Air-ground cooperative robot cooperative positioning system and method

Abstract

The invention discloses an air-ground cooperative robot cooperative positioning system and method. The aerial robot cluster is used for carrying UWB base station beacons to establish base station beacon nodes in a target area; the ground robot cluster is used for carrying the visual module for image acquisition; the UWB base station is used for carrying UWB positioning beacons and carrying out data communication with UWB base station beacons through the UWB positioning beacons to obtain UWB positioning information; the positioning control center is used for receiving the UWB positioning information and the positioning information of the ground robot and fusing the UWB positioning information and the positioning information of the ground robot to obtain fused position information; the target marking module is used for matching the shape and depth distance of a target according to the acquired image and marking the target in a grid map through coordinate conversion; according to the method, the adsorption type unmanned aerial vehicle can be controlled to carry the base station beacon to autonomously establish a positioning system, signal shielding is avoided, hovering and shaking are avoided in combination with the adsorption capacity, more accurate positioning is further achieved, and more accurate external positioning data are provided for a ground robot cluster.

Description

Technical Field

[0001] The present invention relates to the technical field of unknown scenario target positioning, and more specifically, to an air-ground collaborative robot cooperation positioning system and method. Background Art

[0002] Currently, with the improvement of the intelligence level of robots, people gradually begin to use robots to perform some tasks that are relatively dangerous for humans. Among them, positioning target objects in unknown scenarios is an important application scenario for mobile robots. In personnel search and rescue and battlefield exploration, the exploration of a certain target often requires a large amount of manpower and material resources for carpet-style investigation, which not only has low efficiency but also has potential dangers due to unfamiliar terrain. Therefore, we need to explore how to use multiple low-cost mobile robots to replace manual labor to complete tasks.

[0003] The existing similar technology is to combine the LIO radar odometer of a mobile robot and external UWB positioning information to obtain the position of the mobile robot, and then obtain the approximate position of the target object. Such a method is not reliable enough in practical applications. On the one hand, the odometer data of the mobile robot will accumulate large errors as the mobile robot moves. On the other hand, it is often difficult to obtain external UWB positioning information in the scenarios where this type of technology is applied, such as weak communication signals blocked by obstacles and difficult installation of UWB base stations.

[0004] In the existing technology for joint positioning based on base stations, the base stations are often regarded as prerequisites within the detection area and are fixed on the ground of the detection area, unable to be flexibly replaced, which greatly affects the effectiveness of the task. At the same time, when using unmanned equipment for positioning in extreme environments, it is difficult to install base stations, resulting in the inability to effectively provide external positioning information. In addition, in the existing air-ground collaboration, drones often carry relay positioning modules to assist in positioning. Since they need to perform self-positioning first, multiple segments of errors are easily introduced when using positioning algorithms, affecting the positioning accuracy.

[0005] Therefore, how to improve positioning reliability and environmental adaptability is an urgent problem for those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention provides an air-ground collaborative robot cooperation positioning system and method, which can independently establish a positioning system, avoid signal occlusion, and avoid hovering shaking by combining adsorption ability, thereby achieving more accurate positioning and providing more accurate external positioning data for a ground robot cluster.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] An air-ground collaborative robot cooperation positioning system, comprising: an aerial robot cluster, a ground robot cluster, a UWB base station beacon, a UWB positioning beacon, a vision module and a positioning control center.

[0009] The aerial robot cluster is used to carry the UWB base station beacon to establish a base station beacon node in the target area; wherein, the aerial robot cluster includes a plurality of adsorption drones, and each adsorption drone respectively carries the corresponding UWB base station beacon to a designated position in the target exploration area and performs adsorption fixation to construct a local coordinate system.

[0010] The ground robot cluster is used to carry the vision module for image acquisition; it is used to carry the UWB positioning beacon, and perform data communication with the UWB base station beacon through the UWB positioning beacon to confirm its own position in the local coordinate system and obtain UWB positioning information.

[0011] The positioning control center is used to fuse the UWB positioning information and the self-positioning information of the ground robot to obtain the fused position information; it is used to match the target shape and depth distance according to the acquired image, and mark the target in the grid map through coordinate transformation.

[0012] Preferably, the number of the adsorption drones is at least three. The drones confirm the fixed position under the control of the positioning control center, confirm the origin and two coordinate axes to obtain the local coordinate system, and the local coordinate system covers the ground space of the target exploration area.

[0013] Preferably, the ground robot is also equipped with a wheel odometer and an IMU sensor for generating the self-positioning information.

[0014] Preferably, the positioning control center includes a control module, a data receiving module and a data fusion module.

[0015] The control module is communicatively connected to the aerial robot cluster and the ground robot cluster, and is used to control the aerial robot to reach the designated position and form a target exploration area; it is used to control the ground robot to perform image acquisition and explore the target in the target exploration area; the data receiving module is used to receive the self-positioning information and the UWB positioning information; the data fusion module is used to fuse according to the UWB positioning information and the self-positioning information and output the fused position coordinates.

[0016] Preferably, the control center further includes a map fusion module, and the map fusion module is used to obtain the local grid maps generated by each ground robot using the radar module and fuse the local grid maps.

[0017] Preferably, the map fusion module includes a map fusion unit and a status update unit; the map fusion unit is used to receive in real time the local grid maps constructed by the radar modules of different ground robots and update and fuse them into a unified grid map at a certain frequency; the status update unit is used to update in real time the postures and positions of the ground robots in the grid map.

[0018] Preferably, the extended Kalman filter algorithm is used to fuse the UWB positioning information and the self-positioning information, which specifically includes:

[0019] Confirm the predicted state according to the self-positioning information, and calculate the observation residual with reference to the UWB positioning information; calculate the Kalman gain, and update the state according to the observation parameters to obtain the fused position coordinates.

[0020] Preferably, the vision module includes an object recognition unit, a depth perception unit, a coordinate conversion unit and an object marking unit;

[0021] The object recognition unit is used to collect environmental information in real time, run an object recognition algorithm, generate an object bounding box around the object in the environment for bounding selection, and obtain the pixel coordinates and type of the object.

[0022] The depth perception unit is used to confirm the central pixel point according to the pixel points of the object bounding box and obtain the depth information of the central pixel point.

[0023] The coordinate conversion unit is used to convert the coordinates of the object in the pixel coordinate system to the world coordinate system through the camera coordinate system in combination with the fused position information of the ground robot.

[0024] The object marking unit is used to mark the position of the object in the fused grid map according to the coordinates of the object in the world coordinate system.

[0025] Preferably, the coordinate conversion includes:

[0026] Based on the camera internal parameter matrix and depth information, convert the position coordinates of the object from the pixel coordinate system to the camera coordinate system.

[0027] Combine the external parameter matrix and translation vector of the camera to transfer the position coordinates of the object from the camera coordinate system to the pedestal coordinate system.

[0028] Combine the IMU quaternion and the fused position coordinates of the ground robot to obtain the coordinates of the object in the world coordinate system.

[0029] An air-ground collaborative robot cooperative positioning method includes the following steps:

[0030] Deploy an aerial robot cluster, control multiple adsorption drones to carry UWB base station beacons to designated positions in the target exploration area respectively, and construct a position coordinate system under the target exploration area based on the designated positions.

[0031] Deploy a ground robot cluster, control multiple ground robots to carry radar modules, UWB positioning beacons and vision modules to the initial positions; initialize the coordinates of each ground robot according to the initial positions.

[0032] Control the ground robots to move within the target exploration area and identify the target by collecting images; when moving, control the UWB positioning beacons to communicate with the UWB base station beacons, and calculate the UWB positioning information of each ground robot; synchronously collect the UWB positioning information and the self-positioning information of the ground robots for data fusion to obtain the final position information.

[0033] When the target is recognized, combine the position information and the coordinate transformation algorithm to transform the coordinates of the target in the pixel coordinate system into the world coordinate system and mark them in the grid map.

[0034] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses an air-ground collaborative robot cooperative positioning system and method, which can actively control the drone to carry the UWB base station beacon to fly to a suitable adsorption point to build a positioning coordinate system at a high position, can avoid signal occlusion, and combines the adsorption ability to avoid hovering jitter, thereby realizing more accurate positioning, providing accurate external positioning data for the ground robot cluster; at the same time, it avoids the power consumption of long-term hovering, is more suitable for large-scale and long-term positioning operations; can more flexibly adjust the exploration area range. It uses accurate external positioning data and self-positioning data for fusion, and uses the fused data to confirm the target coordinates, avoiding the positioning drift problem caused by relying only on odometer data, and has strong autonomy and environmental adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the provided drawings.

[0036] Figure 1 It is a schematic structural diagram of an air-ground collaborative multi-mobile robot cooperative positioning system provided by the present invention;

[0037] Figure 2Schematic diagram of the fusion process of the data fusion module in an air-ground collaborative multi-mobile robot collaborative positioning system provided by the present invention;

[0038] Figure 3 Schematic diagram of the algorithm process of the visual perception module in an air-ground collaborative multi-mobile robot collaborative positioning system provided by the present invention.

[0039] Figure 4 Schematic diagram of the process of an air-ground collaborative multi-mobile robot collaborative positioning method provided by an embodiment of the present invention. Detailed implementation manners

[0040] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0041] Embodiment 1

[0042] As Figure 1 , an embodiment of the present invention discloses an air-ground collaborative robot collaborative positioning system, including: an aerial robot cluster, a ground robot cluster, a UWB base station beacon, a UWB positioning beacon, a vision module, and a positioning control center;

[0043] The aerial robot cluster is used to carry the UWB base station beacon to establish a base station beacon node in the target area;

[0044] The ground robot cluster is used to carry the vision module for image acquisition; used to carry the UWB positioning beacon and perform data communication with the UWB base station beacon through the UWB positioning beacon to obtain UWB positioning information;

[0045] The positioning control center is used to receive the UWB positioning information and the self-positioning information of the ground robot and perform fusion to obtain the fused position information; used to match the target shape and depth distance according to the collected image, and mark the target in the grid map through coordinate transformation.

[0046] In this embodiment, the aerial robot cluster is composed of multiple drones, which carry UWB positioning beacons to construct a coordinate system in the target exploration area; specifically, the adsorption-type drone is a self-developed horizontal double-rotor drone. The special layout of the double rotors significantly improves the stability and flexibility of the rotorcraft and is also convenient for adsorption. The shell and the vector power system base are made of 3D printing nylon material and carbon fiber composite material, and the suction cup negative pressure adsorption scheme or the gripper grasping scheme can be adopted according to different usage scenarios.

[0047] The ground robot cluster consists of multiple ground robots. Each ground robot is equipped with an Intel NUC as an on-board computer to handle the required computing tasks. The robots are powered by lithium batteries and have a battery life of about 1 to 2 hours. They adopt an omnidirectional wheel structure, which facilitates flexible movement in complex scenarios. The vehicle is equipped with a wheel odometer, an IMU, and a UWB positioning beacon to achieve integrated positioning. At the same time, each vehicle is equipped with a lidar to generate environmental point cloud information for SLAM mapping, and a binocular depth camera as a vision module for target recognition and positioning.

[0048] In addition, a self-organizing network device is configured in the positioning control center and each robot to achieve long-distance self-organizing network communication and mutual transmission within the cluster.

[0049] To further implement the above technical solution, a CUAV V5+ flight controller is set in the positioning control center to control the drone to carry the corresponding equipment and reach the designated position in a preset formation; the drone adopts an adsorption-type drone, which can turn on the air pump after reaching the designated position to fix itself on the nearby adsorbable object and stop the propeller.

[0050] Furthermore, there are four adsorption-type drones in total. They are manually maneuvered to the adsorbable objects at the four top corners of the unknown area in a predetermined rectangular formation, and the air pump is turned on to make them adsorb to the adsorbable objects and stop the propellers. At this time, the UWB on the adsorption-type drones acts as the UWB base station beacon in the positioning module, and the four adsorption-type drones determine the origin, X-axis, and Y-axis of the coordinate system. Drone 0 is used as the origin, the line connecting Drone 0 and Drone 1 is used as the Y-axis, and the line connecting Drone 0 and Drone 3 is used as the X-axis to form a coordinate system. Drone 2 is a redundantly arranged base station, which can improve the positioning reliability on the basis of achieving positioning.

[0051] There are two ground robots in total. Similarly, taking the coordinate system constructed by the adsorption-type drones as the reference coordinate system, they are placed at the corresponding positions according to the requirements. It should be noted that in order to facilitate the subsequent coordinate fusion correction and map fusion of the two ground robots, the original coordinates Odom of the two vehicles need to be modified to match the actual placement positions. For example, Robot 0 starts from the origin, while Robot 1 starts from (0,2). Therefore, Odom 1 needs to be changed from (0,0) to (0,2). At this time, the UWB on the ground robot acts as the UWB positioning beacon in the ground module, and determines its coordinates in the coordinate system by measuring the distance from it to the base station beacon in real time.

[0052] To further implement the above technical solution, the positioning control center includes a control module, a data receiving module, and a position data fusion module.

[0053] The control module is communicatively connected to the aerial robot cluster and the ground robot cluster, and is used to control the aerial robots to reach the designated positions and form a target exploration area; and is used to control the ground robots to perform image acquisition within the target exploration area to explore the target; the data receiving module is used to receive the odometry information and the UWB positioning information; the data fusion module is used to perform data fusion based on the UWB positioning information and its own positioning information to correct the position coordinates.

[0054] In this embodiment, the data fusion module uses the extended Kalman filter algorithm to fuse and calculate the UWB positioning information Odom UWB and the odometer information and the imu data, as Figure 2 , and the specific calculation process is as follows:

[0055] (1) First, define the state vector of the robot: x = [x, y, z] T , where x, y, and z represent the three-dimensional position of the robot.

[0056] (2) Data acquisition.

[0057] Obtain the positioning information of the odometer at time k, specifically x odom,k = [x odom,k , y odom,k , z odom,k T and the UWB positioning information, specifically x UWB,k = [x UWB,k , y UWB,k , z UWB,k T .

[0058] (3) State prediction. At time k, the state prediction is directly provided by the coordinates of the odometer:

[0059]

[0060] Prediction covariance:

[0061] P k = P k-1 + Q

[0062] where P k-1 represents the covariance matrix at time k - 1. The covariance matrix P 0 at the initial time can be set according to the positioning accuracy of the hardware. Q describes the covariance of the process noise, which characterizes the uncertainty of the robot modeling odometer. If the modeling is accurate enough, the diagonal elements of Q can be set to be small enough.

[0063] (4) Data fusion processing.

[0064] ① Define the observation model. For the observations provided by UWB, the measurement model can be defined as follows:

[0065]

[0066] where \(x\) k , \(y\) k , \(z\) k are the true three-dimensional position values of the robot at time \(k\), and \(\delta\) UWB,k is the observation noise. Define the following measurement noise covariance matrix:

[0067]

[0068] where \(\sigma\) x , \(\sigma\) y , \(\sigma\) z are the standard deviations of the UWB measurement noise, and their specific values depend on the hardware accuracy of UWB.

[0069] ② Calculate the observation residual.

[0070] \(e\) k = \(x\) UWB,k - \(x\) odom,k

[0071] ③ Calculate the Kalman gain.

[0072] \(K\) k = \(P\) k \(H\) T (\(HP\) k \(H\) T + \(R\) UWB ) -1

[0073]

[0074] where \(K\) k represents the Kalman gain at time \(k\), and \(H\) is the Jacobian matrix of the measurement equation (the identity matrix under the linear model).

[0075] ④ State update.

[0076]

[0077] ⑤ Covariance update.

[0078] \(P\) k+1 = (1 - \(K\) k \(H\))\(P\) k

[0079] \(x\) k+1 = \(f(x\) k , \(u\) k ) + \(w\) k

[0080] Among them, P k is the covariance matrix at time k, and I represents the three-dimensional unit vector. The fused state is the robot coordinate Odom after data fusion of the robot at time k fusion ; u k+1 is the predicted state at time k + 1, and u k is the control input provided by the IMU in the robot, and w k is the process noise vector. The fusion calculation effectively eliminates the coordinate drift caused by the cumulative error of the odometer.

[0081] Furthermore, the positioning control center further includes a map fusion module, which can combine the local grid maps generated by multiple ground robots using their mounted radar modules to obtain a complete grid map. Furthermore, target marking can be achieved in this map.

[0082] Specifically, the map fusion module includes a map fusion unit and a state update unit; the map fusion unit is used to receive in real time the grid maps constructed by the radar modules of different ground robots and update and fuse them into a unified grid map at a certain frequency; the state update unit is used to update in real time the states of each ground robot in the grid map, including the direction, trajectory and position in the fused map of the robot. The specific implementation mainly depends on the quaternion of the robot IMU and the corrected coordinate Odom fusion .

[0083] Among them, the map fusion unit adopts the multirobot_map_merge function of the m-explore package, optimizes the parameters of the collaborative mapping algorithm according to the computing power level of the robot on-board computer, and makes the coordinate system setting of the grid map Map consistent with the coordinate system constructed by UWB. The specific parameter optimization includes the number of particles of the ion filter, the map update frequency, the scan matching frequency and the number of iterations, etc.

[0084] In addition, in order to ensure the time consistency of the map data of each fused robot, the NTP (Network Time Protocol) timestamp alignment technology is used to synchronize each on-board computer with its ground station, so as to reduce the map misalignment problem caused by time error.

[0085] To further implement the above technical solution, the vision module includes a target recognition unit, a depth perception unit, a coordinate conversion unit and a target marking unit.

[0086] The target recognition unit is used to collect environmental information in real time and run the YOLOv5 target recognition algorithm to accurately frame the target object from the complex environment and give the type information. Specifically, the environment and dependencies required for the operation of YOLOv5 have been pre-configured on the on-board computer of the ground robot. The dataset of the target object has been collected in advance and trained on a high-performance computer. The trained model is the lightweight model YOLOv5s suitable for on-board device operation. After running the algorithm, the binocular depth camera will be turned on to collect the image information of the environment where the ground robot is located in real time and compare it with the trained model at all times. If it is higher than the confidence threshold, the target object will be framed in the image and the type will be given.

[0087] The depth perception unit is used to calculate the center pixel point P of the border from the pixel points of the target border given by the target recognition unit and obtain the depth information Depth of the depth camera at point P. Specifically, P specifically depends on the image resolution. In this embodiment, a resolution of 640*480 is adopted. Let the pixel coordinates xyxy_list[i][0], xyxy_list[i][1], xyxy_list[i][2], xyxy_list[i][3] of the four vertices of the known border be known, where i represents the i-th target object recognized in the image frame, and the coordinates of the center pixel point P of the border are (P x , P y ) are:

[0088] P x = int((xyxy_list[i][0] + xyxy_list[i][2]) / 2)

[0089] P y = int((xyxy_list[i][1] + xyxy_list[i][3]) / 2)

[0090] The coordinate conversion unit is used to convert the target object from the camera coordinate system to the world coordinate system; based on the camera internal parameter matrix K, the coordinates of the center pixel point P of the border, the depth Depth, the external parameter matrix R, the translation vector T, the corrected coordinates Odom of the robot fusion and the IMU quaternion Q of the robot to perform coordinate conversion, convert the coordinates of the center pixel point P of the border from the pixel coordinate system to the camera coordinate system, then from the camera coordinate system to the body coordinate system, and finally from the base coordinate system to the world coordinate system, and calculate the position Pos_Target of the target object in the fused map Map.

[0091] The target marking unit is used to mark the position in the fused grid map according to the coordinates of the target in the world coordinate system.

[0092] Furthermore, the specific coordinate conversion steps include:

[0093] As Figure 3 , first is the conversion from the pixel coordinate system to the camera coordinate system. To convert from pixel coordinate system coordinates to camera coordinate system coordinates, it is necessary to additionally obtain the internal parameter matrix K (camera internal parameters), depth information Depth (the depth camera emits infrared laser so that each pixel point can obtain its corresponding depth information), and distortion parameters. Assuming that the camera has no distortion, the camera internal parameter matrix is K, and the expression of K is:

[0094]

[0095] where f x and f y are the focal lengths of the depth camera, and (c x , c y ) is the optical center coordinate. These parameters are calibrated when the camera leaves the factory and can be directly obtained. The calculation formula for converting from the pixel coordinate system to the camera coordinate system is:

[0096]

[0097] Secondly is the conversion from the camera coordinate system to the body coordinate system. The coordinates in the camera coordinate system are only relative to the visual perception module, and the visual module is installed on the ground robot. There are rotational and translational changes relative to the position estimation point (IMU) of the robot. This change can be described by the external parameter matrix R and translation vector T of the camera. The rotation matrix R is usually obtained through camera calibration or sensor fusion means to ensure that it accurately reflects the actual relationship between the camera and the body coordinate system. The translation vector T can be determined by actual measurement or estimation methods, indicating the position offset of the camera in the body coordinate system. Since the accuracy requirement for this embodiment is at the 1m level, the actual measurement method is used to obtain R and T. Assuming that the coordinates of the target object in the body coordinate system are (X b , Y b , X b ), then the conversion from camera coordinates to body coordinates is:

[0098]

[0099] Finally is the conversion from the base coordinate system to the world coordinate system.

[0100] After obtaining the body coordinate system, it is equivalent to clarifying the coordinates of the target object with the IMU of the robot as the origin and the IMU coordinate axes as the coordinate axes. However, when the robot discovers the target object, it has already left the starting point by a certain distance, and its starting point may not be the origin of the true world coordinate system. Therefore, to convert from the base coordinate system to the world coordinate system, it is necessary to obtain the world coordinate information of the robot, including Odom obtained through the fusion calculation of IMU, wheel odometer, and UWB fusionand the IMU quaternion Q rotated relative to the world coordinate system. Let the world coordinates of the robot be The IMU quaternion is Q = (q x, q y, q z, q w ). The coordinates of the target object in the world coordinate system are Pos_Target = (X w , Y w , Z w ). Then the formula for converting from the pedestal coordinate system to the world coordinate system is:

[0101]

[0102] where R.from_quat(q x, q y, q z, q w ) is to convert the quaternion to a rotation matrix.

[0103] Embodiment 2

[0104] As Figure 4 , based on the same inventive concept, an embodiment of the present invention discloses a method for collaborative positioning of an air-ground collaborative robot, including the following steps:

[0105] Deploy an aerial robot cluster, control multiple drones to carry UWB base station beacons to designated positions in the target exploration area respectively, and construct a local coordinate system under the target exploration area based on the designated positions; Exemplarily, use adsorption drones to carry UWB base station beacons and adsorb them to the adsorption points at the four corners of the fixed unknown area.

[0106] Deploy a ground robot cluster, control multiple ground robots to carry radar modules, UWB positioning beacons and vision modules to the initial positions; Among them, the vision module can use a binocular depth camera, and in addition, equipment such as an on-board computer can also be carried. After reaching the predetermined position, initialize the coordinates of each ground robot according to the initial position.

[0107] Control the ground robots to move within the target exploration area and identify the target by collecting images; When moving, control the UWB positioning beacon to communicate with the UWB base station beacon, and calculate the UWB positioning information of each ground robot; Synchronously collect the UWB positioning information and the self-positioning information of the ground robots for data fusion to obtain the final position information.

[0108] When the target is recognized, combine the position information and the coordinate system conversion algorithm to convert the coordinates of the target in the pixel coordinate system into the world coordinate system and mark them in the grid map.

[0109] In the present specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, reference can be made to the description in the method section.

[0110] The foregoing description of the disclosed embodiments enables those skilled in the art to practice or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Thus, the invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An air-ground collaborative robot collaborative positioning system, characterized in that: Including: aerial robot cluster, ground robot cluster, UWB base station beacon, UWB positioning beacon, vision module and positioning control center; The aerial robot cluster is used to carry the UWB base station beacon to establish a base station beacon node in the target area; wherein the aerial robot cluster includes a plurality of adsorption-type drones, each of which carries the corresponding UWB base station beacon to a designated position in the target exploration area and is adsorbed and fixed to construct a local coordinate system; The ground robot cluster is used to carry the visual module for image acquisition; and is used to carry the UWB positioning beacon, and to communicate data with the UWB base station beacon through the UWB positioning beacon, confirm its own position in the local coordinate system, and obtain UWB positioning information; The positioning control center is used to fuse the UWB positioning information and the ground robot's own positioning information to obtain fused position information; It is used to match the target shape and depth distance according to the acquired image, and mark the target in the grid map through coordinate conversion.

2. The air-ground collaborative robot collaborative positioning system according to claim 1, characterized in that: The number of the suction-type UAVs is at least three. The UAVs confirm a fixed position, an origin and two coordinate axes under the control of the positioning control center to obtain the local coordinate system, which covers the ground space of the target exploration area.

3. The air-ground collaborative robot collaborative positioning system according to claim 1, characterized in that: The ground robot is also equipped with a wheel odometer and an IMU sensor for generating the self-positioning information.

4. The air-ground collaborative robot collaborative positioning system according to claim 1, characterized in that: The positioning control center includes a control module, a data receiving module and a data fusion module; The control module is in communication connection with the aerial robot cluster and the ground robot cluster, and is used to control the aerial robots to reach a designated location and form a target exploration area; Used to control the ground robot to collect images and explore the target in the target exploration area; The data receiving module is used to receive the self-positioning information and the UWB positioning information; The data fusion module is used to fuse the UWB positioning information and the self-positioning information, and output the fused position coordinates.

5. The air-ground collaborative robot collaborative positioning system according to claim 4, characterized in that: The control center also includes a map fusion module, which is used to obtain the local grid map generated by each ground robot using the radar module and fuse the local grid maps.

6. The air-ground collaborative robot collaborative positioning system according to claim 5, characterized in that: The map fusion module includes a map fusion unit and a status update unit; The map fusion unit is used to receive in real time the local grid maps constructed by radar modules of different ground robots, and update and fuse them into a unified grid map at a certain frequency; The state updating unit is used to update the posture and position of each of the ground robots in the grid map in real time.

7. The air-ground collaborative robot collaborative positioning system according to claim 4, characterized in that: The UWB positioning information and the self-positioning information are fused using an extended Kalman filter algorithm, specifically including: Confirm the predicted state according to the own positioning information, and calculate the observation residual with reference to the UWB positioning information; Calculate the Kalman gain and update the state according to the observed parameters to obtain the fused position coordinates.

8. The air-ground collaborative robot collaborative positioning system according to claim 1, characterized in that: The visual module includes a target recognition unit, a depth perception unit, a coordinate conversion unit and a target marking unit; The target recognition unit is used to collect environmental information in real time, and run a target recognition algorithm to generate a target frame around the target object in the environment for selection, and obtain the pixel coordinates and type of the target object; The depth sensing unit is used to determine the central pixel point according to the pixel points of the target frame, and obtain the depth information of the central pixel point; The coordinate conversion unit is used to convert the coordinates of the target object in the pixel coordinate system into the world coordinate system through the camera coordinate system in combination with the fused position information of the ground robot; The target marking unit is used to mark the position of the target object in the fused grid map according to the coordinates of the target object in the world coordinate system.

9. The air-ground collaborative robot collaborative positioning system according to claim 8, characterized in that: The coordinate transformation includes: Based on the camera intrinsic parameter matrix and depth information, the target object position coordinates are converted from the pixel coordinate system to the camera coordinate system; Combining the camera's extrinsic matrix and translation vector, the target object position coordinates are transferred from the camera coordinate system to the base coordinate system; The coordinates of the target object in the world coordinate system are obtained by combining the IMU quaternion and the fused ground robot position coordinates.

10. A collaborative positioning method for air-ground collaborative robots, characterized in that: The positioning system according to any one of claims 1 to 9 is adopted, comprising the following steps: Deploy an aerial robot cluster, control multiple suction-type drones to carry UWB base station beacons to designated locations in the target exploration area, and construct a local coordinate system under the target exploration area based on the designated locations; Deploy a ground robot cluster, control multiple ground robots equipped with radar modules, UWB positioning beacons and vision modules to an initial position; initialize the coordinates of each ground robot according to the initial position; Control the ground robot to move within the target exploration area and identify the target by collecting images; When moving, the UWB positioning beacon is controlled to communicate with the UWB base station beacon, and the UWB positioning information of each of the ground robots is calculated; the UWB positioning information and the ground robot's own positioning information are synchronously collected for data fusion to obtain the final position information; When a target is identified, the coordinates of the target in the pixel coordinate system are converted into the world coordinate system by combining the position information and the coordinate system conversion algorithm, and are marked in the grid map.

Images