质”的雷达点，本实施例中的“相似性质”定义为距离种子点或者每一级的非种子点的距离在预置距离范围内的雷达点。首先种子点一定距离范围内的一级非种子雷达点与种子点聚为一类，然后将此类的一级非种子雷达点重复上一步，寻找其他的雷达点，同样归到这一类，例如，搜寻距离一级非种子雷达点在预置距离范围内的二级非种子雷达点，并同样归到这一类，然后再搜寻距离二级非种子雷达点在预置距离范围内的三级非种子雷达点，也同样归到这一类。一直这样一级一级的搜寻直到此类的雷达点都找不到相邻点为止，这样子这一类就算完成了。
如果还有其他没有归类的雷达点，则从中再随机选择一个作为新的种子点，重复上述过程。
在第一或第二实施例的基础上，图5示出了本申请第三实施例提供的斜坡感知方法的流程，其中，步骤S101与S102与第一、第二实施例相同不再赘述。步骤S103进一步包括：
步骤S1031，将每个疑似斜坡点云聚类拟合成一直线。
步骤S1032，计算每个疑似斜坡点云聚类内部的雷达点到各自对应的直线的距离均差。
步骤S1033，将距离均差低于预置的均差阈值的疑似斜坡点云聚类确认为斜坡点云聚类。
本实施例采用最小二乘法拟合直线，然后计算此点云聚类中的雷达点离这条直线的平均距离误差，当平均距离误差低于一定阈值时，才认为此点云聚类接近直线，才认为是雷达扫描到的斜坡。其中，平均距离误差通过如下方式得到：计算所有雷达点到该直线的距离的平均值，然后计算所有雷达点到该直线的距离与平均值之间的误差，再将所有误差的平均。这样可以反映出该点云聚类是否足够接近直线。
图6示出了本申请第四实施例提供的斜坡感知装置的结构，为了便于描述，仅示出了与本申请相关的部分。
参照图6，该装置包括聚类单元61、筛选单元62和斜坡确认单元63，聚类单元61用于将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类；筛选单元62用于从所述若干点云聚类中筛选出疑似斜坡点云聚类；斜坡确认单元63用于将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。
进一步地，所述装置还包括：过滤单元，用于将所述斜坡点云聚类中的雷达点从在所述斜坡临近区域扫描到的所有雷达点中过滤掉。这样剩下的激光雷达点集合中全部是障碍物，机器人可以更好地据此采取对应的策略，如避障或停止前行并报警。
进一步地，所述装置还包括：斜坡临近区域确定单元，用于记录行进路线上所有斜坡的位置坐标，然后对于每个斜坡，以记录的位置坐标为圆心、以预置的半径创建圆形区域作为斜坡临近区域，再将所有的斜坡临近区域预先标记在创建的地图中。预先确定出斜坡临近区域将有助于大大减少计算量，避免在非斜坡临近区域由于器件的误判而出现误感知。
进一步地，如图7所示，聚类单元61还包括种子点确认模块611、一级非种子雷达点归类模块612、二级以下非种子雷达点归类模块613，其中，种子点确认模块611用于从在所述斜坡临近区域扫描到的所有雷达点中以及未归类的雷达点中随机选择一个雷达点作为区域生长的种子点；一级非种子雷达点归类模块612用于将所述种子点预置距离范围内搜寻到的一级非种子雷达点与所述种子雷达点归至一个点云聚类；二级以下非种子雷达点归类模块613用于以所述一级非种子雷达点为开始，以搜寻预置距离范围内雷达点的方式进行逐级搜寻，直至在预置距离范围内无雷达点，然后将搜寻的各级非种子雷达点也归至同一个点云聚类。
进一步地，筛选单元62具体用于将所述若干点云聚类中雷达点数量大于预置的数量阈值的点云聚类确认为疑似斜坡点云聚类。
进一步地，如图8所示，斜坡确认单元63包括拟合模块631、均差计算模块632、斜坡确认模块633，其中，拟合模块631用于将每个疑似斜坡点云聚类拟合成一直线，均差计算模块632用于计算每个疑似斜坡点云聚类内部的雷达点到各自对应的直线的距离均差；斜坡确认模块633用于将距离均差低于预置的均差阈值的疑似斜坡点云聚类确认为斜坡点云聚类。
上述各模块的技术原理细节请参见上文中各实施例的描述，此处不再赘述。
如图9所示，本申请第五实施例还提供了一种机器人，包括存储器100和处理器200，处理器200可以是上述实施例中的斜坡感知装置。存储器100可以例如硬盘驱动存储器，非易失性存储器（例如闪存或用于形成固态驱动器的其它电子可编程限制删除的存储器等），易失性存储器（例如静态或动态随机存取存储器等）等，本申请实施例不作限制。
存储器100存储有可执行程序代码；与存储器100耦合的处理器200，调用所述存储器中存储的所述可执行程序代码，执行如上所述的斜坡感知方法。
进一步地，本申请第六实施例还提供了一种计算机可读存储介质，该计算机可读存储介质可以是设置于上述各实施例中的机器人中，该计算机可读存储介质可以是前述图9所示实施例中的存储器100。该计算机可读存储介质上存储有计算机程序，该程序被处理器执行时实现前述图2至图5所示实施例中描述的斜坡感知方法。进一步的，该计算机可存储介质还可以是U盘、移动硬盘、只读存储器（ROM，Read-Only Memory）、RAM、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对申请专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。
在此处键入本发明的实施方式描述段落。
在此处键入工业实用性描述段落。
在此处键入序列表自由内容描述段落。
Claims (1)
- 一种斜坡感知方法，所述斜坡感知方法基于激光雷达技术实现，包括下述步骤：将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类；从所述若干点云聚类中筛选出疑似斜坡点云聚类；将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。2 .根据权利要求1所述的方法，其特征在于，还包括：将所述斜坡点云聚类中的雷达点从在所述斜坡临近区域扫描到的所有雷达点中过滤掉。3 .根据权利要求1所述的方法，其特征在于，在所述将在所述斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类的步骤之前，所述斜坡感知方法还包括下述步骤：记录行进路线上所有斜坡的位置坐标；对于每个斜坡，以记录的位置坐标为圆心、以预置的半径创建圆形区域作为斜坡临近区域；将所有的斜坡临近区域预先标记在创建的地图中。4 .根据权利要求1所述的方法，其特征在于，所述将在所述斜坡临近区域扫描到的所有雷达点进行点云聚类，包括：从在所述斜坡临近区域扫描到的所有雷达点中随机选择一个雷达点作为区域生长的种子点；将所述种子点预置距离范围内搜寻到的一级非种子雷达点与所述种子雷达点归至一个点云聚类；以所述一级非种子雷达点为开始，以搜寻预置距离范围内雷达点的方式进行逐级搜寻，直至在预置距离范围内无雷达点，然后将搜寻的各级非种子雷达点也归至同一个点云聚类；若存在未归类的雷达点，则从未归类的雷达点中随机选择一个作为新的种子点，然后重复上述搜寻的步骤，直至所有的雷达点全部完成点云聚类。5 .根据权利要求1所述的方法，其特征在于，所述从所述若干点云聚类中筛选出疑似斜坡点云聚类，包括：将所述若干点云聚类中雷达点数量大于预置的数量阈值的点云聚类确认为疑似斜坡点云聚类。6 .根据权利要求1所述的方法，其特征在于，所述将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，包括：将每个疑似斜坡点云聚类拟合成一直线；计算每个疑似斜坡点云聚类内部的雷达点到各自对应的直线的距离均差；将距离均差低于预置的均差阈值的疑似斜坡点云聚类确认为斜坡点云聚类。7 .一种斜坡感知装置，包括：聚类单元，用于将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类；筛选单元，用于从所述若干点云聚类中筛选出疑似斜坡点云聚类；斜坡确认单元，用于将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。8 .根据权利要求7所述的装置，其特征在于，所述装置还包括：过滤单元，用于将所述斜坡点云聚类中的雷达点从在所述斜坡临近区域扫描到的所有雷达点中过滤掉。9 .根据权利要求7所述的装置，其特征在于，所述装置还包括：斜坡临近区域确定单元，用于记录行进路线上所有斜坡的位置坐标，然后对于每个斜坡，以记录的位置坐标为圆心、以预置的半径创建圆形区域作为斜坡临近区域，再将所有的斜坡临近区域预先标记在创建的地图中。10 .根据权利要求7所述的装置，其特征在于，所述聚类单元包括：种子点确认模块，用于从在所述斜坡临近区域扫描到的所有雷达点中以及未归类的雷达点中随机选择一个雷达点作为区域生长的种子点；一级非种子雷达点归类模块，用于将所述种子点预置距离范围内搜寻到的一级非种子雷达点与所述种子雷达点归至一个点云聚类；二级以下非种子雷达点归类模块，用于以所述一级非种子雷达点为开始，以搜寻预置距离范围内雷达点的方式进行逐级搜寻，直至在预置距离范围内无雷达点，然后将搜寻的各级非种子雷达点也归至同一个点云聚类。11 .根据权利要求7所述的装置，其特征在于，所述筛选单元用于将所述若干点云聚类中雷达点数量大于预置的数量阈值的点云聚类确认为疑似斜坡点云聚类。12 .根据权利要求7所述的装置，其特征在于，所述斜坡确认单元包括：拟合模块，用于将每个疑似斜坡点云聚类拟合成一直线；均差计算模块，用于计算每个疑似斜坡点云聚类内部的雷达点到各自对应的直线的距离均差；斜坡确认模块，用于将距离均差低于预置的均差阈值的疑似斜坡点云聚类确认为斜坡点云聚类。13 .一种机器人，包括：存储器和处理器；所述存储器存储有可执行程序代码；与所述存储器耦合的所述处理器，调用所述存储器中存储的所述可执行程序代码，执行如下步骤：将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类；从所述若干点云聚类中筛选出疑似斜坡点云聚类；将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。14 .根据权利要求13所述的机器人，其特征在于，所述与所述存储器耦合的所述处理器，调用所述存储器中存储的所述可执行程序代码还执行如下步骤：将所述斜坡点云聚类中的雷达点从在所述斜坡临近区域扫描到的所有雷达点中过滤掉。15 .根据权利要求13所述的机器人，其特征在于，在所述将在所述斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类的步骤之前，还包括：记录行进路线上所有斜坡的位置坐标；对于每个斜坡，以记录的位置坐标为圆心、以预置的半径创建圆形区域作为斜坡临近区域；将所有的斜坡临近区域预先标记在创建的地图中。16 .根据权利要求13所述的机器人，其特征在于，所述将在所述斜坡临近区域扫描到的所有雷达点进行点云聚类，包括：从在所述斜坡临近区域扫描到的所有雷达点中随机选择一个雷达点作为区域生长的种子点；将所述种子点预置距离范围内搜寻到的一级非种子雷达点与所述种子雷达点归至一个点云聚类；以所述一级非种子雷达点为开始，以搜寻预置距离范围内雷达点的方式进行逐级搜寻，直至在预置距离范围内无雷达点，然后将搜寻的各级非种子雷达点也归至同一个点云聚类；若存在未归类的雷达点，则从未归类的雷达点中随机选择一个作为新的种子点，然后重复上述搜寻的步骤，直至所有的雷达点全部完成点云聚类。17 .根据权利要求13所述的机器人，其特征在于，所述从所述若干点云聚类中筛选出疑似斜坡点云聚类，包括：将所述若干点云聚类中雷达点数量大于预置的数量阈值的点云聚类确认为疑似斜坡点云聚类。18 .根据权利要求13所述的机器人，其特征在于，所述将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，包括：将每个疑似斜坡点云聚类拟合成一直线；计算每个疑似斜坡点云聚类内部的雷达点到各自对应的直线的距离均差；将距离均差低于预置的均差阈值的疑似斜坡点云聚类确认为斜坡点云聚类。19 .一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时，实现如下步骤：将在斜坡临近区域扫描到的所有雷达点进行点云聚类，得到若干点云聚类；从所述若干点云聚类中筛选出疑似斜坡点云聚类；将筛选出的所述疑似斜坡点云聚类中属于直线型的点云聚类确认为斜坡点云聚类，将所述斜坡点云聚类对应的激光雷达扫描区域确认为斜坡。20 .根据权利要求19所述的计算机可读存储介质，其特征在于，所述将在所述斜坡临近区域扫描到的所有雷达点进行点云聚类，包括：从在所述斜坡临近区域扫描到的所有雷达点中随机选择一个雷达点作为区域生长的种子点；将所述种子点预置距离范围内搜寻到的一级非种子雷达点与所述种子雷达点归至一个点云聚类；以所述一级非种子雷达点为开始，以搜寻预置距离范围内雷达点的方式进行逐级搜寻，直至在预置距离范围内无雷达点，然后将搜寻的各级非种子雷达点也归至同一个点云聚类；若存在未归类的雷达点，则从未归类的雷达点中随机选择一个作为新的种子点，然后重复上述搜寻的步骤，直至所有的雷达点全部完成点云聚类。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110063124.4A CN112904306A (zh) | 2021-01-18 | 2021-01-18 | 斜坡感知方法、装置、机器人和存储介质 |
| CN202110063124.4 | 2021-01-18 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2022152263A1 true WO2022152263A1 (zh) | 2022-07-21 |
Family
ID=76115200
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2022/072121 WO2022152263A1 (zh) | 2021-01-18 | 2022-01-14 | 斜坡感知方法、装置、机器人和存储介质 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112904306A (zh) |
| WO (1) | WO2022152263A1 (zh) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115423835A (zh) * | 2022-11-02 | 2022-12-02 | 中汽创智科技有限公司 | 杆状物点云数据处理方法、装置、电子设备及存储介质 |
| CN116304764A (zh) * | 2023-05-22 | 2023-06-23 | 湖南联智科技股份有限公司 | 一种施工隧道点云中轴线自动拟合方法 |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112904306A (zh) * | 2021-01-18 | 2021-06-04 | 深圳市普渡科技有限公司 | 斜坡感知方法、装置、机器人和存储介质 |
| CN113536958B (zh) * | 2021-06-23 | 2023-08-25 | 华南农业大学 | 导航路径提取方法、装置、农业机器人和存储介质 |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109188459A (zh) * | 2018-08-29 | 2019-01-11 | 东南大学 | 一种基于多线激光雷达的坡道小障碍物识别方法 |
| US20190178989A1 (en) * | 2017-12-11 | 2019-06-13 | Automotive Research & Testing Center | Dynamic road surface detecting method based on three-dimensional sensor |
| CN110441791A (zh) * | 2019-08-14 | 2019-11-12 | 深圳无境智能机器人有限公司 | 一种基于前倾2d激光雷达的地面障碍物检测方法 |
| CN110471086A (zh) * | 2019-09-06 | 2019-11-19 | 北京云迹科技有限公司 | 一种雷达测障系统及方法 |
| CN111077541A (zh) * | 2019-12-30 | 2020-04-28 | 深圳优地科技有限公司 | 一种障碍物识别方法、障碍物识别装置及移动机器人 |
| CN111419116A (zh) * | 2018-12-24 | 2020-07-17 | 珠海市一微半导体有限公司 | 扫地机器人爬坡控制方法和装置、存储介质及扫地机器人 |
| CN112904306A (zh) * | 2021-01-18 | 2021-06-04 | 深圳市普渡科技有限公司 | 斜坡感知方法、装置、机器人和存储介质 |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104950313B (zh) * | 2015-06-11 | 2017-11-07 | 同济大学 | 一种路面提取及道路坡度识别方法 |
| CN106204705B (zh) * | 2016-07-05 | 2018-12-07 | 长安大学 | 一种基于多线激光雷达的3d点云分割方法 |
| CN111352106B (zh) * | 2018-12-24 | 2022-06-14 | 珠海一微半导体股份有限公司 | 扫地机器人坡度识别方法、装置、芯片及扫地机器人 |
| CN111368607B (zh) * | 2018-12-26 | 2023-06-02 | 北京欣奕华科技有限公司 | 一种机器人、障碍物的检测方法及检测装置 |
| CN109959911A (zh) * | 2019-03-25 | 2019-07-02 | 清华大学 | 基于激光雷达的多目标自主定位方法及装置 |
| CN112099494B (zh) * | 2020-09-02 | 2021-09-07 | 西安交通大学 | 一种面向全地形车及其自动检测楼梯与攀爬方法 |
-
2021
- 2021-01-18 CN CN202110063124.4A patent/CN112904306A/zh active Pending
-
2022
- 2022-01-14 WO PCT/CN2022/072121 patent/WO2022152263A1/zh active Application Filing
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190178989A1 (en) * | 2017-12-11 | 2019-06-13 | Automotive Research & Testing Center | Dynamic road surface detecting method based on three-dimensional sensor |
| CN109188459A (zh) * | 2018-08-29 | 2019-01-11 | 东南大学 | 一种基于多线激光雷达的坡道小障碍物识别方法 |
| CN111419116A (zh) * | 2018-12-24 | 2020-07-17 | 珠海市一微半导体有限公司 | 扫地机器人爬坡控制方法和装置、存储介质及扫地机器人 |
| CN110441791A (zh) * | 2019-08-14 | 2019-11-12 | 深圳无境智能机器人有限公司 | 一种基于前倾2d激光雷达的地面障碍物检测方法 |
| CN110471086A (zh) * | 2019-09-06 | 2019-11-19 | 北京云迹科技有限公司 | 一种雷达测障系统及方法 |
| CN111077541A (zh) * | 2019-12-30 | 2020-04-28 | 深圳优地科技有限公司 | 一种障碍物识别方法、障碍物识别装置及移动机器人 |
| CN112904306A (zh) * | 2021-01-18 | 2021-06-04 | 深圳市普渡科技有限公司 | 斜坡感知方法、装置、机器人和存储介质 |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115423835A (zh) * | 2022-11-02 | 2022-12-02 | 中汽创智科技有限公司 | 杆状物点云数据处理方法、装置、电子设备及存储介质 |
| CN116304764A (zh) * | 2023-05-22 | 2023-06-23 | 湖南联智科技股份有限公司 | 一种施工隧道点云中轴线自动拟合方法 |
| CN116304764B (zh) * | 2023-05-22 | 2023-09-19 | 湖南联智科技股份有限公司 | 一种施工隧道点云中轴线自动拟合方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112904306A (zh) | 2021-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022152263A1 (zh) | 斜坡感知方法、装置、机器人和存储介质 | |
| US11157527B2 (en) | Creating clean maps including semantic information | |
| CN107677285B (zh) | 机器人的路径规划系统及方法 | |
| US10984659B2 (en) | Vehicle parking availability map systems and methods | |
| US11670087B2 (en) | Training data generating method for image processing, image processing method, and devices thereof | |
| Kümmerle et al. | A navigation system for robots operating in crowded urban environments | |
| CN104536445B (zh) | 移动导航方法和系统 | |
| WO2017020466A1 (zh) | 基于激光点云的城市道路识别方法、装置、存储介质及设备 | |
| Shang et al. | Lidar based negative obstacle detection for field autonomous land vehicles | |
| CN111542860A (zh) | 用于自主车辆的高清地图的标志和车道创建 | |
| CN114842438A (zh) | 用于自动驾驶汽车的地形检测方法、系统及可读存储介质 | |
| CN108845569A (zh) | 生成三维高清道路图水平弯道行车线的半自动点云方法 | |
| CN108984741B (zh) | 一种地图生成方法及装置、机器人和计算机可读存储介质 | |
| EP3904831A1 (en) | Visual localization using a three-dimensional model and image segmentation | |
| CN109645892A (zh) | 一种障碍物的识别方法及清洁机器人 | |
| CN111308500B (zh) | 基于单线激光雷达的障碍物感知方法、装置和计算机终端 | |
| CN110986945B (zh) | 基于语义高度地图的局部导航方法和系统 | |
| CN114488073A (zh) | 激光雷达采集到的点云数据的处理方法 | |
| CN110544268B (zh) | 一种基于结构光及SiamMask网络的多目标跟踪方法 | |
| CN111295666A (zh) | 一种车道线检测方法、装置、控制设备及存储介质 | |
| CN112393737A (zh) | 障碍地图的创建方法、系统，机器人及可读存储介质 | |
| CN114926809A (zh) | 可通行区域检测方法及装置、移动工具、存储介质 | |
| CN113442908B (zh) | 自动泊车路径规划方法及系统、泊车控制设备 | |
| Yan et al. | RH-Map: Online Map Construction Framework of Dynamic Object Removal Based on 3D Region-wise Hash Map Structure | |
| CN113008249B (zh) | 移动机器人的避让点检测方法、避让方法以及移动机器人 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22739138 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 16/11/2023) |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 22739138 Country of ref document: EP Kind code of ref document: A1 |