阅读说明：本技术 含有激光slam的odom与dm地标组合移动机器人及导航方法 (ODOM and DM landmark combined mobile robot containing laser SLAM and navigation method ) 是由 魏博 杨茸 舒思豪 李艳生 张毅 于 2020-12-23 设计创作，主要内容包括：本发明公开了含有激光SLAM的ODOM与DM地标组合移动机器人及导航方法,首先通过建立里程计运动模型预测粒子位置,并利用DM码数据信息修正其参数,再建立DM地标观测模型更新粒子重要性权重；其后在改进的粒子重采样过程中根据相机观测似然增加固定数目的随机粒子,以获得更准确的粒子分布来有效提高定位精度,同时减弱了蒙特卡洛(MCL)算法粒子退化,改进算法解决了机器人位置漂移与劫持问题；最后通过DM地标轨迹纠偏不断修正其位姿,提高机器人全局导航精度,基于组合导航机器人平台进行实验。(The invention discloses an ODOM and DM landmark combined mobile robot containing laser SLAM and a navigation method, firstly predicting the position of a particle by establishing a motion model of a speedometer, correcting the parameter by using DM code data information, and then establishing an observation model of a DM landmark to update the importance weight of the particle; then, in the improved particle resampling process, a fixed number of random particles are added according to the observation likelihood of the camera, so that more accurate particle distribution is obtained to effectively improve the positioning precision, meanwhile, the particle degradation of a Monte Carlo (MCL) algorithm is weakened, and the problem of robot position drift and hijack is solved by improving the algorithm; and finally, continuously correcting the pose of the robot by DM landmark trajectory rectification, improving the global navigation precision of the robot, and performing experiments based on the combined navigation robot platform.)

1. The navigation method of the ODOM and DM landmark combined mobile robot containing the laser SLAM is characterized by comprising the following steps:

s1: using a laser SLAM navigation system, and enabling the robot to move to an initial DM landmark position by controlling a brushless motor with an encoder;

s2: the robot collects an initial DM landmark by using a positioning camera, the robot system identifies the collected photos to obtain the distance between the robot and the DM landmark and the pose angle deviation alpha, and the robot system calculates the initial position of the robot on a global map;

s3: after the robot system gives a starting point and a target point of the robot, the robot system carries out path planning according to distance information between the robot and the DM coordinates, pose angle deviation information alpha, initial position information of the robot on a global map, starting point information of the given robot and target point information, and the robot obtains a path planning instruction;

s4: the robot moves to the position of the DM landmark of the first target point through the path planning instruction, the robot collects the DM landmark of the first target point by using the positioning camera, and the robot system identifies the collected photo to obtain a relative pose value a1 between the robot and the DM landmark;

s5: the robot system judges whether the relative pose value a1 is larger than alpha or not, if the pose angle deviation value is smaller than alpha, the next step S6 operation is carried out, if the pose deviation value is larger than alpha, the improved Monte Carr positioning algorithm module processes the related parameter information and predicts to obtain a new target point beta, the beta is input into the path planning module to carry out path planning and movement to obtain a pose deviation alpha 2, and the step S5 is carried out to continue judging until the error is smaller than alpha;

s6: and if the pose angle deviation value is smaller than alpha, the mobile robot puts the goods on a goods shelf beside the DM code and returns to the goods sorting table.

2. The method as claimed in claim 1, wherein the step of preprocessing the relevant parameters by the monte carlo location algorithm module in the system comprises: tracking the pose of the robot in a known map using particle filtering that simulates a motion state with a particle representation confidence Xt; distributing a specific number of particles on a map according to the odometer motion model, wherein the larger the odometer error is, the more dispersed the particle swarm is; determining the importance weight of the particles according to the DM landmark observation model, and after recognizing the DM landmark, the robot resamples the particle screening and adds the fixed particles to complete the approximate state estimation; and the robot adjusts the pose to be within the alpha error through path planning according to the state estimation.

3. The method as claimed in claim 2, wherein the relevant parameters include: laser radar parameters, camera parameters, and odometry parameters.

4. The method as claimed in claim 2, wherein the particle filter is used to locate the position of the object to be detected to obtain a presumed pose x_tThe pose x_tObey P (x)_t|u_t,x_t-1) By an odometer motion model method, in whichRepresenting the coordinates embedded in the robot.

5. The method as claimed in claim 2, wherein after the mobile robot recognizes the DM landmark, the positioning error between the ODOM and the DM landmark can be treated as independent gaussian white noise; fusing the positioned position information by using Kalman filtering, wherein the position of the mobile robot is a Kalman filtering state quantity, and the formula X_t＝X_t-1+R(Δx_tΔy_tΔθ_t)^TThe prediction and the update of the robot state quantity are realized by the dead reckoning model.

6. The method as claimed in claim 2, wherein the particle swarm is substituted into the motion model to obtain the next step position of the particle swarm, the geometric difference between the predicted and observed distances is calculated according to the DM landmark measurement model, and the weight is added to each particle according to the difference of the distances, the relationship between the weight and the distance is in accordance with the Gaussian distribution bell curve, in order to obtain the weight of the particleEstablishing DM landmark measuring model to obtain landmark measuring likelihood

7. The method of claim 2, wherein in the resampling process, the less weighted particles are discarded in the new particle group, and in the resampling operation, whether to resample is determined according to the number of effective particles; the set of normalized weights for all the particle numbers N isThe number of effective particles N is determined by placing a certain number of additional particles directly at the corresponding positions_effeCan be defined as:where N is the number of particles and J is the increased number of particles, then the relationship between the number of resampling times and the threshold is as follows:when N is present_effeWhen the system FLAG is smaller than the threshold or the system FLAG identified by the camera is 1, resampling is started, and epsilon is a constant.

8. The ODOM and DM landmark combined mobile robot containing the laser SLAM is characterized in that the mobile robot is internally provided with an execution module for executing the steps S1-S6.

Technical Field

The invention relates to the field of robot navigation, in particular to an ODOM and DM landmark combined mobile robot containing a laser SLAM and a navigation method.

Background

In recent years, with the development of intelligent manufacturing and intelligent logistics, flexible intelligent carrying equipment is widely applied in various industries. The wheel-type mobile robot can efficiently and quickly convey stored articles to a sorting place flexibly in the intelligent warehousing system, a plurality of robots can be used for scheduling in a coordinated mode, the wheel-type mobile robot can automatically plan a route by updating a target point, and a high flexible carrying system is formed. However, the precision and flexibility of the mobile robot are always a difficulty in popularization and application in the industrial field, and the two navigation modes of tape navigation and two-dimensional code navigation are simple to apply, but the tape navigation precision is low, the two-dimensional code navigation flexibility is low, the landmark identification rate is easily influenced by external conditions such as illumination and the like, the two navigation modes cannot be used in the industrial environment, the practicability and the robustness are low, the two navigation modes are difficult to maintain in the later period, and the positioning and navigation performance of the conventional SLAM technology depends on the precision of a sensor and the advantages and disadvantages of an algorithm, so that the problems of robot position drift and hijacking are easily caused.

Disclosure of Invention

The invention aims to solve the technical problems of the problem that the navigation precision of a logistics robot decreases along with the accumulated displacement and the technical problem of the drift and hijack of the position of the robot, and aims to provide an ODOM and DM landmark combined mobile robot containing a laser SLAM and a navigation method, which can avoid the establishment of a complex visual landmark library and improve the positioning precision of the robot to a greater extent.

The invention is realized by the following technical scheme:

the ODOM and DM landmark combined mobile robot containing the laser SLAM and the navigation method comprise the following steps:

s1: using a laser SLAM navigation system, and enabling the robot to move to an initial DM landmark position by controlling a brushless motor with an encoder;

s2: the robot collects an initial DM landmark by using a positioning camera, the robot system identifies the collected photos to obtain the distance between the robot and the DM landmark and the pose angle deviation alpha, and the robot system calculates the initial position of the robot on a global map;

s3: after the robot system gives a starting point and a target point of the robot, the robot system carries out path planning according to distance information between the robot and a DM coordinate, pose angle deviation information alpha, initial position information of the robot on a global map, starting point information of the given robot and target point information, and the robot obtains a path planning instruction to solve the drifting problem of the robot;

s4: the robot moves to the position of the DM landmark of the first target point through the path planning instruction, the robot collects the DM landmark of the first target point by using the positioning camera, and the robot system identifies the collected photo to obtain a relative pose value a1 between the robot and the DM landmark;

s5: the robot system judges whether the relative pose value a1 is larger than alpha or not, if the pose angle deviation value is smaller than alpha, the next step S6 operation is carried out, if the pose deviation value is larger than alpha, the improved Monte Carr positioning algorithm module processes the related parameter information and predicts to obtain a new target point beta, the beta is input into the path planning module to carry out path planning and movement to obtain a pose deviation alpha 2, and the step S5 is carried out to continue judging until the error is smaller than alpha;

s6: and if the pose angle deviation value is smaller than alpha, the mobile robot puts the goods on a goods shelf beside the DM code and returns to the goods sorting table.

Wherein, the SLAM-based odometer (Odom) and Data Matrix (DM) landmark combined navigation method; firstly, predicting the particle position by establishing a odometer motion model, correcting the parameter by using DM code data information, and then establishing a DM landmark observation model to update the particle importance weight; then, in the improved particle resampling process, a fixed number of random particles are added according to the observation likelihood of the camera, so that more accurate particle distribution is obtained to effectively improve the positioning precision, meanwhile, the particle degradation of a Monte Carlo (MCL) algorithm is weakened, and the problem of robot position drift and hijack is solved by improving the algorithm; and finally, continuously correcting the pose of the robot by DM landmark trajectory rectification, improving the global navigation precision of the robot, and performing experiments based on the combined navigation robot platform.

Further, the process of preprocessing the relevant parameters by the monte carlo location algorithm module in the system comprises: tracking the pose of the robot in a known map using particle filtering that simulates a motion state with a particle representation confidence Xt; distributing a specific number of particles on a map according to the odometer motion model, wherein the larger the odometer error is, the more dispersed the particle swarm is; determining the importance weight of the particles according to the DM landmark observation model, and after recognizing the DM landmark, the robot resamples the particle screening and adds the fixed particles to complete the approximate state estimation; and the robot adjusts the pose to be within the alpha error through path planning according to the state estimation.

Further, the relevant parameters include: laser radar parameters, camera parameters, and odometry parameters.

Further, the particle filter is used for positioning to obtain a presumed pose x_tThe pose x_tObey P (x)_t|u_t,x_t-1) By an odometer motion model method, in whichRepresenting the coordinates embedded in the robot.

Further, after the mobile robot identifies the DM landmark, the positioning error between the odometer and the DM landmark can be treated as mutually independent white gaussian noise; fusing the positioned position information by using Kalman filtering, wherein the position of the mobile robot is a Kalman filtering state quantity, and the formula X_t＝X_t-1+R(Δx_tΔy_tΔθ_t)^TThe prediction and the update of the robot state quantity are realized by the dead reckoning model.

Further, the particle swarm is brought into a motion model to obtain the next step position of the particle swarm, a geometric difference value of the prediction distance and the observation distance is calculated according to the DM landmark measurement model, weight is added to each particle according to the difference of the distances, the relation between the weight and the distance conforms to a Gaussian distribution bell-shaped curve, and in order to obtain the weight of the particleEstablishing DM landmark measuring model to obtain landmark measuring likelihood

Furthermore, in the importance sampling process, an importance weight and a sample set are generated according to each observation information recursion, but as the iteration times increase, a plurality of particle weights are degenerated to 0, so that the particle degeneration is caused. After DM landmark information is introduced, additional particles are directly placed at corresponding positions according to observation likelihood, particle degradation can be effectively inhibited, and positioning accuracy is improved. In the resampling process, the particles with smaller weight may be discarded in a new particle group, and in the resampling operation, whether to resample or not is determined according to the number of effective particles. The set of normalized weights for all the particle numbers N isThe number of effective particles N is determined by placing a certain number of additional particles directly at the corresponding positions_effeCan be defined as:

where N is the number of particles and J is the increased number of particles, then the relationship between the number of resampling times and the threshold is as follows:

when N is present_effeAnd when the FLAG is smaller than the threshold value or the FLAG of the system identified by the camera is 1, resampling is started, wherein epsilon is a constant and can be selected according to the actual situation. According to the method, the particle degradation can be effectively avoided and more accurate particle distribution can be obtained.

Further, the mobile robot is built with an execution module for executing the steps S1-S6.

Compared with the prior art, the invention has the following advantages and beneficial effects:

the invention relates to an ODOM and DM landmark combined mobile robot containing a laser SLAM and a navigation method, wherein the ODOM and DM landmark combined mobile robot corrects the motion model parameters of a speedometer by identifying the position information encapsulated by a DM code, establishes a DM landmark observation model, modifies the weight and resampling of particles, and completes accurate global positioning by an improved Monte Carlo positioning algorithm; the navigation method can avoid establishing a complex visual landmark library, improve the positioning accuracy of the robot to a greater extent, and solve the problems of robot drifting, hijacking and the like.

Drawings

The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:

FIG. 1 is a flow chart of the Monte Carlo algorithm;

FIG. 2 is a graph comparing state estimates for DM-MCL and KLD-MCL particle filters;

FIG. 3 is a schematic diagram of a motion path of a combined robot simulating a warehousing environment;

FIG. 4 is a schematic view of particle distribution of the combined navigation DM-MCL algorithm;

fig. 5 is a comparison graph of the combined navigation effect.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to examples and accompanying drawings, and the exemplary embodiments and descriptions thereof are only used for explaining the present invention and are not meant to limit the present invention.

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that: it is not necessary to employ these specific details to practice the present invention. In other instances, well-known structures, circuits, materials, or methods have not been described in detail so as not to obscure the present invention.

Throughout the specification, reference to "one embodiment," "an embodiment," "one example," or "an example" means: the particular features, structures, or characteristics described in connection with the embodiment or example are included in at least one embodiment of the invention. Thus, the appearances of the phrases "one embodiment," "an embodiment," "one example" or "an example" in various places throughout this specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics may be combined in any suitable combination and/or sub-combination in one or more embodiments or examples. Further, those of ordinary skill in the art will appreciate that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.

In the description of the present invention, it is to be understood that the terms "front", "rear", "left", "right", "upper", "lower", "vertical", "horizontal", "high", "low", "inner", "outer", etc. indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience of description and simplicity of description, and do not indicate or imply that the referenced devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and therefore, are not to be construed as limiting the scope of the present invention.

Examples

As shown in fig. 1, the ODOM and DM landmark combined mobile robot with laser SLAM and navigation method of the present invention includes the following steps:

s1: using a laser SLAM navigation system, and enabling the robot to move to an initial DM landmark position by controlling a brushless motor with an encoder;

s2: the robot collects an initial DM landmark by using a positioning camera, the robot system identifies the collected photos to obtain the distance between the robot and the DM landmark and the pose angle deviation alpha, and the robot system calculates the initial position of the robot on a global map;

s3: after the robot system gives a starting point and a target point of the robot, the robot system carries out path planning according to distance information between the robot and a DM coordinate, pose angle deviation information alpha, initial position information of the robot on a global map, starting point information of the given robot and target point information, and the robot obtains a path planning instruction to solve the drifting problem of the robot;

s4: the robot moves to the position of the DM landmark of the first target point through the path planning instruction, the robot collects the DM landmark of the first target point by using the positioning camera, and the robot system identifies the collected photo to obtain a relative pose value a1 between the robot and the DM landmark;

s5: the robot system judges whether the relative pose value a1 is larger than alpha or not, if the pose angle deviation value is smaller than alpha, the next step S6 operation is carried out, if the pose deviation value is larger than alpha, the improved Monte Carr positioning algorithm module processes the related parameter information and predicts to obtain a new target point beta, the beta is input into the path planning module to carry out path planning and movement to obtain a pose deviation alpha 2, and the step S5 is carried out to continue judging until the error is smaller than alpha;

s6: and if the pose angle deviation value is smaller than alpha, the mobile robot puts the goods on a goods shelf beside the DM code and returns to the goods sorting table.

Wherein, the odometer (Odom) based on SLAM technology and the Data Matrix (DM) landmark combined navigation method; firstly, predicting the particle position by establishing a odometer motion model, correcting the parameter by using DM code data information, and then establishing a DM landmark observation model to update the particle importance weight; then, in the improved particle resampling process, a fixed number of random particles are added according to the observation likelihood of the camera, so that more accurate particle distribution is obtained to effectively improve the positioning precision, meanwhile, the particle degradation of a Monte Carlo (MCL) algorithm is weakened, and the problem of robot position drift and hijack is solved by improving the algorithm; and finally, continuously correcting the pose of the robot by DM landmark trajectory correction, and improving the global navigation precision of the robot. And carrying out experiments based on the combined navigation robot platform.

In a real warehousing environment, the surrounding environment information of the mobile robot is relatively fixed, and the data of the odometer can be optimized by using a small amount of information such as DM landmarks.

Particle filtering for positioning to obtain a presumed pose x_tThe pose obeys P (x)_t|u_t,x_t-1) By means of odometer motion models(motion _ model _ order) method for implementing samplingRepresenting coordinates embedded in the robot, and motion information u can be extracted for extracting relative distance_tThe method comprises the following three steps: initial rotation delta_r1D translation delta_tAnd a second rotation δ_r2. Two rotation values and one translation value are respectively calculated, and the calculation results are as follows:

robot passes x_t-1Angle of rotation of positionThen passes through a translation distanceFinally following another rotation angleObtaining the final pose x with actual errors_t：

Robot passes x_t-1Angle of rotation of positionThen passes through a translation distanceFinally following another rotation angleObtaining the final pose x with actual errors_t：

By outputting the final pose X with errors_t(same as above x)_t) And combining with the DM data information, and correcting the error, wherein the odometer track reckoning model is as follows:

X_t＝(x_T,y_T,θ_T)^T (8)

X_t＝X_t-1+R(Δx_t Δy_t Δθ_t)^T (9)

in the above formula X_tAnd X_t-1Corresponding to the position coordinates of the robot at the time t and the time t-1, R is a rotation matrix from a robot coordinate system to a world coordinate system, theta, alpha and beta are rotation angles of the rotation matrix, and delta x_t,Δy_t,Δθ_t) And calculating the difference value of the data of the two positioning moments of the odometer.

When the mobile robot identifies the DM landmark, the positioning error of the odometer and the DM landmark can be regarded as mutually independent white Gaussian noiseSound processing^[15]. The method is characterized in that the positioned position information is fused by Kalman filtering (Kalman Filter), the position of the mobile robot is a Kalman filtering state quantity, the prediction and the update of the robot state quantity are realized by a dead reckoning model of a formula (9), and information after DM landmark extraction processing provides an observation value for an observation model. The system state update equation is reconstructed as follows:

Z_t＝BX_t+V_t (12)

X_t＝X^- _t+K_t(Z_t-AX^- _t) (13)

in the above formula, A is a robot state transition matrix, B is an observation matrix, Z_tFor locating observations of DM codes, W_tAnd V_tRespectively, the noise of the system and the measurement error noise in the observation process. After data fusion, X^- _tIs a state estimator, K_tIs a gain matrix for kalman filtering. And X_tThe position information after the odometer is fused with the DM landmark.

When the particle swarm is brought into a motion model, the next step position of the particle swarm is obtained, a geometric difference value of prediction and observation distances is calculated according to a DM landmark measurement model, weights are added to each particle according to the difference of the distances, and the relation between the weights and the distances accords with a Gaussian distribution bell-shaped curve; to obtain the weight of particlesThe method needs to find out the landmark measurement likelihood by establishing a DM landmark measurement model (landmark _ measurement _ model)In the equations (14) and (15), the map is m, j is the corresponding landmark, i is the landmark feature, r is the apparent measured distance of the azimuth phi, s is the number,for features observed by the camera, assuming that their noise is independent of each other, there are:

when the robot does not read the DM landmark in the motion process, the status flag bit is set to be 0, and when the DM landmark is read to be 1, the likelihood P is measured for the laser sensor_g(z_t|x_tM), the sensor measurement probability is as follows:

n is the total number of particles, then each particleThe weight of (A) is:

and satisfies the normalization condition

In the importance sampling process, an importance weight and a sample set are generated according to each observation information recursion, but as the iteration times increase, a plurality of particle weights are degenerated to 0, so that the particle degeneration is caused. After DM landmark information is introduced, additional particles are directly placed at corresponding positions according to observation likelihood, particle degradation can be effectively inhibited, and positioning accuracy is improved. During resampling, the less weighted particles may be discarded in a new population of particles, and the resampling may be performedIn the sampling operation, whether to resample or not is determined according to the number of effective particles. The set of normalized weights for all the particle numbers N isThe number of effective particles N is determined by placing a certain number of additional particles directly at the corresponding positions_effeCan be defined as:

where N is the number of particles and J is the increased number of particles, then the relationship between the number of resampling times and the threshold is as follows:

when N is present_effeAnd when the FLAG is smaller than the threshold value or the FLAG of the system identified by the camera is 1, resampling is started, wherein epsilon is a constant and can be selected according to the actual situation. According to the method, the particle degradation can be effectively avoided and more accurate particle distribution can be obtained.

In order to improve the positioning accuracy of the algorithm, the importance of sampling is evaluated according to the observation likelihood of the camera, and additional particles are directly placed at corresponding positions according to the observation likelihood in the resampling process; when no DM landmarks are identified, camera sampling failures will be evaluated by laser likelihood for particle weight. The basic steps of the DM-MCL algorithm are as follows:

inputting: weighted subset X at time t-1_t-1Control amount u_tAnd a sensor observation value z_t。

And (3) outputting: weighted particle subset X at time t_t。

Step 1X_t-1And setting a null value, wherein n is 0.

Step 2 from step 1 to step M, from X_t-1In which a sample is randomly takenThrough the motion model equationObtaining the pose of the particle by algorithmLandmark measurement according to DM landmarkThe algorithm performs importance sampling and calculates the corresponding particle weights, as shown in equation (17).

And step 3: particle weight normalization process

And 4, step 4: the updated particle set is resampled, and the number of sampled particles is updated according to equations (18) and (19).

And 5: the robot outputs the state of the pose at the moment t by taking the mathematical expectation of the particle set as:

step 6: and returning to the second step when n is n + 1.

The combined navigation robot adopts two driving wheels for differential driving in the middle, the two driving wheels are respectively provided with two universal wheels for supporting and guiding in front and at the back of the same reference of the driving wheels, the motor adopts a direct current brushless motor with a Z5BLDXX belt encoder, and the moving speed of the robot is V-0.5 m/s. A WISS-AGVXF high-frequency camera is arranged downwards in the center of the robot, positioning information in the walking process is obtained by sampling and identifying DM codes at a high speed of 50 times per second, and a laser radar is arranged right above the camera and used for map construction and navigation. The front end of the robot is provided with an infrared obstacle avoidance device to prevent the robot from colliding.

The controller adopts an autonomously developed special combined navigation controller STM32F407ZGT6, and a CPU is designed based on a Cortex-M432-bit RISC kernel and has a master frequency of 168 MHz. The controller integrates various control and communication interfaces such as motor control and drive, Ethernet NUC communication, laser radar, infrared obstacle avoidance, a vision sensor and WIFI communication, a CPU of a PC of the Inter NUC is used for Ubuntu system and ROS communication based on Intel core i 78809G and dominant frequency 3.1GHz, and all modules and functions required by combined navigation are realized by the system.

In order to better prove the positioning performance of the navigation method and the related algorithm, a simulated logistics storage environment is set up in a laboratory field, a DM landmark is pasted beside a sorting table in the simulated field, and the front of each goods shelf is pasted with (i) to (iii) number DM codes to simulate different goods shelves to store different goods. And verifying the robot positioning algorithm and the navigation control precision of the combined navigation.

To verify the effectiveness of the improved monte carlo localization algorithm (DM-MCL) for integrated navigation, a study was first conducted by simulation experiments, and compared with adaptive monte carlo (KLD-MCL) simulation results, and computational simulation was conducted on Matlab2016a, with the DM-MCL set parameters as follows: c is 0.7, u is 0.1, m is 5, z is 0.01, the initial number of particles used in the experiment is 2000, and the fixed number of particles is increased by 500. FIG. 2 is a comparison of the error of the state estimation of DM-MCL and KLD-MCL algorithms, with DM-MCL being the closest to the true state under the same conditions; table 1 shows that the filter performance of the two algorithms is compared, and it can be seen that the DM-MCL algorithm has a high percentage of effective particles and a relatively short operation time, which weakens the particle degradation and improves the processing capability. Thereby illustrating the advantages of the improved monte carlo algorithm over the adaptive monte carlo algorithm.

TABLE 1 comparison of Filter Performance for two algorithms

In order to perform experimental verification on the DM-MCL algorithm of the Robot, a simulated warehousing environment is mapped by using the combined navigation Robot Operating System (ROS) mapping node. The robot performs verification through the path movement of the multitask navigation A → B → C → D point, and DM landmarks are respectively stuck at the positions of (i) - (v) in the figure 3.

The change of the particles after the combined navigation robot recognizes the DM landmark is shown in fig. 4(a) - (C). The yellow dots represent the robot, the surrounding red arrows represent the particles, the arrow direction is the direction of the robot movement, and the green areas represent increasing fixed particle counts. Through contrast of different color particles in the graph, when the DM landmark is detected for the first time, the number of the green particles is increased and converged, if the DM landmark is detected for the first time, as shown in FIG. 5(a), the green particles disappear after the DM landmark is separated, and the pose drifts along with the movement of the robot, and if the DM landmark is identified again, the robot corrects the pose of the robot by combining the control value and the observation value, increases and converges the green particles, as shown in FIG. 5(b), the process is circulated, and the positioning accuracy of the combined navigation robot is improved.

Table 2 shows that the errors of the positioning by the two algorithms vary with the number of steps of the mobile node, and after repeated measurement and averaging by the laser range finder and the DM landmark dedicated camera, it can be seen that the error of the MCL positioning algorithm in the motion process is about 8.90cm, while the average value of the positioning errors of the improved monte carlo algorithm is 5.73cm, and compared with the KLD-MCL algorithm, the positioning accuracy of the DM-MC algorithm is improved by 35.6%. Therefore, after the positioning information of the DM landmarks and the odometer information are fused, the accumulated error is corrected by the positioning algorithm, and the fixed particle number is increased after the robot moves to each DM landmark, so that the phenomenon of positioning distortion caused by particle loss is avoided, and the advantages of the DM-MCL positioning algorithm are further verified.

TABLE 2 two Algorithms positioning error comparison

The accuracy and the feasibility of the integrated navigation are verified, the navigation accuracy of the laser SLAM navigation robot with the KLD-MCL positioning algorithm is compared with the navigation accuracy of the laser SLAM integrated navigation robot with the DM-MCL positioning algorithm, the measured values of the tracks of the two navigation modes are respectively recorded in the experimental process, and the analysis of a navigation path diagram shows that: FIG. 5(a) shows that the laser SLAM navigation robot with KLD-MCL positioning algorithm has obvious positioning error, can not reach the requirement of accurate positioning, and increases with the accumulated error of the moving mileage; and (b) the path of the combined navigation robot approaches the target planning path of fig. 5, so that the precision is obviously improved.

The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.