CN109000649B - Omni-directional mobile robot pose calibration method based on right-angle bend characteristics - Google Patents

Omni-directional mobile robot pose calibration method based on right-angle bend characteristics Download PDF

Info

Publication number
CN109000649B
CN109000649B CN201810529469.2A CN201810529469A CN109000649B CN 109000649 B CN109000649 B CN 109000649B CN 201810529469 A CN201810529469 A CN 201810529469A CN 109000649 B CN109000649 B CN 109000649B
Authority
CN
China
Prior art keywords
line segment
fitted
abstractly
coordinate system
bend
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.)
Active
Application number
CN201810529469.2A
Other languages
Chinese (zh)
Other versions
CN109000649A (en
Inventor
孙棣华
赵敏
廖孝勇
王俊祥
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University
Original Assignee
Chongqing University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chongqing University filed Critical Chongqing University
Priority to CN201810529469.2A priority Critical patent/CN109000649B/en
Publication of CN109000649A publication Critical patent/CN109000649A/en
Application granted granted Critical
Publication of CN109000649B publication Critical patent/CN109000649B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/20—Instruments for performing navigational calculations
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C25/00—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass
    • 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
    • G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
    • G01S17/88—Lidar systems specially adapted for specific applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Electromagnetism (AREA)
  • Manufacturing & Machinery (AREA)
  • Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)

Abstract

The invention discloses an omnibearing mobile robot pose calibration method based on right-angle bend characteristics, which comprises the following steps: s1, establishing a right-angle bend model; s2, standardizing the right-angle bend model, and then establishing a global reference point set of the right-angle bend; s3, establishing a trolley initial state coordinate system, and calculating initial pose estimation of the AGV in a global coordinate system; s4, identifying characteristics of the straight corner bend; and S5, carrying out global positioning and pose updating on the AGV based on the characteristics of the right-angled bend. Under the condition of fully analyzing the typical warehouse channel environment, the invention describes the environmental characteristics of the right-angle curve by using the laser radar data, and carries out global positioning and pose updating on the omnibearing trolley under the channel environment by using the characteristics, thereby eliminating the accumulated error in relative positioning and improving the overall positioning precision of the robot.

Description

Omni-directional mobile robot pose calibration method based on right-angle bend characteristics
Technical Field
The invention belongs to the technical field of mobile robot navigation and positioning, and particularly relates to an all-directional mobile robot pose calibration method based on right-angle bend characteristics.
Background
The omni-directional mobile robot, i.e. the AGV, is rapidly and widely applied in the logistics industry and other industries due to its advantages of high automation degree, good flexibility, high space utilization rate, etc., and becomes an important logistics conveying device of the manufacturing system and the automatic storage system.
Pose estimation of the intelligent vehicle, positioning for short, is an indispensable important link in automatic control and navigation of the ACV. In a general environment, the pose of a mobile robot is typically represented using three-dimensional vectors, i.e., lateral and longitudinal translational components relative to a global coordinate position and a rotational angle component representing its orientation. Accurate pose estimation has important significance for automatic map generation, path planning and control, target detection and tracking and the like of the mobile robot.
Common positioning technologies mainly include relative positioning based on a ranging method and inertial navigation, and absolute positioning technologies such as GPS positioning, landmark positioning, map matching, and the like. The existing positioning method mostly has the contradiction between the technical cost and the precision, and the relative positioning technical method mainly based on dead reckoning is simple and has low technical cost, but has the problem of accumulated error; the absolute positioning technology has no accumulated error, but the technology has the problems of complex technology, poor method portability, high technical cost and the like.
Due to the fact that the independent relative positioning technology and the independent absolute positioning technology have certain defects, the positioning result is not ideal in some complex environments with high positioning accuracy requirements. In practical application, two methods are combined, and the absolute pose information is used for calibrating and updating the relative pose information, so that the positioning accumulated error is eliminated, and the accuracy of the estimation of the overall pose is improved.
In the existing pose calibration method, the technology is mature and mainly based on GPS data fusion, navigation beacons, labels, Landmarks and the like, but the GPS data is mostly used in outdoor environment, indoor signals are easy to be shielded and interfered, the positioning accuracy is more in the order of meters, and the indoor accurate positioning accuracy is insufficient; the positioning method based on the beacon and the label has high positioning precision and good real-time performance, but the positioning method needs to place a large number of artificial road signs or labels with obvious characteristics in the detection environment range of the sensor, has large workload, is easy to damage the road signs in scenes with complex environments such as a warehouse and the like, and has high equipment maintenance cost; secondly, the map matching-based pose calibration method is more researched in an indoor environment, typically including laser radar SLAM, visual SLAM technology and the like, but the algorithm complexity of the method is higher, the calculated amount is large, and in an environment with higher similarity, such as a warehouse channel, the map identification and matching process is easy to generate errors; in addition, in some environment feature-based pose calibration methods, technologies based on columnar features and indoor right-angle features are typical, the method is high in flexibility, but the former method has a good effect in an open indoor environment and has a poor effect in some complex environments; in the method based on the indoor right-angle feature, although the method has good applicability in an indoor environment with obvious structuralization, the method is easily influenced by sensor noise or environmental interference factors in the process of extracting corner point information, and the precision is easily influenced.
Therefore, a pose calibration method which is suitable for the omnidirectional mobile robot, good in applicability and feasible in an automatic storage channel environment is urgently needed.
Disclosure of Invention
In view of the above, in order to solve the above problems, the present invention provides a method for calibrating a pose of an omnidirectional mobile robot based on a feature of a right-angle curve, which uses laser radar data to describe an environmental feature of the right-angle curve under the condition of sufficiently analyzing a typical warehouse channel environment, and performs global positioning and pose updating on an omnidirectional trolley under the channel environment according to the environmental feature, thereby eliminating an accumulated error in relative positioning and improving the overall positioning accuracy of the robot.
The purpose of the invention is realized by the following technical scheme: the invention provides an omnibearing mobile robot pose calibration method based on right-angle bend characteristics, which comprises the following steps:
s1, establishing a right-angle bend model, which comprises an L-shaped right-angle bend model, an I-shaped right-angle bend model and a T-shaped right-angle bend model;
the L-shaped right-angle bending model comprises two line segments which are abstractly fitted by two wall surfaces and are in a vertical intersection state and a first line segment which is abstractly fitted by the goods package stack, wherein one line segment of the two line segments which are abstractly fitted by the two wall surfaces and are in the vertical intersection state is parallel to the first line segment which is abstractly fitted by the goods package stack;
the T-shaped right-angled bend model comprises a second line segment abstractly fitted by one wall surface and two line segments in a parallel state abstractly fitted by two goods pack stacks, and the two line segments in the parallel state abstractly fitted by the two goods pack stacks are perpendicular to but not intersected with the second line segment abstractly fitted by one wall surface;
the I-shaped right-angle bending model comprises a third line segment which is abstracted and fitted by one wall surface and two line segments which are abstractly fitted by the two goods package stacks and are in a parallel state, and the third line segment abstracted and fitted by one wall surface is perpendicular to and intersected with the two line segments which are abstractly fitted by the two goods package stacks and are in the parallel state; or the I-shaped right-angle bending model comprises a fourth line segment which is abstractly fitted by a goods pack stack and two line segments which are abstractly fitted by two wall surfaces and are in a parallel state, and the fourth line segment which is abstractly fitted by a goods pack stack is perpendicular to and intersected with the two line segments which are abstractly fitted by the two wall surfaces and are in the parallel state;
s2, standardizing the right-angle bend model, and then establishing a global reference point set of the right-angle bend;
s3, establishing a trolley initial state coordinate system, and calculating initial pose estimation of the AGV in a global coordinate system;
s4, identifying characteristics of the straight corner bend;
and S5, carrying out global positioning and pose updating on the AGV based on the characteristics of the right-angled bend.
Preferably, the L-shaped quarter bend model is:
Figure BDA0001676860370000031
the T-shaped quarter bend model is as follows:
Figure BDA0001676860370000032
the I-shaped quarter bend model is as follows:
Figure BDA0001676860370000033
line segment L under two-dimensional rectangular coordinate systemi＝(pi,θiStart, end, len), where start _ q and end _ q represent the coordinates of the starting point and the starting point of the fitting line segment of the wall in front of the trolley, start (q +1) represents the coordinates of the starting point of the fitting line segment of the inner wall of the left side channel of the trolley, end (q-1) represents the coordinates of the starting point of the fitting line segment of the inner wall of the right side channel of the trolley, piRepresenting a straight line L from the originiA distance of (a), thetaiRepresents the origin and LiThe perpendicular line of (A) and the positive direction of the x-axis, start and end respectively represent a line segment LiLen represents the line segment LiLength of (d), deltadminAnd deltadmaxMinimum and maximum thresholds, delta, representing the Euclidean distance between two pointsdmaxUsed for judging whether the head and the tail of two adjacent line segments are intersected or not, deltadminAnd the method is used for judging whether the distance between the head positions of two adjacent line segments is larger than a certain threshold value.
Preferably, in the step S2, the method for normalizing the L-shaped quarter-turn model features is as follows:
two line segments which are in a vertical intersection state and are abstractly fitted by two wall surfaces and one line segment which is in a vertical intersection state and is abstractly fitted by two wall surfaces are taken as fixed characteristics;
the method for standardizing the characteristics of the T-shaped quarter bend model comprises the following steps:
firstly, respectively calculating the coordinates of the intersection points of two parallel section extension lines and a second section, which are abstractly fitted by two bale stacks;
then, the arithmetic mean value of the coordinates of the two intersection points is obtained, and the arithmetic mean value is used as a Key point Key _ P;
and the Key point Key _ P and the second line segment are used as Key characteristics of positioning;
the method for standardizing the characteristics of the I-shaped quarter bend model comprises the following steps:
firstly, respectively solving the intersection point coordinates of two parallel line segment extensions and a third line segment abstractly fitted by stacking two goods packages;
then, the arithmetic mean value of the coordinates of the two intersection points is obtained, and the arithmetic mean value is used as a Key point Key _ P';
and the Key point Key _ P' and the third line segment are used as Key characteristics of positioning;
or the method for standardizing the characteristics of the I-shaped right-angle bending model comprises the following steps:
firstly, respectively solving the intersection point coordinates of two parallel line segment extensions and a fourth line segment abstractly fitted by two wall surfaces;
then, the arithmetic mean value of the coordinates of the two intersection points is calculated, and the arithmetic mean value is used as a Key point Key _ P ";
and the Key point Key _ P' and the fourth line segment are used as Key features of positioning.
Preferably, in the step S2, when the global coordinate system is established; taking the abstractly fitted point of the warehouse entrance as an origin, taking the direction perpendicular to the abstractly fitted line segment of the bale stack as a y axis, and taking the abstract line segment parallel to the bale stackEstablishing a global coordinate system xoy by taking the direction of the fitted line segment as an x axis; and establishing a global reference point set P ═ P { P } of the right-angle bend positioning key points under a global coordinate system1,P2,...,PnIn which P isi＝(xi,yi,θri)，i∈[1,n]，(xi,yi) Represents the coordinates of the intersection point of the two intersecting straight lines of the normalized quarter bend in the global coordinate system xoy, thetariRepresenting the angle between a straight line fitted abstractly to the wall and the x-axis of the global coordinate system xoy.
Preferably, the step S3 specifically includes the following sub-steps:
s31, establishing a body coordinate system x ' o ' y ' of the initial state of the AGV by taking the geometric center of the AGV as an original point, and establishing an omnidirectional mobile trolley kinematics model:
Figure BDA0001676860370000041
wherein (w)1,w2,w3,w4) Representing the rotating speed of four wheels of the AGV trolley, r is the radius of the wheels, m and n respectively represent half of the width and the length of the AGV body, (v)x,vyW) represents the moving speed of the mass center along the x axis, the y axis and the vehicle head direction respectively under the vehicle body initial coordinate system;
s32, calculating the barycenter coordinate (x ') under the body coordinate system x ' o ' y ' at the time t-kT according to the dead reckoning principle 'k,y′k,θ′k)，θkThe' represents the included angle between the direction of the head of the trolley and the positive y direction of the coordinate system of the mass center of the trolley, T represents the dead reckoning sampling period, and k represents a positive integer;
s33, calculating the barycenter coordinate (x) of the trolley under the global coordinate system xoy according to the conversion relation between the global coordinate system xoy and the body coordinate system x ' o ' y ' of the trolleyk,yk,θk)，θkAnd the included angle between the direction of the head of the trolley and the positive y direction of the global coordinate system is shown.
Preferably, the step S4 specifically includes the following sub-steps:
s41, firstly, manually remotely controlling the AGV to learn a primary path in a channel environment, and generating a global path key point map preliminarily identified by a right-angled bend; in the process of carrying out subsequent autonomous navigation on the AGV, judging whether the AGV enters the neighborhood range of the global path key point map preliminarily identified by the right angle bend or not in real time by adopting a pose tracking method according to the prior map, wherein the neighborhood range is defined by taking the key point as the center and taking the radius as r1The circular region of (a);
s42, extracting line segment characteristics based on the laser radar, and extracting the line segment characteristics of laser radar scanning data by using a split _ merge algorithm to obtain a line segment set L'; removing invalid data of the line segments extracted from the environment and combining the line segments again to obtain a new line segment set L';
s43, performing model discrimination on a line segment set L 'in a frame of laser radar near the right-angle bend by using a right-angle bend model and classifying the line segment set L'.
Preferably, the step S5 specifically includes the following sub-steps:
s51, associating reference coordinates; after the AGV finishes the identification of the right-angled bend characteristics, the right-angled bend characteristics at the position of the AGV need to be associated and matched with a preset global reference coordinate point to obtain a correct associated reference point;
s52, resolving the position of the AGV based on the line segment characteristics, and resolving the position of the laser radar to obtain the position o' (x) of the laser radar under the global coordinate system xoy by utilizing the standardized quarter bend model line segment and the key characteristics extracted from the scanning data of the laser radaro,yo,θo)；
And S53, updating the position and attitude angles (x) of the AGV under the global coordinate system xoy through the conversion of the laser radar and the AGV coordinate systemr,yr,θr) And updating the relative positioning result of the AGV by using the global positioning result of the right-angled bend.
Due to the adoption of the technical scheme, the invention has the following advantages:
under the condition of fully analyzing the typical warehouse channel environment, the invention describes the environmental characteristics of the right-angle curve by using the laser radar data, and carries out global positioning and pose updating on the omnibearing trolley under the channel environment by using the characteristics, thereby eliminating the accumulated error in relative positioning and improving the overall positioning precision of the robot.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention may be realized and attained by the means of the instrumentalities and combinations particularly pointed out hereinafter.
Drawings
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail with reference to the accompanying drawings:
FIG. 1 is a typical automated warehouse application environment;
FIG. 2 is a schematic view of an L-bend;
FIG. 3 is a schematic view of a T-bend;
FIG. 4 is a schematic view of a type I bend;
FIG. 5 is an L-bend normalization;
FIG. 6 illustrates T-bend normalization;
FIG. 7 illustrates type I bend normalization;
FIG. 8 is a schematic view of a set of reference coordinate points;
FIG. 9 is a coordinate system of the omni-directional mobile cart body;
FIG. 10 is a schematic diagram of a quarter turn preliminary location based on keypoint discrimination;
FIG. 11 illustrates the partitioning principle of the IEPF algorithm;
FIG. 12 is a schematic diagram of channel segments extracted by the split _ merge algorithm;
FIG. 13 is a schematic diagram of AGV positioning based on line segment characteristics;
FIG. 14 is a flow chart of the method of the present invention.
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention.
Please refer to fig. 1 to 14. It should be noted that the drawings provided in the present embodiment are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number, shape and size of the components in actual implementation, and the type, quantity and proportion of the components in actual implementation may be changed freely, and the layout of the components may be more complicated.
The method for calibrating the pose of the omnibearing mobile robot based on the characteristics of the right-angle bend comprises the following steps:
step 1: and (5) establishing a right-angle bend model.
The warehouse passage environment is mostly composed of three-dimensional shelves and wall surfaces, a typical structural geometric characteristic is formed inside the passage due to the general and regular arrangement mode of the goods packages, and a relatively obvious right-angle bending characteristic composed of line segments is formed at the passage turning position, as shown in fig. 1.
Through environmental analysis, the typical right angle bends existing in the warehouse access environment are mainly classified into the following types:
step 11: establishing an L-shaped quarter bend model
FIG. 2 is a schematic view showing an L-shaped quarter turn, in which the path consists of two wall surfaces and a stack of parcels, and the typical L-shaped quarter turn consists essentially of three segments (L)1,L2,L3) The L-shaped right-angle bending model comprises two line segments which are abstractly fitted by two wall surfaces and are in a vertical intersection state and a first line segment which is abstractly fitted by the goods package stack, wherein one line segment of the two line segments which are abstractly fitted by the two wall surfaces and are in the vertical intersection state is parallel to the first line segment which is abstractly fitted by the goods package stack. Line segment L under two-dimensional rectangular coordinate systemi＝(pi,θiStart, end, len), where piRepresenting from the originTo a straight line LiA distance of (a), thetaiRepresents the origin and LiThe perpendicular line of (A) and the positive direction of the x-axis, start and end respectively represent a line segment LiLen represents the line segment LiBased on the above line segment parameters and the channel combination characteristics in the actual environment, the model of the L-shaped quarter bend in this scenario is described as:
Figure BDA0001676860370000071
Figure BDA0001676860370000072
in the above formula, formula (1) is a rough description of the profile of the L-shaped quarter bend model, and formula (2) is a further precise description of the model satisfying the rough profile, and is mainly used for judging a right-angle feature formed by two segments in a vertical intersection state abstractly fitted by two wall surfaces in the L-shaped quarter bend model; num _ L represents the total number of line segments extracted from one frame of scanning data of the laser radar, len _ min represents the shortest length of L-shaped bent line segments, epsilon represents the minimum neighborhood for judging the included angle between two line segments, and deltadminAnd deltadmaxMinimum and maximum threshold values, delta, representing the Euclidean distance between two pointsdmaxUsed for judging whether the head and the tail of two adjacent line segments are intersected or not, deltadminThe method is used for judging whether the distance between the head positions of two adjacent line segments is larger than a certain threshold value, and in the formula (2), "|" represents a mathematical operator "or".
Step 12: building T-shaped quarter bend model
Fig. 3 is a schematic view of a T-shaped right-angle bending model, and a channel in the scene is composed of a wall surface and two stacks of bags. The curve abstract model under the scene also comprises three line segments, the T-shaped right-angle curve model comprises a second line segment abstractly fitted by one wall surface and two line segments abstractly fitted by two goods pack stacks and in a parallel state, and the two line segments abstractly fitted by the two goods pack stacks and in the parallel state are perpendicular to but not intersected with the second line segment abstractly fitted by one wall surface. By combining the line segment parameters and the channel combination characteristics in the actual environment, the model description of the T-shaped right-angle bend in the scene is as follows:
Figure BDA0001676860370000073
Figure BDA0001676860370000074
the formula 4 is mainly used for identifying the line segment fitted on the wall surface, and the parameter meaning is consistent with the parameter description in the T-shaped right-angled bend.
Step 13: i-shaped right-angle bending model
Fig. 4 shows that the type I channel ends belong to a special type of quarter bend, which the present example attributes to a typical quarter bend for global positioning due to its segment combination properties similar to those of a typical quarter bend. According to different surrounding environments forming the right-angle bend, the right-angle bend is mainly divided into two types, the I-type right-angle bend model comprises a third line segment which is abstracted and fitted by one wall surface and two line segments which are abstractly fitted by two goods pack stacks and are in a parallel state, and the third line segment abstracted and fitted by one wall surface is vertical to and intersected with the two line segments which are abstractly fitted by the two goods pack stacks and are in the parallel state; or the I-shaped right-angle bending model comprises a fourth line segment which is abstracted and fitted by a goods pack stack and two line segments which are abstractly fitted by two wall surfaces and are in a parallel state, and the fourth line segment which is abstracted and fitted by a goods pack stack is perpendicular to and intersected with the two line segments which are abstractly fitted by the two wall surfaces and are in the parallel state. This type of quarter bend model is described as:
Figure BDA0001676860370000081
Figure BDA0001676860370000082
step 2: right angle bend characteristic standardization and global reference point set establishment
In order to enable the positioning method based on the line segment characteristics to have universality in different models, firstly, right-angled bends subjected to model identification and classification are standardized, and the same characteristics used for positioning the AGV are extracted; and then establishing a global reference point set at the right angle bend.
Step 21: right angle bend feature normalization
Step 211: l-shaped quarter bend standardization
As shown in FIG. 5, (a) and (b) are an actual quarter bend model and a quarter bend model after table conversion, respectively, and the quarter bend of this type is a straight line L fitted to a wall surface with two intersecting surfaces2And L3The intersection point Key _ P and one of the fitting straight lines are used as fixed features for positioning, and in practical application, a straight line closest to the slope of the body abscissa axis direction is generally used as a Key straight line feature for positioning.
Step 212: standardization of T-shaped quarter bend
As shown in fig. 6, in the process of normalizing the characteristic of the quarter bend in this scenario, first, two parallel segment extensions and a second segment L which are abstractly fitted by two stacks of parcels and are parallel are obtained2Coordinate P of intersection point23And P21Then from P23And P21The arithmetic mean value is used for solving the Key point Key _ P and the solved Key point Key _ P and a second line segment L abstractly fitted by the wall surface2As a key feature of positioning, among others:
Figure BDA0001676860370000083
step 213: standardization of I-type quarter bend
As shown in fig. 7, in the process of normalizing the characteristic of the quarter bend in this scenario, first, two parallel segment extensions and a second segment L which are abstractly fitted by two stacks of parcels and are parallel are obtained2Coordinate P of intersection point23And P21Then from P23And P21The arithmetic mean value calculates the Key point Key _ P and calculates the Key point KeyP and a second line segment L obtained by abstract fitting of the wall surface2As a key feature of positioning, among others:
Figure BDA0001676860370000091
step 22: establishing a global set of reference points
In a standardized warehouse environment, the spatial position of the shelf is generally perpendicular to or parallel to one wall surface in the warehouse, and based on the current situation, when a global coordinate system is established, a global coordinate system xoy is established by taking a point abstractly fitted to a warehouse entrance as an origin, taking a direction perpendicular to a line segment abstractly fitted to a parcel stack as a y-axis and taking a direction parallel to the line segment abstractly fitted to the parcel stack as an x-axis; and establishing a global reference point set P ═ P { P } of the right-angle bend positioning key points under a global coordinate system1,P2,...,PnIn which P isi＝(xi,yi,θri)，i∈[1,n]，(xi,yi) Represents the coordinates of the intersection point of the two intersecting straight lines of the normalized quarter bend in the global coordinate system xoy, thetariRepresenting the angle between a straight line fitted abstractly to the wall and the x-axis of the global coordinate system xoy. As shown in fig. 8, the reference point sets are stored in the database in order.
Step 3: and establishing a trolley initial state coordinate system, acquiring trolley motion parameters by using an encoder inside a trolley body, and calculating initial pose estimation of the trolley in a global coordinate system by combining an omnidirectional mobile robot kinematics model.
As shown in FIG. 9, the geometrical center of the AGV car is used as the origin to establish the initial state body coordinate system x 'o' y 'of the car, wherein the y-axis direction is parallel to the direction of the front of the car, and the laser radar coordinate system x "o" y "in the figure is the body coordinate system x' o 'y' horizontally shifted by a distance n along the positive direction of the front of the car.
Step 31, establishing a kinematics model of the omnibearing moving trolley
Figure BDA0001676860370000092
In the above formula, (w)1,w2,w3,w4) The rotating speed of four wheels is calculated according to the wheel encoder data, r is the radius of the wheel, m and n respectively represent half of the width and the length of the vehicle body, and (v)x,vyAnd w) represents the moving speed of the mass center along the x axis, the y axis and the direction of the vehicle head respectively under the initial coordinate system of the vehicle body.
Step 32: and calculating the coordinate (x ') of the center of mass of the AGV at the moment t-kT under a coordinate system x ' o ' y ' of the vehicle body according to the dead reckoning principle 'k,y′k,θ′k) Comprises the following steps:
Figure BDA0001676860370000101
(Δxi，Δyi，Δθi) Indicating the offset of the AGV car center of mass during the i-th cycle. Since T is sufficiently small, it is assumed that within the period T, at the initial state coordinate (Δ x)i,Δyi,Δθi) Constant values, then there are:
Figure BDA0001676860370000102
step 33: calculating the barycenter coordinate (x) of the trolley under the global coordinate system xoy according to the conversion relation between the global coordinate system xoy and the coordinate system x ' o ' y ' of the trolley bodyk,yk,θk)：
Figure BDA0001676860370000103
Wherein (x)0,y0,θ0) Is the initial coordinate of the center of mass under the global coordinate system xoy, and theta is the included angle between the direction of the head of the trolley and the positive y direction of the global coordinate system xoy, (x'k,y′k,θ′k) And the coordinates of the center of mass of the trolley under the coordinate system x ' o ' y ' of the trolley body are represented.
Step 4: quarter turn feature identification
In the process of carrying out global positioning on the AGV based on the characteristics of the right angle bends, the accurate identification of the characteristics of the right angle bends is the first step of carrying out the global positioning, and the process of identifying the right angle bends is mainly divided into two steps of primary positioning and accurate judgment.
Step 41: preliminary positioning of quarter bend
Firstly, the AGV is manually remotely controlled to learn a preliminary path in a passage environment. When the AGV car travels near a quarter turn, the operating central control computer records the position (x) of the car at that position based on dead reckoningi,yi) And the positioning result is used as a primary judgment key point when the AGV trolley moves to the vicinity of the right-angled bend, and the primary judgment key point is stored in the database in sequence to generate a global path key point map for primary identification of the right-angled bend.
In the process of carrying out subsequent autonomous navigation on the AGV, adopting a pose tracking method to judge whether the AGV enters a neighborhood range of a global path key point map or not in real time, wherein the neighborhood range is defined by taking a key point as a center and having a radius of r1A detailed schematic view of the circular region of (a) is shown in fig. 10.
Step 42: quarter bend accurate determination
After the quarter bend is preliminarily positioned and judged, the environment in the preliminary positioning neighborhood range is further identified by utilizing the quarter bend judgment model in front, and whether the quarter bend characteristic conforming to the model exists in the neighborhood range or not and which type of quarter bend characteristic belongs to the neighborhood range is judged. The process mainly comprises two links: and (4) performing line segment feature extraction and right-angle bending model judgment based on the laser radar.
Step 421: line segment feature extraction based on laser radar
The method utilizes a split _ merge algorithm to extract line segment characteristics of laser radar scanning data, and the algorithm is mainly divided into three steps of segmentation, fitting and combination.
A segmentation stage: dividing point sets which do not belong to the same straight line, and dividing point cloud data scanned by a laser radar frame near a right-angled bend by adopting an IEPF (edge-edge particle Filter) clustering method to obtainA plurality of point set sets D ═ D (D) belonging to different straight lines1,D2,...,Dn)。
As shown in fig. 11, for laser data point D in a certain areaiFirstly, selecting head and tail points in the region as end points to fit a straight line liWhile the set point-to-straight distance threshold dthdTraversing all points between the head point and the tail point, and calculating the distance d from the point to the straight lineiAnd find out the point to the straight line liMaximum distance d ofmaxIf d ismax＜dthdIf not, fitting the point P with the maximum straight line distance from the initial point in the regionmaxFor the demarcation point, the point set in the area is divided into two point sets { Pj(xj,yj) 1, 2.. i } and { P |j(xj,yj) I j ═ i +1, i +2,.. and n }, then repeating the segmentation steps in each point set respectively, and repeating the operation until all the point sets meet the segmentation conditions, and finally obtaining a plurality of point set sets D ═ D (D) belonging to different straight lines1,D2,...,Dn)。
And (3) fitting: the set D (D) of the point sets after the division1,D2,...,Dn) The least squares line fitting is performed to obtain line segment representation L ═ L (L) for each set1,L2,...,Ln)，i∈[1,n]，Li＝(pi,θiStart, end, len), where piRepresenting a straight line L from the originiA distance of (a), thetaiRepresents the origin and LiThe perpendicular line of (A) and the positive direction of the x-axis, start and end respectively represent a line segment LiLen represents the line segment LiLength of (d).
A merging stage: set of line segments L ═ L (L)1,L2,...,Ln) The line segments meeting a certain threshold range are merged to prevent the same line segment from being divided into multiple segments to cause line segment overfitting, and two adjacent straight lines L are combinedi,LjThe merging criteria of (1) are: | pi-pj|＜δpAnd | θi-θj|＜δθWherein, deltapAnd deltaθRespectively as the merging threshold of the segment parameters, and merging the segment sets to obtain a new segment set L ═ (L ═ L)1,L2,...,Lm)。
Accurate extraction of scanning data of a frame of laser radar environment can be completed through three steps of segmentation, fitting and combination, but in an actual channel environment formed by a goods package stack, due to the fact that a certain gap exists between goods packages, a plurality of discontinuous collinear line segments can be extracted from scanning data of a frame of laser radar environment, as shown in figure 12.
In order to make the description of the quarter turn based on the line segment characteristics in the actual scene conform to the foregoing typical quarter turn model, the line segments extracted in the environment need to be subjected to invalid data elimination and line segment recombination again.
Invalid data elimination: mainly eliminates line segments fitted in the cracks between stacked goods and packages, and eliminates any line segment L under a laser radar coordinate systemi＝(pi,θiStart, end, len), there are the following two rejection principles: 1) for satisfying | | thetai|-90°|＜thdθIn the line segment of (1), if pi＜thdpRemoving the line segment; if p isi＞thdpAnd len _ i < dminRemoving; 2) for satisfying | | thetai|-180°|＜thdθOr | | | θi|-0°|＜thdθIf len _ i < dminRemoving; wherein the parameter thdθIs a discrimination threshold between the line segment polar diameter angle and 90 DEG, thdpThreshold value, d, to be satisfied for the length of the line segment pole diameterminIs the minimum segment length threshold.
And merging again: and adopting a split _ merge algorithm segment merging method to merge the segments without the invalid data to finally obtain a new segment set L' (L is equal to L)1,L2,...,Lk)。
Step 422: model decision based on line segment characteristics
Utilizing a typical quarter bend model to align a line segment set L' in a frame of laser radar near a quarter bend＝(L1,L2,...,Lk) And (5) carrying out model discrimination and classifying the model discrimination.
Step 5: global positioning and pose updating based on quarter turn characteristics
Step 51: reference coordinate association
After the AGV finishes the identification of the right-angle bend characteristics, the right-angle bend characteristics at the position need to be associated and matched with a preset global reference coordinate point.
In the embodiment, a nearest neighbor method is adopted to perform correlation matching on the reference coordinates, and in the matching process, the reference key point coordinates P in the quarter turn standardized model are estimated by combining with the dead reckoning result of the AGVi′(Xi′,Yi') search for a point P in a set of reference coordinate points using the "nearest neighbor" methodi', when a certain reference point P is searchedjAnd point PiWhen the Euclidean distance between P and P is minimum, P is consideredjIs the correct associated reference point for this right angle bend.
Step 52: AGV pose resolving based on line segment characteristics
As shown in FIG. 13, there are two Cartesian coordinate systems, the global coordinate system xoy and the lidar coordinate system x ' o ' y ', for which two intersecting straight lines L are known1＝(k1,b1,p1,θ1)，L2＝(k2,b2,p2,θ2) Coordinates A (x) of the intersection of straight linesA,yA) Wherein k isi,biRespectively representing the slope and intercept of the line, pi,θiRespectively is the corresponding polar diameter and polar angle of the straight line; the linear parameters obtained by measurement under the laser radar coordinate system are respectively L1′＝(k1′,b1′,p1′,θ1′)，L2′＝(k2′,b2′,p2′,θ2') straight line intersection coordinate A ' (x 'A,y′A) Let the position and attitude of the origin of the laser radar coordinate system under the global coordinate system be o' (x)o″,yo″,θo") wherein θoRepresenting the attitude orientation of the lidar in a global coordinate systemThe angle, which is represented by the angle between the positive x-axis direction of the radar coordinate system and the positive x-axis direction of the global coordinate system, can be obtained by combining fig. 13 and the above conditions:
Figure BDA0001676860370000131
Figure BDA0001676860370000132
the pose o ″ (x) of the laser radar in the global coordinate system can be obtained by bringing formula 14 into formula 13o,yo,θo)。
Step 53: AGV pose update
Herein, the laser radar and the AGV body belong to a rigid body connection state, and as shown in fig. 9, similarly, the position coordinates and attitude angles of the AGV in the global coordinate system can be obtained through the conversion of the laser radar and the AGV trolley coordinate system.
In fig. 9, the pose of the lidar obtained by using the quarter-turn feature in the global coordinate system is o ″ (x)o,yo,θo) Obtaining the position coordinate and attitude angle (x) of the AGV trolley under the global coordinate system xoy through coordinate conversionr,yr,θr)：
Figure BDA0001676860370000133
And finally, updating the relative positioning result of the AGV by using the global positioning result of the right-angled bend, eliminating accumulated errors in the relative positioning result and completing the global pose calibration of the AGV.
The method for calibrating the pose of the omnidirectional mobile robot based on the right-angle bend characteristic provides a solution for accurately positioning in an omnidirectional mobile robot channel in engineering by combining the structural linearization characteristic of a typical right-angle bend under the condition of fully utilizing environmental information acquired by a laser radar.
Finally, the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions, and all of them should be covered in the protection scope of the present invention.

Claims (6)

1. An omnibearing mobile robot pose calibration method based on right-angle bend characteristics is characterized by comprising the following steps:
s1, establishing a right-angle bend model, which comprises an L-shaped right-angle bend model, an I-shaped right-angle bend model and a T-shaped right-angle bend model;
the L-shaped right-angle bending model comprises two line segments which are abstractly fitted by two wall surfaces and are in a vertical intersection state and a first line segment which is abstractly fitted by the goods package stack, wherein one line segment of the two line segments which are abstractly fitted by the two wall surfaces and are in the vertical intersection state is parallel to the first line segment which is abstractly fitted by the goods package stack;
the T-shaped right-angled bend model comprises a second line segment abstractly fitted by one wall surface and two line segments in a parallel state abstractly fitted by two goods pack stacks, and the two line segments in the parallel state abstractly fitted by the two goods pack stacks are perpendicular to but not intersected with the second line segment abstractly fitted by one wall surface;
the I-shaped right-angle bending model comprises a third line segment which is abstracted and fitted by one wall surface and two line segments which are abstractly fitted by the two goods package stacks and are in a parallel state, and the third line segment abstracted and fitted by one wall surface is perpendicular to and intersected with the two line segments which are abstractly fitted by the two goods package stacks and are in the parallel state; or the I-shaped right-angle bending model comprises a fourth line segment which is abstractly fitted by a goods pack stack and two line segments which are abstractly fitted by two wall surfaces and are in a parallel state, and the fourth line segment which is abstractly fitted by a goods pack stack is perpendicular to and intersected with the two line segments which are abstractly fitted by the two wall surfaces and are in the parallel state;
s2, standardizing the right-angle bend model, and then establishing a global reference point set of the right-angle bend;
s3, establishing a trolley initial state coordinate system, and calculating initial pose estimation of the AGV in a global coordinate system;
s4, identifying characteristics of the straight corner bend;
s5, carrying out overall positioning and pose updating on the AGV based on the characteristics of the right-angled bend;
the L-shaped quarter bend model is as follows:
Figure FDA0003373514770000011
the T-shaped quarter bend model is as follows:
Figure FDA0003373514770000012
the I-shaped quarter bend model is as follows:
Figure FDA0003373514770000013
line segment L under two-dimensional rectangular coordinate systemi＝(pi,θiStart, end, len), where start _ q and end _ q represent the coordinates of the starting point and the starting point of the fitting line segment of the wall in front of the trolley, start (q +1) represents the coordinates of the starting point of the fitting line segment of the inner wall of the left side channel of the trolley, end (q-1) represents the coordinates of the starting point of the fitting line segment of the inner wall of the right side channel of the trolley, piRepresenting a straight line L from the originiA distance of (a), thetaiRepresents the origin and LiThe perpendicular line of (A) and the positive direction of the x-axis, start and end respectively represent a line segment LiLen represents the line segment LiLength of (d), deltadminAnd deltadmaxMinimum and maximum thresholds, delta, representing the Euclidean distance between two pointsdmaxUsed for judging whether the head and the tail of two adjacent line segments are intersected or not，δdminAnd the method is used for judging whether the distance between the head positions of two adjacent line segments is larger than a certain threshold value.
2. The method for calibrating the pose of the omni-directional mobile robot based on the characteristics of the right-angled bend according to claim 1, wherein in the step S2, the method for normalizing the characteristics of the L-shaped right-angled bend model comprises the following steps:
two line segments which are in a vertical intersection state and are abstractly fitted by two wall surfaces and one line segment which is in a vertical intersection state and is abstractly fitted by two wall surfaces are taken as fixed characteristics;
the method for standardizing the characteristics of the T-shaped quarter bend model comprises the following steps:
firstly, respectively calculating the coordinates of the intersection points of two parallel section extension lines and a second section, which are abstractly fitted by two bale stacks;
then, the arithmetic mean value of the coordinates of the two intersection points is obtained, and the arithmetic mean value is used as a Key point Key _ P;
and the Key point Key _ P and the second line segment are used as Key characteristics of positioning;
the method for standardizing the characteristics of the I-shaped quarter bend model comprises the following steps:
firstly, respectively solving the intersection point coordinates of two parallel line segment extensions and a third line segment abstractly fitted by stacking two goods packages;
then, the arithmetic mean value of the coordinates of the two intersection points is obtained, and the arithmetic mean value is used as a Key point Key _ P';
and the Key point Key _ P' and the third line segment are used as Key characteristics of positioning;
or the method for standardizing the characteristics of the I-shaped right-angle bending model comprises the following steps:
firstly, respectively solving the intersection point coordinates of two parallel line segment extensions and a fourth line segment abstractly fitted by two wall surfaces;
then, the arithmetic mean value of the coordinates of the two intersection points is calculated, and the arithmetic mean value is used as a Key point Key _ P ";
and the Key point Key _ P' and the fourth line segment are used as Key features of positioning.
3. The method for calibrating the pose of an omni-directional mobile robot based on the characteristics of right-angled curves according to claim 2, wherein in the step S2, when a global coordinate system is established; establishing a global coordinate system xoy by taking the abstractly fitted point of the warehouse entrance as an origin, taking the direction perpendicular to the abstractly fitted line segment of the parcel stack as a y-axis and taking the direction parallel to the abstractly fitted line segment of the parcel stack as an x-axis; and establishing a global reference point set P ═ P { P } of the right-angle bend positioning key points under a global coordinate system1,P2,...,PnIn which P isi＝(xi,yi,θri)，i∈[1,n]，(xi,yi) Represents the coordinates of the intersection point of the two intersecting straight lines of the normalized quarter bend in the global coordinate system xoy, thetariRepresenting the angle between a straight line fitted abstractly to the wall and the x-axis of the global coordinate system xoy.
4. The method for calibrating the pose of the omni-directional mobile robot based on the characteristics of the right-angled bend according to claim 3, wherein the step S3 specifically comprises the following sub-steps:
s31, establishing a body coordinate system x ' o ' y ' of the initial state of the AGV by taking the geometric center of the AGV as an original point, and establishing an omnidirectional mobile trolley kinematics model:
Figure FDA0003373514770000031
wherein (w)1,w2,w3,w4) Representing the rotating speed of four wheels of the AGV trolley, r is the radius of the wheels, m and n respectively represent half of the width and the length of the AGV body, (v)x,vyW) represents the moving speed of the mass center along the x axis, the y axis and the vehicle head direction respectively under the vehicle body initial coordinate system;
s32, calculating the barycenter coordinate (x ') under the body coordinate system x ' o ' y ' at the time t-kT according to the dead reckoning principle 'k,y′k,θ′k)，θ'kCoordinates representing the direction of the head of the trolley and the mass center of the trolleyAn included angle of a positive y direction of the system, T represents a dead reckoning sampling period, and k represents a positive integer;
s33, calculating the barycenter coordinate (x) of the trolley under the global coordinate system xoy according to the conversion relation between the global coordinate system xoy and the body coordinate system x ' o ' y ' of the trolleyk,yk,θk)，θkAnd the included angle between the direction of the head of the trolley and the positive y direction of the global coordinate system is shown.
5. The method for calibrating the pose of the omni-directional mobile robot based on the characteristics of the right-angled bend according to claim 4, wherein the step S4 specifically comprises the following sub-steps:
s41, firstly, manually remotely controlling the AGV to learn a primary path in a channel environment, and generating a global path key point map preliminarily identified by a right-angled bend; in the process of carrying out subsequent autonomous navigation on the AGV, judging whether the AGV enters the neighborhood range of the global path key point map or not in real time by adopting a pose tracking method according to a prior map, wherein the neighborhood range is defined by taking the key point as the center and the radius of the key point as r1The circular region of (a);
s42, extracting line segment characteristics based on the laser radar, and extracting the line segment characteristics of laser radar scanning data by using a split _ merge algorithm to obtain a line segment set L'; removing invalid data of the line segments extracted from the environment and combining the line segments again to obtain a new line segment set L';
s43, performing model discrimination on a line segment set L 'in a frame of laser radar near the right-angle bend by using a right-angle bend model and classifying the line segment set L'.
6. The method for calibrating the pose of the omni-directional mobile robot based on the characteristics of the right-angled bend according to claim 5, wherein the step S5 specifically comprises the following sub-steps:
s51, associating reference coordinates; after the AGV finishes the identification of the right-angled bend characteristics, the right-angled bend characteristics at the position of the AGV need to be associated and matched with a preset global reference coordinate point to obtain a correct associated reference point;
s52. baseAnd performing AGV pose calculation on line segment characteristics, and performing pose calculation on the laser radar by using the standardized quarter bend model line segment and the key characteristics extracted from the scanning data of the laser radar to obtain a pose o' of the laser radar under a global coordinate system xoy (x is the x characteristic)o,yo,θo)；
And S53, updating the position and attitude angles (x) of the AGV under the global coordinate system xoy through the conversion of the laser radar and the AGV coordinate systemr,yr,θr) And updating the relative positioning result of the AGV by using the global positioning result of the right-angled bend.
CN201810529469.2A 2018-05-29 2018-05-29 Omni-directional mobile robot pose calibration method based on right-angle bend characteristics Active CN109000649B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810529469.2A CN109000649B (en) 2018-05-29 2018-05-29 Omni-directional mobile robot pose calibration method based on right-angle bend characteristics

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810529469.2A CN109000649B (en) 2018-05-29 2018-05-29 Omni-directional mobile robot pose calibration method based on right-angle bend characteristics

Publications (2)

Publication Number Publication Date
CN109000649A CN109000649A (en) 2018-12-14
CN109000649B true CN109000649B (en) 2022-02-15

Family

ID=64573689

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810529469.2A Active CN109000649B (en) 2018-05-29 2018-05-29 Omni-directional mobile robot pose calibration method based on right-angle bend characteristics

Country Status (1)

Country Link
CN (1) CN109000649B (en)

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110146110B (en) * 2019-05-20 2022-11-15 哈尔滨工程大学 Mismatching judgment method for ICNN data association of robot line characteristics in indoor environment
CN110260875A (en) * 2019-06-20 2019-09-20 广州蓝胖子机器人有限公司 A kind of method in Global motion planning path, Global motion planning device and storage medium
CN112230256B (en) * 2019-07-15 2024-04-09 苏州宝时得电动工具有限公司 Autonomous robot, positioning calibration method and device thereof, and storage medium
CN110442136A (en) * 2019-08-12 2019-11-12 北京特种机械研究所 A kind of any attitude air navigation aid of Mecanum wheel AGV
CN110823225A (en) * 2019-10-29 2020-02-21 北京影谱科技股份有限公司 Positioning method and device under indoor dynamic situation
CN111060093B (en) * 2019-12-31 2021-07-30 芜湖哈特机器人产业技术研究院有限公司 Mobile robot positioning method and system based on single corner folding plate
CN111142117B (en) * 2019-12-31 2021-11-05 芜湖哈特机器人产业技术研究院有限公司 Hybrid navigation map construction method and system fusing corner plates and occupied grids
CN112327325B (en) * 2020-09-16 2024-05-14 安歌智慧科技(上海)有限公司 Method for improving 2D-SLAM precision and stability based on characteristic road sign
CN112068128B (en) * 2020-09-19 2024-02-02 重庆大学 Straight-line scene line segment type radar data processing and pose acquisition method
CN112577494B (en) * 2020-11-13 2022-11-18 上海航天控制技术研究所 SLAM method, electronic device and storage medium suitable for lunar vehicle
CN112731354B (en) * 2020-12-24 2024-04-05 芜湖哈特机器人产业技术研究院有限公司 Self-calibration method for laser radar pose on AGV
CN112699979A (en) * 2021-01-04 2021-04-23 北京国腾联信科技有限公司 Cargo movement detection method and device and electronic equipment
CN113391318B (en) * 2021-06-10 2022-05-17 上海大学 Mobile robot positioning method and system

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103941264B (en) * 2014-03-26 2017-04-19 南京航空航天大学 Positioning method using laser radar in indoor unknown environment
CN104729500B (en) * 2015-02-15 2017-09-08 南京航空航天大学 A kind of laser navigation AGV global localization method
CN105987697B (en) * 2016-04-26 2019-01-29 重庆大学 The wheeled AGV navigation locating method of Mecanum and system under a kind of quarter bend
CN106643805B (en) * 2016-12-30 2020-07-14 上海交通大学 Method for calibrating position of laser positioning sensor in AGV
CN107065887B (en) * 2017-05-26 2020-10-30 重庆大学 Reverse navigation method in channel of omnidirectional mobile robot
CN107933557B (en) * 2017-11-17 2019-09-13 吉林大学 A kind of vehicle right-angled bend control system and its control method based on multi-cam
CN108036792A (en) * 2017-12-11 2018-05-15 苏州中德睿博智能科技有限公司 A kind of data fusion method of odometer for mobile robot and measurement pose

Also Published As

Publication number Publication date
CN109000649A (en) 2018-12-14

Similar Documents

Publication Publication Date Title
CN109000649B (en) Omni-directional mobile robot pose calibration method based on right-angle bend characteristics
CN111337941B (en) Dynamic obstacle tracking method based on sparse laser radar data
CN107741234B (en) Off-line map construction and positioning method based on vision
Jeong et al. Visual SLAM with line and corner features
Nieto et al. Recursive scan-matching SLAM
CN112014857A (en) Three-dimensional laser radar positioning and navigation method for intelligent inspection and inspection robot
CN109186606B (en) Robot composition and navigation method based on SLAM and image information
Ye et al. Object detection and tracking using multi-layer laser for autonomous urban driving
CN111158355A (en) Automatic navigation cloud server and automatic navigation control method
JP2005070043A (en) Self-position recognizing device and method of intelligent system using artificial mark, and the intelligent system using the same
CN109100731B (en) Mobile robot positioning method based on laser radar scanning matching algorithm
Eppenberger et al. Leveraging stereo-camera data for real-time dynamic obstacle detection and tracking
US20200088882A1 (en) Real-time 3d mapping using 3d-lidar
Yin et al. Radar-on-lidar: metric radar localization on prior lidar maps
Zhao et al. BIM-based indoor mobile robot initialization for construction automation using object detection
Zhu et al. Indoor multi-robot cooperative mapping based on geometric features
CN114998276A (en) Robot dynamic obstacle real-time detection method based on three-dimensional point cloud
Avgeris et al. Single vision-based self-localization for autonomous robotic agents
Wei et al. GCLO: Ground constrained LiDAR odometry with low-drifts for GPS-denied indoor environments
Wei et al. Ground-SLAM: Ground constrained LiDAR SLAM for structured multi-floor environments
Cui et al. Monte-Carlo localization in underground parking lots using parking slot numbers
Lu et al. A lightweight real-time 3D LiDAR SLAM for autonomous vehicles in large-scale urban environment
Li et al. Pallet detection and localization with RGB image and depth data using deep learning techniques
Jung et al. Graph SLAM for AGV using geometrical arrangement based on lamp and SURF features in a factory environment
Wu et al. Global localization in repetitive and ambiguous environments

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
GR01 Patent grant
GR01 Patent grant