TWI795306B - Localization failure detection system for autonomous mobile robots using deep learning based 6d pose regression - Google Patents

Abstract

Disclosed is a localization failure detection system for autonomous mobile robots using deep learning based 6d pose regression comprising: a visual image input device, a map coordinate regression device, and a localization failure detection device, wherein the localization failure detection system detects a localization failure of a primary localization device of the autonomous mobile robot based on the deviation over time between a localization result of the map coordinate regression device and a localization result of the primary localization device.

Description

Positioning failure detection system for autonomous mobile robot based on deep learning six-dimensional attitude regression

The present invention relates to an autonomous mobile robot, in particular to a positioning failure detection system for an autonomous mobile robot based on deep learning six-dimensional attitude regression.

In the current epidemic situation, in order to reduce the contact and gathering between people, the demand for automated food delivery and delivery services is increasing. The introduction of mobile robots is one of the solutions. Mobile robots can be roughly divided into two types: Automated Guided Vehicle (AGV) and Autonomous Mobile Robot (AMR). The difference between the two is that the former requires construction in the operating field, adding magnetic strips or tracks to allow the AGV to operate, while the latter only needs sensors to build a map in the field in advance to navigate and avoid obstacles in the map. Since most of the delivery locations are ordinary homes and outdoors, it is impossible to carry out pre-construction, so AMR is relatively flexible and flexible, making it more vigorous in recent years.

In the system of autonomous mobile robots, the positioning system can be said to be particularly important. If the positioning system fails, it means that the positioning system cannot find the AMR position on the map, or the error of the positioning result is greater than the positioning error defined by the system. When the positioning system fails, it often causes the path planning system to fail to operate normally, making the AMR stop in place and even crash into obstacles.

Therefore, the purpose of the present invention is to provide a positioning failure detection system for an autonomous mobile robot based on deep learning six-dimensional attitude regression, so as to solve the problem of positioning failure of an autonomous mobile robot.

The technical means adopted by the present invention to solve the problems of the prior art is to provide a positioning failure detection system for an autonomous mobile robot based on deep learning six-dimensional attitude regression, which is used to detect the location of a main positioning device of the autonomous mobile robot Positioning failure, the positioning failure detection system includes: a visual image input device, arranged on the autonomous mobile robot, the visual image input device is configured to obtain a current visual image of the autonomous mobile robot for a predetermined positioning environment Image information; a map coordinate regression device, connected to the visual image input device, the map coordinate regression device uses the trained map coordinate regression neural network to perform feature extraction processing on the current visual image information, and based on The feature map information obtained by the feature extraction process is subjected to attitude estimation processing to obtain a current six-dimensional attitude estimation information of the autonomous mobile robot, wherein the trained map coordinate regression neural network is regressed at the map coordinates The current visual image information input module of the neural network and a visual odometry neural network are in the state of sharing weights, and the map coordinate regression neural network and the visual odometry neural network are trained simultaneously. , the trained map coordinate regression neural network includes a feature extractor for performing the feature extraction process and a pose regressor for performing the pose estimation process, the feature extractor includes a geometric constraint module, the geometric constraint module is configured to add the feature map information of the previous moment obtained by the feature extraction process of the previous moment to input during the feature extraction process of the feature extractor at the moment into the feature extractor; and a positioning failure detection device connected to the map coordinate regression device and the main positioning device to obtain the six-dimensional attitude estimation information estimated by the map coordinate regression device and the A six-dimensional attitude positioning information obtained by the main positioning device for positioning the autonomous mobile robot. A deviation change function value is obtained, and when the deviation change function value exceeds a preset failure judgment threshold, it is judged that the main positioning device of the autonomous mobile robot is currently a positioning failure.

In one embodiment of the present invention, a positioning failure detection system is provided, wherein the visual image input device is a camera, and the current visual image information is a single RGB image.

In an embodiment of the present invention, a localization failure detection system is provided, wherein the feature extractor is a deep separable convolutional neural network.

In an embodiment of the present invention, a positioning failure detection system is provided, wherein the feature extractor includes a MobileNetV2 module.

In an embodiment of the present invention, a positioning failure detection system is provided, wherein the feature extractor includes a global average pooling layer.

An embodiment of the present invention provides a positioning failure detection system, wherein the attitude regressor includes a fully connected layer.

In an embodiment of the present invention, a positioning failure detection system is provided, wherein the positioning failure detection device includes a long-term variation calculation unit and a short-term variation calculation unit, and the long-term variation calculation unit is configured to calculate the The deviation between the six-dimensional attitude estimation information and the six-dimensional attitude positioning information is a long-term variation function value in a long period of time, and the short-term variation calculation unit is configured to calculate the six-dimensional attitude estimation information and the six-dimensional attitude The deviation between the positioning information is a short-term variation function value in a short time, and the positioning failure detection device is configured so that the difference between the long-term variation function value and the short-term variation function value exceeds the failure judgment threshold, it is judged that the main positioning device of the autonomous mobile robot is currently in failure of positioning.

In one embodiment of the present invention, a positioning failure detection system is provided, wherein the positioning failure detection device includes a positioning result output unit connected to a control system of the autonomous mobile robot, and the positioning result output unit is configured And when the main positioning device is judged to be a current positioning failure, the six-dimensional attitude estimation information is used to replace the six-dimensional attitude positioning information, and output to the control system as a positioning result information.

In one embodiment of the present invention, a positioning failure detection system is provided, wherein the positioning failure detection device includes an initial positioning giving unit connected to the main positioning device of the autonomous mobile robot, and the initial positioning giving unit is passed configured to provide the primary positioning device with the six-dimensional attitude estimation information as an initial attitude positioning information for the primary positioning device to use the initial attitude positioning information for subsequent positioning when the primary positioning device is judged to be a current positioning failure The information is used to locate the autonomous mobile robot to obtain the six-dimensional attitude positioning information.

Through the technical means adopted in the present invention, the positioning failure detection system can effectively detect the positioning failure of the main positioning device of the autonomous mobile robot. Furthermore, in a preferred embodiment, when the positioning failure detection system of the present invention is abnormal in the positioning of the autonomous mobile robot, it can use the six-dimensional posture estimation information obtained by posture estimation as the positioning result information, and output it to the autonomous mobile robot. The control system of the robot to maintain the correct functioning of the autonomous mobile robot. On the other hand, the six-dimensional attitude estimation information obtained by the attitude estimation can also be used as the initial attitude positioning information, and given to the main positioning device of the autonomous mobile robot, so that the main positioning device can be repositioned. In addition, in a preferred embodiment, the positioning failure detection system of the present invention estimates whether there is a current positioning failure by observing long-term and short-term changes in the deviation, so as to effectively reduce the possibility of misjudgment.

Embodiments of the present invention will be described below based on FIGS. 1 to 6 . This description is not intended to limit the implementation of the present invention, but is one of the examples of the present invention.

As shown in FIG. 1 , a positioning failure detection system 100 for an autonomous mobile robot based on deep learning six-dimensional attitude regression according to an embodiment of the present invention is used to detect a main component of the autonomous mobile robot (not shown). The positioning failure of the positioning device A, the positioning failure detection system 100 includes: a visual image input device 1 , a map coordinate return device 2 , and a positioning failure detection device 3 .

As shown in Figure 1, in the positioning failure detection system 100 of the present invention, the visual image input device 1 is arranged on the autonomous mobile robot, and the visual image input device 1 is configured to obtain the predetermined position of the autonomous mobile robot. A current visual image information I1 of a positioning environment.

Specifically, the visual image input device 1 is installed on the autonomous mobile robot, and the obtained current visual image information I1 refers to the image information of the positioning environment seen from the perspective of the autonomous mobile robot. Therefore, the autonomous mobile robot does not need to be included in the current visual image information I1. In this embodiment, the visual image input device 1 is a camera, especially a monocular camera, and the current visual image information I1 is a single RGB image, that is, a single RGB bitmap .

As shown in Figure 1 and Figure 2, in the positioning failure detection system 100 of this embodiment, the map coordinate regression device 2 is connected to the visual image input device 1, and the map coordinate regression device 2 is trained The map coordinate regression neural network 20 is used to perform feature extraction processing on the current visual image information I1, and perform pose estimation processing based on the feature map information M1 obtained by the feature extraction processing to obtain the current state of the autonomous mobile robot. A six-dimensional attitude estimation information I2 of .

Specifically, the map coordinate regression method of the map coordinate regression device 2 learns the relationship between the scene and the world coordinates, and only needs to rely on the current visual image information I1 of a single RGB image. The six-dimensional pose of the camera in world coordinates can be output. The trained map coordinate regression neural network 20 used by the map coordinate regression device 2 mainly includes two parts: a feature extractor 21 for performing the feature extraction process, and a feature extractor 21 for performing the attitude estimation A pose regressor 22 is processed. The feature extraction process performed by the feature extractor 21 is mainly responsible for extracting the features in the image (the current visual image information I1). For example, when the positioning environment is the internal environment of a building, the extracted features can be Features such as wall corners, corridor disappearing lines, doors and windows, etc., the distribution of these features in the image can be used to judge the pose in the subsequent stage of the pose estimation process. The attitude estimation process performed by the attitude regressor 22 is mainly to convert the features extracted by the feature extraction process into values conforming to the world coordinates, and finally return to the six-dimensional attitude (6D POSE) to obtain the six-dimensional attitude estimation information I2.

The feature extractor 21 can usually use a convolutional neural network (Convolutional Neural Network; CNN) or a recurrent neural network (Recurrent Neural Network; RNN), and the attitude regressor 22 can usually be a fully connected neural network (Fully Connected Neural Network; FNN), but the present invention is not limited thereto. In this embodiment, the feature extractor 21 is a Depthwise Separable Convolution Neural Network (Depthwise Separable Convolution Neural Network), and the attitude regressor 22 is a neural network composed of multiple fully connected layers. network. The feature extractor 21 includes MobileNetV2 modules 211 , 212 , 213 and a global average pooling layer 214 , and especially includes a geometric constraint module 23 . The MobileNetV2 modules 211, 212, and 213 are the second-generation mobile device version of the computer vision neural network launched by "Google", which achieves the purpose of compressing the model through depth-separable convolution to reduce parameters and improve In addition to the computing speed, it also has the binomial characteristics of linear bottlenecks between layers and shortcut connections between bottlenecks. The architecture and technical details of the MobileNetV2 modules 211 , 212 , 213 are easily known to those skilled in the technical field of the present invention, so no further details are given here. The global average pooling layer (Global Average Pooling Layer) 214 is used as the last layer of the feature extractor 21, in order to avoid the feature map after the feature extraction of the MobileNetV2 module 211, 212, 213 due to the excessive quantity. Subsequent problems in the attitude estimation processing.

The Geometric Constrained Module (GCM) 23 is configured so that when the feature extractor 21 performs the feature extraction process at the moment, the feature extraction process at the previous moment is obtained by the feature extraction process at the previous moment. The feature map information M0 is added to the feature extractor 21 . The reason for setting the geometric restriction module 23 is that the posture change mode of the autonomous mobile robot in progress is limited by the mechanism, which affects the corresponding relationship between the image position at the moment and the previous moment. If the map coordinates cannot be returned to the neurological The network 20 effectively learns such geometric constraints during training, and the obtained six-dimensional pose estimation information I2 will have insufficient accuracy. Although the posture change of the autonomous mobile robot can be learned by using two images, it will encounter the problem that the scale of the scene cannot be correctly estimated and the initial position is required for positioning. Therefore, the present invention connects the feature map information M0 at the previous moment with the feature map information M1 at the present moment through the geometric restriction module 23, performs a point-wise convolution, and then adds a nonlinear activation function to each pixel, Filter out unwanted information. As shown in Figure 2, the feature map information M1 obtained by the MobileNetV2 modules 211 and 212, and the feature map information M0 obtained by the same module as the previous moment, enter the convolution layer with a convolution kernel of 1×1 231, and then enter the feature extraction of the MobileNetV2 module 213 in the next stage after passing through the linear rectification function 232 (Rectified Linear Unit; ReLU).

As shown in Fig. 3, in the map coordinate regression device 2, the trained map coordinate regression neural network 20 is in the current state of both the map coordinate regression neural network 20 and a visual odometer neural network VO. The visual image information input module is obtained by simultaneously training the map coordinate regression neural network 20 and the visual odometer neural network VO under the condition of sharing weights. The visual odometry neural network VO refers to a neural network based on visual odometry. Visual odometry (Visual Odometry) is a way of positioning a robot by image analysis. Its architecture and technical details are the technology of the present invention. Those with common knowledge in the field can easily know it, so no further details will be given here. The present invention trains the map coordinate regression neural network 20 in an auxiliary learning manner. The auxiliary learning method is to simultaneously train the visual odometry neural network VO and the map coordinate regression neural network 20 to make the characteristics of the two Extractors can learn each other's propensity to extract features. The trained map coordinate regression neural network 20 can learn the geometric constraints between each image through the visual odometry neural network VO to increase the prediction accuracy. The trained map coordinate regression neural network 20 after assisted learning will be separated from the visual odometry neural network VO when applied (as shown in Figure 2), so it can be used without increasing the amount of parameters and model architecture , to produce better results. In the embodiment shown in FIG. 3 , the visual odometry neural network VO has a structure similar to a twin neural network, and is used to extract the current image information T1 and the previous image information T0 respectively. The current visual image information input module of the map coordinate regression neural network 20 (that is, the MobileNetV2 module 211) and the current visual image information input module of the visual odometry neural network VO (that is, input at the moment Modules of image information T1) share weights (indicated by dotted line links in Figure 3). In this way, whenever the parameters are updated, the shared weight will consider the error between the visual odometry part and the map coordinate regression part for gradient descent, so it can learn to extract the required features on both sides. It should be noted that the present invention is not limited thereto, and the map coordinate regression neural network 20 and the visual odometry neural network VO may also have other architectures than the embodiment shown in FIG. 3 .

As shown in Figures 1 to 6, the positioning failure detection device 3 is connected to the map coordinate returning device 2 and the main positioning device A to obtain the six-dimensional attitude estimated by the map coordinate returning device 2 The estimated information I2 and a six-dimensional attitude positioning information I _A obtained by the main positioning device A for positioning the autonomous mobile robot. The positioning failure detection device 3 is configured to obtain a deviation variation function value F _B according to the variation over time of the deviation B between the six-dimensional attitude estimation information I2 and the six-dimensional attitude positioning information I _A , and when the deviation variation function value F _B exceeds a preset failure judgment threshold Th, it is judged that the main positioning device A of the autonomous mobile robot is currently in a positioning failure. In this embodiment, the main positioning device A is a map-based positioning device, which obtains the six-dimensional attitude positioning information I _A by sensing the lidar of the environment and the pre-built point cloud map for positioning. Of course, the main positioning device A can also be a device for positioning in other ways, as long as the six-dimensional posture positioning information I A obtained by the main positioning device A is consistent with the six-dimensional posture positioning information I _A estimated by the map coordinate regression device 2 It only needs to correspond to the three-dimensional attitude estimation information I2, so that it can be used by the positioning failure detection device 3 to perform the above-mentioned relative calculation and judgment of the positioning failure.

Preferably, as shown in Figure 4, in this embodiment, the positioning failure detection device 3 includes a long-term variation calculation unit 31 and a short-term variation calculation unit 32, and the long-term variation calculation unit 31 is configured To calculate the long-term variation function value F1 of the deviation B between the six-dimensional attitude estimation information I2 and the six-dimensional attitude positioning information _IA , the short-term variation calculation unit 32 is configured to calculate the six-dimensional A short-term variation function value F2 of the deviation B between the attitude estimation information I2 and the six-dimensional attitude positioning information I _A in a short period of time, and a judgment unit 33 of the positioning failure detection device 3 is configured to be in the When the difference between the long-term variation function value F1 and the short-term variation function value F2 (that is, the deviation variation function value F _B ) exceeds the failure judgment threshold Th, it is judged that the main positioning device A of the autonomous mobile robot is currently Invalid for positioning.

Since if the deviation B between the six-dimensional attitude estimation information I2 and the six-dimensional attitude positioning information I _A is simply used as a judgment, as long as it exceeds a specific threshold, it will be judged as a positioning failure. In this case, the map must be determined. Each estimation result of the coordinate regression device 2 (the six-dimensional attitude estimation information I2 ) is extremely accurate, otherwise it is easy to cause a high rate of false positives. Therefore, in this embodiment, by emphasizing the long-term and short-term changes of the deviation, by accumulating the deviation and comparing the long-term and short-term effects of the deviation accumulation, it is judged whether there is a positioning failure. Specifically, both the long-term variation function value F1 and the short-term variation function value F2 represent a cost obtained by a set calculation method. In this embodiment, by calculating the current deviation B The difference from the previous cost is accumulated to obtain the current cost. Please refer to FIG. 5 and FIG. 6, which show the variation relationship between the deviation B and the long-term variation function value F1 and the short-term variation function value F2. As shown in the figure, when the deviation B changes small, the cost will fluctuate in a small range, but when the deviation B becomes larger, the cost can also rise accordingly. This approach can avoid the occurrence of misjudgment caused by a certain excessive deviation B. When the robot kidnaps or the optimization failure of the main positioning device A occurs, the deviation B will increase, and the cost will gradually increase. The short-term change amount function value F2 shows the cost of being sensitive to the short-term change amount of the deviation B, and can respond to the change of the deviation B more quickly. Compared with the short-term variation function value F2, the long-term variation function value F1 needs a longer time to accumulate cost, which reduces the chance of misjudgment. When the difference between the long-term variation function value F1 and the short-term variation function value F2 (the deviation variation function value F _B ) exceeds the failure judging threshold Th, it is judged as a positioning failure. As can be seen from Figure 6, even if the deviation B suddenly increases at the fifth data point on the horizontal axis, the deviation variation function value F _B will not increase excessively, avoiding the estimation caused by the map coordinate regression device 2 Misjudgments caused by mistakes. When the positioning abnormality occurs at the 11th data point on the horizontal axis, the deviation variation function value F _B shows a gradual increase, and the positioning failure can be judged by the failure judgment threshold Th.

Preferably, as shown in Figure 4, in this embodiment, the positioning failure detection device 3 includes a positioning result output unit 34, which is connected to a control system C of the autonomous mobile robot, and the positioning result output unit 34 It is configured to replace the six-dimensional attitude positioning information I A with the six-dimensional attitude estimation information _I2 when the main positioning device A is judged to be a current positioning failure, and output it to the control system C as a positioning result information I3 , to maintain the correct operation of the autonomous mobile robot.

Preferably, as shown in Figure 4, in this embodiment, the positioning failure detection device 3 includes an initial positioning unit 35 connected to the main positioning device A of the autonomous mobile robot, the initial positioning unit 35 is configured so that when the main positioning device A is judged to be a current positioning failure, the six-dimensional attitude estimation information I2 is used as an initial attitude positioning information I0 and given to the main positioning device A for the main positioning The device A subsequently uses the initial attitude positioning information I0 to position the autonomous mobile robot to obtain the six-dimensional attitude positioning information I _A .

The above descriptions and descriptions are only descriptions of the preferred embodiments of the present invention. Those who have common knowledge of this technology may make other modifications according to the scope of the patent application defined below and the above descriptions, but these modifications should still be It is for the inventive spirit of the present invention and within the scope of rights of the present invention.

100: Positioning failure detection system 1: Visual image input device 2: Map coordinate regression device 20: Map coordinate regression neural network 21: Feature extractor 211: MobileNetV2 module 212: MobileNetV2 module 213: MobileNetV2 module 214 : Global average pooling layer 22: Attitude regressor 23: Geometric restriction module 231: Convolution layer 232: Linear rectification function 3: Positioning failure detection device 31: Long-term variation calculation unit 32: Short-term variation calculation unit 33: Judging unit 34: positioning result output unit 35: initial positioning giving unit A: main positioning device B: deviation C: control system F1: long-term variation function value F2: short-term variation function value F _B : deviation variation function value I0: Initial attitude positioning information I1: current visual image information I2: six-dimensional attitude estimation information I3: positioning result information I _A : six-dimensional attitude positioning information M0: feature map information M1: feature map information VO: visual odometry neural network

[Fig. 1] is a schematic block diagram showing a positioning failure detection system for an autonomous mobile robot based on deep learning six-dimensional attitude regression according to an embodiment of the present invention; [Fig. 2] is a schematic diagram showing the trained map coordinate regression neural network of the map coordinate regression device in the positioning failure detection system according to an embodiment of the present invention; [Fig. 3] is a schematic diagram showing a map coordinate regression neural network and a visual odometry neural network according to an embodiment of the present invention; [Fig. 4] is a schematic block diagram showing a location failure detection device in a location failure detection system according to an embodiment of the present invention; [Fig. 5] shows the deviation between the six-dimensional attitude estimation information and the six-dimensional attitude positioning information obtained by the positioning failure detection device of the embodiment of the present invention, the long-term variation function value, and the short-term variation function value schematic diagram; [Figure 6] shows the deviation between the six-dimensional attitude estimation information and the six-dimensional attitude positioning information obtained by the positioning failure detection device according to the embodiment of the present invention, and the difference between the long-term variation function value and the short-term variation function value The schematic diagram of the difference between and the failure judgment threshold.

100: Positioning failure detection system

1: Visual image input device

2: Map coordinate return device

20: Map Coordinate Regression Neural Network

3: Positioning failure detection device

A: Main positioning device

I0: Initial attitude positioning information

I1: Current visual image information

I2: Six-dimensional attitude estimation information

I3: Positioning result information

I _A : Six-dimensional attitude positioning information

Claims (9)

A positioning failure detection system for an autonomous mobile robot based on deep learning six-dimensional attitude regression is used to detect a positioning failure of a main positioning device of the autonomous mobile robot. The positioning failure detection system includes: A visual image input device is arranged on the autonomous mobile robot, and the visual image input device is configured to obtain a current visual image information of the autonomous mobile robot for a predetermined positioning environment; A map coordinate regression device, connected to the visual image input device, the map coordinate regression device uses the trained map coordinate regression neural network to perform feature extraction processing on the current visual image information, and based on the feature extraction Taking the processed feature map information and performing attitude estimation processing to obtain the current six-dimensional attitude estimation information of the autonomous mobile robot, wherein the trained map coordinate regression neural network is based on the map coordinate regression neural network The current visual image information input module and a visual odometry neural network are obtained by simultaneously training the map coordinate regression neural network and the visual odometry neural network in the state of sharing weights. After training The map coordinate regression neural network includes a feature extractor for performing the feature extraction process and a pose regressor for performing the pose estimation process, the feature extractor includes a geometric constraint module, The geometric constraint module is configured to add the feature map information of the previous moment obtained by the feature extraction process of the previous moment to the feature when the feature extractor is performing the feature extraction process of the current moment in the fetcher; and A positioning failure detection device, connected to the map coordinate returning device and the main positioning device, to obtain the six-dimensional attitude estimation information estimated by the map coordinate returning device and the main positioning device to locate the autonomous mobile robot For the obtained six-dimensional attitude positioning information, the positioning failure detection device is configured to obtain a deviation change according to the variation over time of the deviation between the six-dimensional attitude estimation information and the six-dimensional attitude positioning information and when the deviation variation function value exceeds a preset failure judgment threshold, it is judged that the main positioning device of the autonomous mobile robot is currently a positioning failure. The positioning failure detection system as described in Claim 1, wherein the visual image input device is a camera, and the current visual image information is a single RGB image. The positioning failure detection system as claimed in claim 1, wherein the feature extractor is a deep separable convolutional neural network. The positioning failure detection system as described in claim 3, wherein the feature extractor includes a MobileNetV2 module. The location failure detection system as claimed in claim 3, wherein the feature extractor includes a global average pooling layer. The positioning failure detection system according to claim 1, wherein the attitude regressor includes a fully connected layer. The positioning failure detection system as described in claim 1, wherein the positioning failure detection device includes a long-term variation calculation unit and a short-term variation calculation unit, and the long-term variation calculation unit is configured to calculate the six-dimensional attitude estimation The deviation between the measurement information and the six-dimensional attitude positioning information is a long-term variation function value in a long period of time, and the short-term variation calculation unit is configured to calculate the difference between the six-dimensional attitude estimation information and the six-dimensional attitude positioning information The deviation of a short-term variation function value in a short period of time, and the positioning failure detection device is configured so that when the difference between the long-term variation function value and the short-term variation function value exceeds the failure judgment threshold, it is judged The primary positioning device of the autonomous mobile robot is currently a positioning failure. The positioning failure detection system as described in claim 1 or 7, wherein the positioning failure detection device includes a positioning result output unit connected to a control system of the autonomous mobile robot, and the positioning result output unit is configured to When the main positioning device is judged that the current positioning fails, the six-dimensional attitude estimation information is used to replace the six-dimensional attitude positioning information, and output to the control system as a positioning result information. The positioning failure detection system as described in claim 1 or 7, wherein the positioning failure detection device includes an initial positioning giving unit connected to the main positioning device of the autonomous mobile robot, and the initial positioning giving unit is configured to When the main positioning device is judged to be a current positioning failure, the six-dimensional attitude estimation information is given to the main positioning device as an initial attitude positioning information, so that the main positioning device can subsequently use the initial attitude positioning information to pair The autonomous mobile robot performs positioning to obtain the six-dimensional attitude positioning information.

Images