Autonomous multi-machine cooperative SLAM method and apparatus, and electronic device

Abstract

The invention provides an autonomous multi-machine cooperative SLAM (Simultaneous Localization and Mapping) method and device and electronic equipment, and the method comprises the steps: receiving detection data from different sensors of a robot, and carrying out the preprocessing of the detection data; carrying out undistorted fusion on the preprocessed detection data to obtain robot state estimation; generating an obstacle avoidance navigation path for each robot in combination with the robot state estimation; integrating the obstacle avoidance navigation paths of the robots to form a complete environment map; and in combination with the environment map, allocating an exploration task to each robot, and carrying out environment modeling on a target area. According to the invention, the security and task efficiency of multi-machine collaborative operation can be improved, the practicability and applicability are high, and the problem of poor detection effect in a complex environment in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of inspection robot control, and in particular to a self-driven multi-robot cooperative SLAM method and device and electronic equipment. BACKGROUND

[0002] With the rapid development of inspection robot technology, inspection robots have been widely used in military, civilian and scientific research fields. In the field of application of inspection robots, the simultaneous localization and mapping (SLAM) algorithm is an important research direction. The goal of SLAM is to estimate the position and orientation of the intelligent agent simultaneously, and to create an environment map according to data from various sensors such as cameras, lidar and sonar.

[0003] At present, the academic and industrial circles have made great progress in single-robot SLAM, but the traditional single-robot SLAM algorithm has the problems of low precision and poor robustness. With the advantages brought by information exchange and sharing between different intelligent agents, the multi-robot cooperative SLAM algorithm has higher robustness, accuracy and efficiency. At present, the research on multi-robot cooperative SLAM is still relatively less, including feature-based methods, extended methods based on Kalman filter and graph-based methods. These methods integrate the measurement information of multiple robots to improve the positioning and mapping accuracy of the system, but mainly aim at some special application scenarios such as underwater robots and aircrafts. Due to the need to process a large amount of data, the computational complexity is very high. In addition, in multi-robot cooperative SLAM, in order to achieve an accurate and efficient solution, several challenges need to be solved, including coordination between inspection robots, data communication, sensor data fusion and efficient autonomous search.

[0004] There is no effective solution to the problem of poor detection effect in complex environments in the prior art. SUMMARY

[0005] The present application provides a self-driven multi-robot cooperative SLAM method, device and electronic equipment to solve the defect of poor detection effect in complex environments in the prior art.

[0006] In a first aspect, the present application provides a self-driven multi-robot cooperative SLAM method, comprising: receiving detection data from different sensors of the robot, and preprocessing the detection data; non-distortion fusion of the preprocessed detection data to obtain robot state estimation; Based on the robot state estimation, an obstacle avoidance navigation path is generated for each robot; The obstacle avoidance navigation paths of each robot are integrated to form a complete environmental map; Based on the environmental map, each robot is assigned an exploration task, and environmental modeling is performed on the target area.

[0007] According to an autonomous multi-machine collaborative SLAM method provided by the present invention, detection data from different sensors of a robot is received, and the detection data is preprocessed, including: Receive detection data from the robot's inertial odometry and binocular camera; The detection data is then subjected to soft synchronization alignment in time and space. The weight values ​​of each of the detection data are determined by the breadth weighting algorithm, and the aligned binocular depth map is obtained. The pixel layer is determined, and the binocular depth map is converted into two-dimensional laser point cloud data.

[0008] According to the autonomous multi-machine collaborative SLAM method provided by the present invention, the preprocessed detection data is fused without distortion to obtain robot state estimation, including: Based on the robot's state at the previous moment, predict the robot's state at the current moment; the robot state includes position information and posture information. The detection data from different sensors are fused together to correct the predicted robot state, thus obtaining the robot state estimate.

[0009] According to the autonomous multi-machine cooperative SLAM method provided by the present invention, in conjunction with the robot state estimation, an obstacle avoidance navigation path is generated for each robot, including: Based on hybrid The path planning algorithm and the B-spline soft-constraint path optimization algorithm generate obstacle avoidance navigation paths for the robot.

[0010] The present invention provides an autonomous multi-machine collaborative SLAM method based on hybrid... The path planning algorithm and the B-spline soft-constraint path optimization algorithm generate an obstacle avoidance navigation path for the robot, including: A grid map is generated based on the robot's position information, posture information, and binocular depth map. Combining the aforementioned raster map, a hybrid approach is adopted. The path planning algorithm and the soft-constraint path optimization algorithm of B-spline are used to determine the obstacle avoidance navigation path of the robot in the target area.

[0011] The autonomous multi-robot cooperative SLAM method provided by the application comprises the following steps of: after generating an obstacle avoidance navigation path for each robot, performing path smoothing processing on the obstacle avoidance navigation path.

[0012] The autonomous multi-robot cooperative SLAM method provided by the application comprises the following steps of: after generating an obstacle avoidance navigation path for each robot, performing path smoothing processing on the obstacle avoidance navigation path. The key frames are acquired through distributed communication of a robot operating system, and the local map of the grid map component is combined; The repeated map points are deleted, the co-visibility relationship of the deleted map points is reserved, and BA is performed on the key frames and the map points in the merging process; The key frames with the co-visibility relationship are repaired, and the pose of other key frames in the local map is optimized to obtain the environment map.

[0013] The autonomous multi-robot cooperative SLAM method provided by the application comprises the following steps of: after generating an obstacle avoidance navigation path for each robot, performing path smoothing processing on the obstacle avoidance navigation path. The exploration boundary of the target area is read by the robot; The task grid is divided based on the exploration boundary to obtain a plurality of exploration tasks; The exploration tasks are distributed to each robot, and the exploration tasks are executed on the target area to model the environment.

[0014] In a second aspect, the application further provides an autonomous multi-robot cooperative SLAM device, comprising: A processing module is configured to receive detection data from different sensors of a robot and pre-process the detection data; A fusion module is configured to perform lossless fusion on the pre-processed detection data to obtain robot state estimation; A planning module is configured to generate an obstacle avoidance navigation path for each robot in combination with the robot state estimation; An integration module is configured to integrate the obstacle avoidance navigation paths of each robot to form a complete environment map; An exploration module is configured to distribute exploration tasks to each robot in combination with the environment map and model the environment of a target area.

[0015] In a third aspect, the application further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the autonomous multi-robot cooperative SLAM method of the first aspect when executing the program.

[0016] In a fourth aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the autonomous multi-robot cooperative SLAM method according to the first aspect.

[0017] In a fifth aspect, the present application also provides a computer program product comprising a computer program which, when executed by a processor, implements the autonomous multi-robot cooperative SLAM method according to the first aspect.

[0018] Compared with the prior art, the present application has the following beneficial effects: The autonomous multi-robot cooperative SLAM method provided by the present application can generate corresponding obstacle avoidance navigation paths for each robot according to the fusion data of multiple sensors, and the path generation result is more detailed and can better adapt to more complex environments. Then, the multiple robots are integrated and task allocation is performed to coordinate synchronous exploration, which can improve the safety and task efficiency of multi-robot cooperative operation, and has strong practicability and applicability, and solves the problem of poor detection effect in complex environments in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0020] Figure 1 is a flow chart of the autonomous multi-robot cooperative SLAM method provided by the present application; Figure 2 is a loose coupling relationship diagram of the multi-sensor odometer in the embodiment of the present application; Figure 3 is a process schematic diagram of single-machine path planning in the embodiment of the present application; Figure 4 is a multi-robot cooperative exploration configuration layout schematic diagram in the embodiment of the present application; Figure 5 is a structural block diagram of the autonomous multi-robot cooperative SLAM device provided by the present application; Figure 6 is a structural block diagram of the autonomous multi-robot cooperative SLAM system provided by the present application; Figure 7 is a structural schematic diagram of the electronic device provided by the present application. DETAILED DESCRIPTION

[0021] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be clearly and completely described below in combination with the drawings in the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0022] The present application provides a self-autonomous multi-robot cooperative SLAM method, Figure 1 The present application provides a flow chart of the self-autonomous multi-robot cooperative SLAM method, as shown in Figure 1 The method comprises the following steps: Step S101, receiving detection data from different sensors of the robot, and pre-processing the detection data; Step S102, performing distortion-free fusion on the pre-processed detection data to obtain robot state estimation; Step S103, combining the robot state estimation to generate an obstacle avoidance navigation path for each robot; Step S104, integrating the obstacle avoidance navigation path of each robot to form a complete environment map; Step S105, combining the environment map to assign an exploration task to each robot, and modeling the target area.

[0023] In the method, firstly, detection data from different sensors of the robot is received, and the acquired detection data is preprocessed. Then, the preprocessed detection data is fused without distortion to obtain robot state estimation, thereby improving the accuracy of positioning. In combination with the robot state estimation, obstacle avoidance planning is performed for the robot to generate an obstacle avoidance navigation path. Then, based on an Oriented FAST and Rotated BRIEF Simultaneous Localization and Mapping-Atlas Map Fusion Algorithm (ORB-SLAM Atlas), all the arm chapter navigation paths of the robots are integrated to form a complete environment map. Finally, based on the environment map, exploration tasks are assigned to each robot, the motions of the robots are coordinated, and the target area is explored to realize environment modeling. In the above process, according to the fused data of multiple sensors, obstacle avoidance planning is performed for each robot to generate a corresponding obstacle avoidance navigation path, the path generation result is more detailed, and the path generation result can better adapt to a more complex environment. Then, integration and task assignment are performed to coordinate synchronous exploration of multiple robots, which can improve the safety and task efficiency of multi-robot cooperative operation, has strong practicality and applicability, and solves the problem of poor detection effect in a complex environment in the prior art.

[0024] In some embodiments, step S101 of receiving detection data from different sensors of the robot and preprocessing the detection data includes: receiving detection data from an inertial odometer and a binocular camera of the robot; performing time and space soft synchronization alignment processing on the detection data; determining respective weight values of the detection data by a luminosity weight algorithm, and acquiring an aligned binocular depth map; determining a pixel layer, and converting the binocular depth map into two-dimensional laser point cloud data.

[0025] For example, firstly, detection data of an inertial measurement unit (IMU) and a binocular camera is received, and then time and space soft synchronization is performed on the two kinds of detection data. Weight values of the aligned detection data are calculated by a luminosity weight algorithm for subsequent use of a visual inertial odometer (VIO) module. Meanwhile, an aligned binocular depth map is acquired, a suitable pixel layer is selected, and the binocular depth map is converted into two-dimensional laser point cloud data.

[0026] In the embodiment, an adaptive photometric weight algorithm is designed to maintain an accurate pose output in the case of visual odometry failure. Meanwhile, in order to save the computing resources of the inspection robot for long-term stable operation, the binocular depth map combined with the infrared structured light is converted into two-dimensional laser point cloud data, which is adapted to the laser obstacle avoidance algorithm.

[0027] In some embodiments, the preprocessed detection data is fused without distortion to obtain a robot state estimation, including: combining the robot state of the robot at the previous time, predicting the robot state of the robot at the current time; the robot state includes position information and attitude information; the detection data of different sensors is fused, and the predicted robot state is corrected to obtain the robot state estimation.

[0028] For example, the Extended Kalman Filter (EKF) is used to fuse the odometer data of multiple sensors. Let the robot state be , the state transition model be , and the observation model be , wherein includes the position information and the attitude information of the robot. The Extended Kalman Filter includes a prediction step and an update step, and the prediction step is as follows:

[0029] wherein, represents the predicted state, represents the posterior state estimation at time k 1, represents the control input, represents the predicted covariance matrix, represents the state transition model at time , , represents the posterior estimation covariance matrix at time k 1, represents the transpose matrix of the Jacobian matrix , represents the process noise covariance matrix.

[0030] The update step is as follows:

[0031] wherein, represents the innovation, the difference between the actual observation value and the predicted observation value , i.e., the deviation between the sensor measurement and the model prediction, represents an observation, represents an observation model, represents a predicted state, S k represents an innovation covariance matrix, H k represents a Jacobian matrix of an observation matrix, represents a predicted covariance matrix, superscript T represents a transpose of a matrix, R k represents an observation noise covariance matrix, K k represents a Kalman gain, represents an innovation covariance matrix S k an inverse matrix of, represents a posterior state estimation at k represents represents k represents a posterior covariance matrix at I represents an identity matrix. In this embodiment, a robust multi-odometer loose coupling method is designed for the complexity of the scene, Figure 2 is a multi-sensor odometer loose coupling relationship diagram in the embodiment of the application, as shown in Figure 2 by extending the Kalman filter, visual, infrared and IMU data are fused to obtain more accurate robot state estimation.

[0032] In some embodiments, step S103, in combination with the robot state estimation, generates an obstacle avoidance navigation path for each robot, including: generating an obstacle avoidance navigation path for the robot based on a hybrid path planning algorithm and a soft constraint path optimization algorithm based on B-spline.

[0033] Specifically, an obstacle avoidance navigation path is generated for the robot based on a hybrid path planning algorithm and a soft constraint path optimization algorithm based on B-spline, including: generating a grid map based on robot position information, attitude information and binocular depth map; in combination with the grid map, a hybrid path planning algorithm and a soft constraint path optimization algorithm based on B-spline are used to determine the obstacle avoidance navigation path of the robot in the target area. After generating an obstacle avoidance navigation path for each robot, including: performing path smoothing processing on the obstacle avoidance navigation path.

[0034] Figure 3 is a process schematic diagram of single-machine path planning in the embodiment of the application, as shown in Figure 3As shown, since the inspection robot needs to autonomously complete the exploration modeling task, it needs to have an obstacle avoidance function in a complex environment. Each unmanned inspection robot performs real-time path planning on its own on-board processor to complete the obstacle avoidance exploration task. First, the position information, attitude information and depth map information are received, and a grid map is generated while controlling spatial discretization to better transmit data information. Then, through the hybrid The path planning algorithm and the soft constraint path optimization algorithm of B-spline generate an obstacle avoidance and target navigation path for the robot in a complex environment. Further, the obstacle avoidance and target navigation path can be effectively smoothed to improve the motion performance and safety of the robot.

[0035] In some embodiments, step S104, the obstacle avoidance and target navigation path of each robot is integrated to form a complete environment map, including: obtaining key frames through distributed communication of the Robot Operating System (ROS), and combining with the local map of the grid map component; deleting repeated map points, retaining the co-visibility relationship of the deleted map points, and performing BA on the key frames and map points in the merging process; repairing the key frames with co-visibility relationship, and optimizing the poses of other key frames in the local map to obtain the environment map.

[0036] For example, the key frames and grid map information of the single unmanned aerial vehicle are received through the distributed communication of the Robot Operating System to complete the modeling of the environment map. In order to integrate the local maps of multiple robots, the map fusion algorithm based on ORB-SLAM3 Atlas is adopted. First, the detected frame, the co-visibility frame of the detected frame, the candidate frame and the co-visibility frame of the candidate frame, and the map points corresponding to the above key frames are combined to form a local map. Then, the repeated points are deleted, but the co-visibility relationship of the deleted points is retained. Then, the key frames and map points in the merging process are processed by Bundle Adjustment (BA). In order to ensure consistency, the key frames with co-visibility relationship are repaired, and finally the remaining key frames in the entire map are optimized in pose to obtain the environment map.

[0037] In some embodiments, step S105, in combination with the environment map, the exploration task is assigned to each robot, and the target area is modeled, including: reading the exploration boundary of the robot on the target area; dividing the task grid based on the exploration boundary to obtain a plurality of exploration tasks; assigning the exploration tasks to each robot to execute the exploration tasks on the target area to model the environment.

[0038] In actual application, due to the limited computing power of a single machine, it cannot bear the task allocation work of multiple machines, so the multi-machine task allocation algorithm based on front information is deployed on the server. Figure 4is a schematic diagram of multi-machine cooperative exploration configuration layout in the embodiment of the application, as Figure 4 As shown in the figure, first, the set exploration boundary is read from the configuration file, then the task grid is divided based on the front information, and then the exploration task is allocated to each sub-machine, and the exploration task is executed by the single machine autonomously and cooperatively, and the environment of the target area is modeled.

[0039] The application also provides an autonomous multi-machine cooperative SLAM device, which is described as follows. Figure 5 is a structural block diagram of the autonomous multi-machine cooperative SLAM device provided by the application, as Figure 5 As shown in the figure, the device comprises: The processing module 501 is configured to receive detection data from different sensors of the robot and pre-process the detection data. The fusion module 502 is configured to fuse the pre-processed detection data without distortion to obtain a robot state estimation. The planning module 503 is configured to generate an obstacle avoidance navigation path for each robot in combination with the robot state estimation. The integration module 504 is configured to integrate the obstacle avoidance navigation path of each robot to form a complete environment map. The exploration module 505 is configured to allocate an exploration task to each robot in combination with the environment map and model the environment of the target area.

[0040] In use, firstly, the processing module 501 receives detection data from different sensors of the robot, and pre-processes the obtained detection data. Then, the fusion module 502 performs distortion-free fusion on the pre-processed detection data to obtain robot state estimation, thereby improving the positioning accuracy. The planning module 503 combines the robot state estimation to perform obstacle avoidance planning for the robot, and generates an obstacle avoidance navigation path. Then, the integration module 504 integrates all the arm chapter navigation paths of the robots based on the ORB-SLAM Atlas sub-map fusion algorithm to form a complete environment map. Finally, the exploration module 505 assigns exploration tasks to each robot based on the environment map, coordinates the motion of each robot, and explores the target area to realize environment modeling. In the above process, according to the fusion data of multiple sensors, obstacle avoidance planning is performed for each robot to generate a corresponding obstacle avoidance navigation path, and the path generation result is more detailed, which can well adapt to more complex environments. Then, integration and task allocation are performed to coordinate the synchronous exploration of multiple robots, which can improve the safety and task efficiency of multi-machine cooperative operation, and has strong practicality and applicability, and solves the problem of poor detection effect in complex environments in the prior art.

[0041] The application also provides an autonomous multi-machine cooperative SLAM system, Figure 6 is a structural block diagram of the autonomous multi-machine cooperative SLAM system provided by the application, as Figure 6 shown, the system is composed of a sensor data processing module, a multi-sensor odometer fusion module, a multi-machine data communication module, a single-machine obstacle avoidance planning module, a server map fusion module, a multi-machine task allocation module based on a front information structure algorithm and a multi-machine process module. The above modules are described in detail as follows: The sensor data processing module is responsible for pre-processing data from various sensors. The system designs an adaptive luminosity weight algorithm, which still maintains a relatively accurate pose output in the case of visual odometer failure. At the same time, in order to save the computing resources of the inspection robot for long-term stable operation, the binocular depth map combined with infrared structured light is converted into two-dimensional laser point cloud data, which is adapted to the laser obstacle avoidance algorithm.

[0042] The multi-sensor fusion module is responsible for distortion-free fusion of the processed multi-sensor data. Extended Kalman filter is a commonly used multi-sensor data fusion method. In this module, a robust multi-odometer loosely coupled module is designed according to the complexity of the scene, and EKF is used to fuse the odometer data of multiple sensors.

[0043] The multi-machine data communication module is mainly responsible for multi-machine cooperative communication and data exchange. Considering the possible existence of satellite rejection and other complex indoor environmental conditions, distributed communication between multiple machines is considered, the communication mode is established, and then the data communication message format is established. The ROS multi-machine communication mechanism based on 4G link is designed to transmit key frames, poses, and sparse map data between multiple machines.

[0044] The single-machine path planning module is responsible for receiving position information, attitude information, and depth map information, and is mainly composed of a hybrid path planning algorithm based on The path planning algorithm and B-spline soft constraint path optimization algorithm. Since the inspection robot needs to autonomously complete the exploration modeling task, it needs to have obstacle avoidance function in complex environment. Each inspection robot performs real-time path planning on its own on-board processor to complete the obstacle avoidance exploration task.

[0045] The map fusion module runs on the server and receives key frames and grid map information from single unmanned aerial vehicles through ROS distributed communication to complete environment modeling.

[0046] The multi-machine cooperative exploration module is deployed on the server, which first reads the set exploration boundary from the configuration file, then divides the task grid based on the front information, and then assigns the exploration task to each sub-machine, which is executed autonomously by the single machine. The environment is modeled.

[0047] In summary, when the system is used, the sensor data processing module is responsible for preprocessing the detection data from various sensors, and the multi-machine process module fuses the preprocessed detection data to improve the positioning accuracy. The 4G link between multiple machines realizes data transmission between each machine through the ROS multi-machine strategy. At the same time, the single-machine obstacle avoidance planning module plans an obstacle avoidance path for each machine. Then, based on the ORB-SLAM Atlas sub-map fusion algorithm, the map information of each machine is integrated in the map fusion module to form a complete map. Finally, the multi-machine task allocation module based on the front information structure algorithm coordinates the actions of each machine to improve the detection efficiency of the entire system. The association between modules is reflected in the data flow transmission and processing process, and finally realizes the fusion of multiple sensor data and multi-machine cooperative environment modeling task.

[0048] Figure 7 An example of an entity structure diagram of an electronic device is shown in FIG. 1. Figure 7As shown, the electronic device can include a processor 701, a communications interface 702, a memory 703, and a communications bus 704, wherein the processor 701, the communications interface 702, and the memory 703 complete mutual communication through the communications bus 704. The processor 701 can invoke a logic instruction in the memory 703 to execute an autonomous multi-robot collaborative SLAM method, which includes: receiving detection data from different sensors of the robot, and preprocessing the detection data; fusing the preprocessed detection data without distortion to obtain robot state estimation; combining the robot state estimation to generate an obstacle avoidance navigation path for each robot; integrating the obstacle avoidance navigation path of each robot to form a complete environment map; combining the environment map to assign an exploration task to each robot and model the target area.

[0049] In addition, the logic instruction in the memory 703 described above can be implemented in the form of a software function unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium, includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various storage program codes.

[0050] On the other hand, the present application also provides a computer program product, which includes a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program is executed by a processor, the computer can execute the autonomous multi-robot collaborative SLAM method provided by the above-mentioned method, which includes: receiving detection data from different sensors of the robot, and preprocessing the detection data; fusing the preprocessed detection data without distortion to obtain robot state estimation; combining the robot state estimation to generate an obstacle avoidance navigation path for each robot; Integrate the obstacle avoidance navigation path of each robot to form a complete environment map; In combination with the environment map, assign an exploration task to each robot and model the target region.

[0051] In another aspect, the application also provides a non-transitory computer readable storage medium having stored thereon a computer program, which, when executed by a processor, implements the autonomous multi-robot collaborative SLAM method provided by the above method, which comprises: Receiving detection data from different sensors of the robot and pre-processing the detection data; Non-distortionally fusing the pre-processed detection data to obtain robot state estimation; In combination with the robot state estimation, generating an obstacle avoidance navigation path for each robot; Integrate the obstacle avoidance navigation path of each robot to form a complete environment map; In combination with the environment map, assign an exploration task to each robot and model the target region.

[0052] The device embodiments described above are only schematic, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment scheme. Those skilled in the art can understand and implement without creative labor.

[0053] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software plus the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions, essentially or in terms of the contribution to the prior art, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0054] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit the same; and although the present application has been described in detail with reference to the foregoing embodiments, it should be appreciated by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features thereof can be replaced equivalently; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An autonomous multi-machine collaborative SLAM method, characterized in that, include: Receive detection data from different sensors of the robot and preprocess the detection data; The preprocessed detection data is then fused without distortion to obtain a robot state estimate. Based on the robot state estimation, an obstacle avoidance navigation path is generated for each robot; The obstacle avoidance navigation paths of each robot are integrated to form a complete environmental map; Based on the environmental map, each robot is assigned an exploration task, and environmental modeling is performed on the target area.

2. The autonomous multi-machine collaborative SLAM method according to claim 1, characterized in that, Receive detection data from different sensors of the robot, and preprocess the detection data, including: Receive detection data from the robot's inertial odometry and binocular camera; The detection data is then subjected to soft synchronization alignment in time and space. The weight values ​​of each of the detection data are determined by the breadth weighting algorithm, and the aligned binocular depth map is obtained. The pixel layer is determined, and the binocular depth map is converted into two-dimensional laser point cloud data.

3. The autonomous multi-machine collaborative SLAM method according to claim 1, characterized in that, The preprocessed detection data is then fused without distortion to obtain a robot state estimate, including: Based on the robot's state at the previous moment, predict the robot's state at the current moment; the robot state includes position information and posture information. The detection data from different sensors are fused together to correct the predicted robot state, thus obtaining the robot state estimate.

4. The autonomous multi-machine collaborative SLAM method according to claim 1, characterized in that, Based on the robot state estimation, an obstacle avoidance navigation path is generated for each robot, including: Based on hybrid The path planning algorithm and the B-spline soft-constraint path optimization algorithm generate obstacle avoidance navigation paths for the robot.

5. The autonomous multi-machine collaborative SLAM method according to claim 4, characterized in that, Based on hybrid The path planning algorithm and the B-spline soft-constraint path optimization algorithm generate an obstacle avoidance navigation path for the robot, including: A grid map is generated based on the robot's position information, posture information, and binocular depth map. Combined with the aforementioned raster map, a hybrid approach is adopted. The path planning algorithm and the soft-constraint path optimization algorithm of B-spline are used to determine the obstacle avoidance navigation path of the robot in the target area.

6. The autonomous multi-machine collaborative SLAM method according to claim 1, characterized in that, After generating an obstacle avoidance navigation path for each robot, the process includes: smoothing the obstacle avoidance navigation path.

7. The autonomous multi-machine collaborative SLAM method according to claim 5, characterized in that, The obstacle avoidance navigation paths of each robot are integrated to form a complete environmental map, including: Keyframes are acquired through distributed communication of the robot operating system and combined with the local map of the grid map component; Duplicate map points are deleted, while the co-view relationships of the deleted map points are preserved, and keyframes and map points in the merging process are subjected to BA (Balanced Analytical Processing). The keyframes with shared viewing relationships are repaired, and the poses of other keyframes in the local map are optimized to obtain the environment map.

8. The autonomous multi-machine collaborative SLAM method according to claim 1, characterized in that, Based on the aforementioned environment map, exploration tasks are assigned to each robot, and environmental modeling is performed on the target area, including: Read the robot's exploration boundary of the target area; Based on the exploration boundary, a task grid is divided to obtain several exploration tasks; The exploration task is assigned to each robot, which then performs the exploration task in the target area and performs environmental modeling.

9. An autonomous multi-machine collaborative SLAM device, characterized in that, include: The processing module is used to receive detection data from different sensors of the robot and preprocess the detection data; The fusion module is used to fuse the preprocessed detection data without distortion to obtain a robot state estimate; The planning module is used to generate an obstacle avoidance navigation path for each robot by combining the robot state estimation. The integration module is used to integrate the obstacle avoidance navigation paths of each robot to form a complete environmental map; The exploration module, in conjunction with the environment map, assigns exploration tasks to each robot and performs environmental modeling of the target area.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the autonomous multi-machine collaborative SLAM method as described in any one of claims 1 to 8.

Images