A space manipulator on-orbit parameter identification system based on binocular vision

Abstract

A space manipulator on-orbit parameter identification system based on binocular vision is used for identifying on-orbit kinematics and dynamics parameters of the manipulator, comprising an information processing unit, a GNC controller, a TT&C subsystem, a manipulator motion planning and control subsystem, and a ground information processing subsystem. Based on high-precision visual measurement of a binocular camera, target measurement points are arranged on the manipulator, and visual measurement information of the motion process of the manipulator is acquired by the binocular camera when the manipulator performs different operation tasks in the field of view of the binocular camera. The actual on-orbit motion process of the manipulator is compared with a simulation model to acquire the actual on-orbit operation characteristics of the manipulator and identify on-orbit parameters of the manipulator that cannot be accurately measured on the ground. The problem of the difference between on-orbit dynamics parameters of the space manipulator and those on the ground is solved.

Description

TECHNICAL FIELD

[0001] The application relates to a space mechanical arm in-orbit parameter identification system based on binocular vision and belongs to the field of mechanical arm kinematics and dynamics visual measurement and parameter identification. BACKGROUND

[0002] The space mechanical arm is one of important support equipment of a space station and undertakes tasks such as cabin section transfer docking, suspended aircraft capturing and auxiliary docking, support for extravehicular activity of astronauts, extravehicular cargo carrying, extravehicular state inspection, exposure platform experimental load care and optical platform care. The space mechanical arm is inevitably caused to change in structure and other performance parameters due to severe vibration in a launching process and great difference between an extreme environment in an in-orbit operation and a ground environment, so it is very important to obtain in-orbit parameters of the mechanical arm. Based on flight data of a space shuttle mechanical arm in the past nearly 30 years, MDA company has realized task verification based on full simulation in the development and use of the Canadian II arm and replaced verification based on hardware. The Canadian II arm captures a "Dragon" spaceship task is realized by full simulation, that is, task simulation and verification.

[0003] The above technical methods all obtain motion and dynamics information of the mechanical arm through mechanical arm self sensors (such as joint torque, position sensors, a mechanical arm end camera and the like), the accuracy and measurement range of the mechanical arm self sensors are limited, so that only effective motion information of the mechanical arm in a limited range can be obtained; meanwhile, the measurement results of the mechanical arm self sensors are inevitably coupled with the mechanical arm motion, further influencing the effectiveness of the measurement results of the mechanical arm. SUMMARY

[0004] The technical problem solved by the application is that, in the prior art, the effective motion range of mechanical arm parameter identification is limited, the measurement information is coupled with the mechanical arm motion, the in-orbit identification time period is long and the measurement accuracy is low, and a space mechanical arm in-orbit parameter identification system based on binocular vision is provided.

[0005] The application solves the above technical problem by the following technical scheme:

[0006] The space mechanical arm in-orbit parameter identification system based on binocular vision identifies in-orbit kinematics and dynamics parameters of the mechanical arm, and is characterized by comprising an information processing unit, a GNC controller, a measurement and control subsystem, a mechanical arm motion planning and control subsystem and a ground information processing subsystem.

[0007] The space mechanical arm in-orbit parameter identification system based on binocular vision identifies in-orbit kinematics and dynamics parameters of the mechanical arm, and is characterized by comprising an information processing unit, a GNC controller, a measurement and control subsystem, a mechanical arm motion planning and control subsystem and a ground information processing subsystem.

[0008] The information processing unit receives an instruction signal sent by the GNC controller, controls the binocular camera to measure target motion information, and transmits measurement image information sent by the binocular camera to the measurement and control subsystem after compression processing; the instruction signal includes a camera exposure index and sampling frequency information.

[0009] The GNC controller controls the information processing unit and the mechanical arm motion planning and control subsystem according to ground measurement and control information, controls the motion of the mechanical arm through the mechanical arm motion planning and control subsystem, and measures the motion of the mechanical arm through the information processing unit.

[0010] The measurement and control subsystem receives image information sent by the information processing unit, communicates with the ground information processing subsystem during the measurement and control arc segment, and transmits the image information.

[0011] The mechanical arm motion planning and control subsystem controls the motion trajectory of the mechanical arm according to a preset mechanical arm path, and realizes the motion of the mechanical arm within the measurement effective range of the information processing unit.

[0012] The ground information processing system obtains a spatial motion information sequence of a measurement point on the mechanical arm from image information, and obtains mechanical arm motion information and a mechanical arm dynamics model by processing the spatial motion information sequence, to determine actual dynamics characteristic parameters of the mechanical arm in orbit.

[0013] The watchdog module is further included, which is used for monitoring the core voltage during the operation of the information processing unit, resetting and restarting the information processing unit when the core voltage is lower than the normal working demand voltage, generating an error code and sending the GNC controller.

[0014] The information processing unit includes an instruction receiving unit, an image synthesis unit, an exposure control unit and an image compression unit.

[0015] The instruction receiving unit receives a control instruction of the GNC controller, controls the binocular camera, sends an exposure control instruction to adjust the exposure mode of the binocular camera, and includes a manual exposure mode and an automatic exposure mode.

[0016] The image synthesis unit receives measurement image information and measurement noise data output by the binocular camera at the current time, synthesizes the current time data and the compressed measurement image information into a single clear photo.

[0017] The exposure control unit receives a control instruction of the GNC controller, adjusts the exposure time and exposure gain of the binocular camera.

[0018] The image compression unit receives the single clear photo sent by the image synthesis unit for gray correction, compresses the image, and transmits the processed image information to the measurement and control subsystem.

[0019] The GNC controller receives ground remote control instructions, extracts ground measurement and control information, turns on and off the binocular camera, powers on, collects and transmits data, calls the mechanical arm motion planning and control subsystem to control the in-orbit motion of the mechanical arm, including mechanical arm motion, stopping, braking, advancing and reversing.

[0020] The ground information processing system comprises a binocular image ground solution platform and a dynamics parameter identification platform, wherein:

[0021] The binocular image ground solution platform acquires image information and acquires a spatial motion information sequence of a measurement point on the mechanical arm.

[0022] The dynamics parameter identification platform determines actual end pose information of the mechanical arm in orbit according to the spatial motion information sequence, compares with a known mechanical arm dynamics model, and acquires actual dynamics characteristic parameters of the mechanical arm in orbit.

[0023] In the binocular image ground solution platform, a non-cooperative target at different distances is imaged by using a binocular camera, a stereo image pair of the target at the same time is obtained, a common point is acquired by binocular camera vision matching, a parallax of the binocular camera and three-dimensional information of all measurement points of the mechanical arm are calculated, and a spatial motion information sequence of the measurement point on the mechanical arm is acquired.

[0024] In the dynamics parameter identification platform, a target coordinate system is newly established in an end coordinate system by using known coordinates of four target points at the end of the mechanical arm and by observing data of any three target points selected by the binocular camera, a transformation matrix from the end coordinate system to the target coordinate system is calculated, a target coordinate system is newly established in a core cabin body geometric coordinate system by observing data of any three target points selected by the binocular camera, a transformation matrix from the core cabin body geometric coordinate system to the target coordinate system is calculated, a transformation matrix of the end coordinate system relative to the core cabin body geometric coordinate system is calculated according to the two obtained transformation matrices, the actual end pose information of the target at the end of the mechanical arm is converted and calculated according to the final transformation matrix, actual dynamics characteristic parameters of the selected target point of the mechanical arm in the end coordinate system and the core cabin body geometric coordinate system in orbit are determined, including a vector matrix of the target point in the end coordinate system and the core cabin body geometric coordinate system, and a relative attitude matrix of the end coordinate system and the core cabin body geometric coordinate system.

[0025] In the binocular image ground solution platform, the measurement method of the parallax of the binocular camera and the three-dimensional information of all measurement points of the mechanical arm is as follows:

[0026] It is determined that the imaging planes of the left camera and the right camera of the binocular camera are located in the same plane, a world coordinate system is established, an arbitrary point M in the world coordinate system is selected, and projection points m in the imaging planes of the left camera and the right camera are determinedl , m r ;

[0027] The position parameters of the binocular camera are measured by measuring the camera, and the camera head position of the camera is adjusted to make the optical axis parallel forward;

[0028] The depth distance of M to the measuring camera and the parallax value of the left camera and the right camera are calculated;

[0029] According to the relative positions and focal lengths of the left camera and the right camera, the three-dimensional position of the M point in space is calculated;

[0030] The spatial three-dimensional position of the target point on the mechanical arm at any time is obtained, and the spatial motion information sequence of the measuring point on the mechanical arm is determined.

[0031] In the dynamic parameter identification platform, the target points A, B, C and D at the end of the mechanical arm are selected and the coordinates in the end coordinate system E' are determined, a target coordinate system O1 is newly established at the target point A, and the coordinate conversion matrix of the end coordinate system E' to the target coordinate system O1 is calculated

[0032] The coordinates of the end target points A, B, C and D in the core cabin body geometric coordinate system O0 are determined, and the coordinate conversion matrix of the target coordinate system O1 to the core cabin body geometric coordinate system O0 is calculated

[0033] According to the coordinate conversion matrix The coordinate conversion matrix The coordinates of the end point E' of the mechanical arm in the core cabin body geometric coordinate system O0 are solved

[0034] When the target points A, B and C are selected, the position vector and the coordinate matrix I of the selected target points in the end coordinate system E' are determined, and the position vector of the selected target points in the core cabin body geometric coordinate system O0 is measured.

[0035] According to the calculated coordinates of the target points A, B and C in the core cabin body geometric coordinate system O0 and the coordinates The position vector matrix J is determined;

[0036] According to the coordinate matrix I and the position vector matrix J, the relative attitude matrix of the end coordinate system relative to the core cabin body geometric coordinate system O0 is solved

[0037] In the binocular camera, the three axes of the near-field camera are all better than 5mm, the three axes of the far-field camera are all better than 2mm, and the maximum compressed image output is not greater than 210K bytes; the refresh frequency of the output image is 40f / s; the image information download time is less than 0.5 hours.

[0038] The application has the advantages compared with the prior art:

[0039] (1) The space manipulator on-orbit parameter identification system provided by the application first solves the problem that the effective motion information of the manipulator in a limited range can be obtained only through the sensor of the manipulator in the prior art, and the motion information of the manipulator can be obtained in the field of view of the binocular camera, so that the motion range of the manipulator in the on-orbit identification process is greatly improved, and the on-orbit identification cycle of the manipulator is significantly shortened;

[0040] (2) The space manipulator on-orbit parameter identification system provided by the application is different from the prior art in that the measurement result and the motion of the manipulator are coupled when the motion information of the manipulator is obtained through the sensor of the manipulator, the motion of the manipulator is decoupled from the measurement result through the external sensor of the manipulator, the motion of the manipulator no longer affects the measurement information, and the on-orbit parameter identification accuracy of the space manipulator is greatly improved;

[0041] (3) The space manipulator on-orbit parameter identification system provided by the application adopts the working mode of the dual processors of FPGA and DSP through integrated design, the DSP has multi-core processing capability, the instruction receiving unit, the image synthesis unit, the exposure control unit and the image compression unit can perform real-time parallel operation, the problem that high-precision real-time image real-time simulation operation under multiple simulation requirements cannot be realized by using a single processing chip is solved. Meanwhile, the high-speed parallel processing capability of the FPGA is utilized, the image synthesis and image compression algorithms are operated at high speed, and the accuracy of the on-orbit measurement image can be further improved;

[0042] (4) The on-orbit measurement capability of the high-precision binocular camera is fully utilized, the camera contains 2048*2048 pixels, the monochrome pixel size is 5.5*5.5*mu m, the three-axis measurement accuracy of the near-field camera is better than 5mm, the three-axis measurement accuracy of the far-field camera is better than 2mm, and the space manipulator on-orbit parameter identification system can be widely applied to the professional field of structures and mechanisms that need to identify on-orbit motion and dynamic parameters. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 The space manipulator on-orbit parameter identification system provided by the application is shown in the schematic diagram;

[0044] Figure 2 The binocular camera image ground solution provided by the application is shown in the schematic diagram;

[0045] Figure 3 The manipulator end and the target structure thereof provided by the application are shown in the schematic diagram;

[0046] Figure 4 The manipulator end structure provided by the application is shown in the schematic diagram;

[0047] Figure 5 A mechanical arm end coordinate system origin calculation flowchart is provided for the invention;

[0048] Figure 6 An end posture solution flowchart is provided for the invention; DETAILED DESCRIPTION

[0049] A space mechanical arm in-orbit parameter identification system based on binocular vision is used to identify the in-orbit kinematics and dynamics parameters of the mechanical arm, which includes an information processing unit, a GNC controller, a measurement and control subsystem, a mechanical arm motion planning and control subsystem, and a ground information processing subsystem. Based on high-precision visual measurement of a binocular camera, target measurement points are set on the mechanical arm, and visual measurement information of the mechanical arm motion process is obtained by the binocular camera when the mechanical arm performs different operation tasks in the field of view of the binocular camera. The actual in-orbit motion process of the mechanical arm is compared with the simulation model to obtain the actual in-orbit operation characteristics of the mechanical arm and identify the in-orbit parameters of the mechanical arm that cannot be accurately measured on the ground. The problem of the difference between the dynamics parameters of the space mechanical arm in space and on the ground is solved, specifically as follows:

[0050] The information processing unit receives the instruction signal sent by the GNC controller, controls the binocular camera to measure the target motion information, and transmits the measurement image information sent by the binocular camera to the measurement and control subsystem after compression processing. The instruction signal includes camera exposure index and sampling frequency information. The information processing unit adopts a working mode of dual processors of FPGA and DSP, wherein the DSP has multi-core processing capability.

[0051] The GNC controller cooperates with the information processing unit and the mechanical arm motion planning and control subsystem according to the ground measurement and control information, controls the motion of the mechanical arm through the mechanical arm motion planning and control subsystem, and measures the motion of the mechanical arm through the information processing unit.

[0052] The measurement and control subsystem receives the image information sent by the information processing unit, communicates with the ground information processing subsystem during the measurement and control arc segment, and transmits the image information.

[0053] The mechanical arm motion planning and control subsystem controls the motion trajectory of the mechanical arm according to the preset mechanical arm path, and realizes the motion of the mechanical arm within the measurement effective range of the information processing unit.

[0054] The ground information processing system obtains the spatial motion information sequence of the measurement points on the mechanical arm from the image information, and obtains the mechanical arm motion information and the mechanical arm dynamics model after data processing according to the spatial motion information sequence, to determine the actual dynamics characteristic parameters of the mechanical arm in orbit.

[0055] A watchdog module is further included for monitoring a core voltage during operation of the information processing unit, and when the core voltage is lower than a normal working demand voltage, the information processing unit is reset and restarted, an error code is generated and sent to the GNC controller.

[0056] The selected binocular camera includes 2048*2048 pixels, a monochrome pixel size of 5.5*5.5 μm, and a three-axis measurement accuracy of less than 5 mm for a near-field camera and less than 2 mm for a far-field camera.

[0057] Further description is made below in combination with the accompanying drawings and preferred embodiments:

[0058] In the current embodiment, the on-orbit parameter identification system is as shown in the figure, and the information processing unit includes an instruction receiving unit, an image synthesis unit, an exposure control unit, and an image compression unit. Figure 1

[0059] The instruction receiving unit receives control instructions of the GNC controller, controls the binocular camera, and sends exposure control instructions to adjust the exposure mode of the binocular camera, including an active exposure mode and an automatic exposure mode.

[0060] The image synthesis unit receives measurement image information and measurement noise data output by the binocular camera at the current time, and synthesizes the current time data and the compressed measurement image information into a single clear photo.

[0061] The exposure control unit receives control instructions of the GNC controller, and adjusts the exposure time and exposure gain of the binocular camera.

[0062] The image compression unit receives the single clear photo sent by the image synthesis unit for gray correction, compresses the image, and transmits the processed image information to the measurement and control subsystem.

[0063] The GNC controller receives ground remote control instructions, extracts ground measurement and control information, controls the on-off, power-up, data acquisition and transportation of the binocular camera, calls the mechanical arm motion planning and control subsystem to control the on-orbit motion of the mechanical arm, including mechanical arm motion, stop, brake, forward, and reverse.

[0064] The ground information processing system includes a binocular image ground solution platform and a dynamics parameter identification platform.

[0065] The binocular image ground solution platform acquires image information and acquires a spatial motion information sequence of a measurement point on the mechanical arm.

[0066] ​The dynamic parameter identification platform: according to the spatial motion information sequence, the actual end pose information of the manipulator in orbit is determined, the known dynamic model of the manipulator is simulated and compared, and the actual dynamic characteristic parameters of the manipulator in orbit are obtained;

[0067] In the binocular image ground solution platform, the non-cooperative target at different distances is imaged by using the binocular camera, the stereo image pair of the target at the same time is obtained, and the common points are obtained by binocular camera visual matching, the parallax of the binocular camera and the three-dimensional information of all the measurement points of the manipulator are calculated, and the spatial motion information sequence of the measurement points on the manipulator is obtained;

[0068] In the dynamic parameter identification platform, through the known coordinates of the four target points at the end of the manipulator, the data observed by the binocular camera on any selected three target points is observed, a target coordinate system is newly established in the end coordinate system, the transformation matrix from the end coordinate system to the target coordinate system is calculated, a target coordinate system is newly established in the core cabin body geometric coordinate system through the data observed by the binocular camera on any selected three target points, the transformation matrix from the core cabin body geometric coordinate system to the target coordinate system is calculated, the transformation matrix of the end coordinate system relative to the core cabin body geometric coordinate system is calculated according to the two obtained transformation matrices, and the actual end pose information of the end target of the manipulator is converted and calculated according to the final transformation matrix, the actual dynamic characteristic parameters of the selected target points of the manipulator in the end coordinate system and the core cabin body geometric coordinate system in orbit are determined, including the vector matrix of the target points in the end coordinate system and the core cabin body geometric coordinate system, and the relative attitude matrix of the end coordinate system and the core cabin body geometric coordinate system;

[0069] In the binocular image ground solution platform, the measurement method of the parallax of the binocular camera and the three-dimensional information of all the measurement points of the manipulator is:

[0070] The imaging planes of the left camera and the right camera of the binocular camera are determined to be located in the same plane, a world coordinate system is established, an arbitrary point M in the world coordinate system is selected, and the projection points m l 、m r in the imaging planes of the left camera and the right camera are determined;

[0071] The position parameters of the binocular camera are measured by measuring the camera, and the camera position of the camera is adjusted to make the optical axis parallel forward;

[0072] The depth distance from M to the measuring camera and the parallax values of the left camera and the right camera are calculated;

[0073] According to the relative positions and focal lengths of the left camera and the right camera, the three-dimensional position of the M point in space is calculated;

[0074] The spatial three-dimensional position of the target points on the manipulator at any time is obtained, and the spatial motion information sequence of the measurement points on the manipulator is determined.

[0075] In the dynamic parameter identification platform, target points A, B, C and D at the end of the robot arm are selected and coordinates in the end coordinate system E' are determined, a target coordinate system O1 is newly established at the target point A, and a coordinate conversion matrix of the end coordinate system E' to the target coordinate system O1 is calculated

[0076] Coordinates of the end target points A, B, C and D in the core cabin body geometric coordinate system O0 are determined, and a coordinate conversion matrix of the target coordinate system O1 to the core cabin body geometric coordinate system O0 is calculated

[0077] According to the coordinate conversion matrix The coordinate conversion matrix The coordinates of the end point E' of the robot arm in the core cabin body geometric coordinate system O0 are solved

[0078] When the target points A, B and C are selected, the position vectors and coordinate matrix I of the selected target points in the end coordinate system E' are determined, and the position vectors of the selected target points in the core cabin body geometric coordinate system O0 are measured;

[0079] According to the calculated coordinates of the target points A, B and C in the core cabin body geometric coordinate system O0 and the coordinates The position vector matrix J is determined;

[0080] According to the coordinate matrix I and the position vector matrix J, the relative attitude matrix of the end coordinate system relative to the core cabin body geometric coordinate system O0 is solved

[0081] The binocular monitoring measurement sensor is based on the stereo vision measurement principle, and two probes are arranged on a certain baseline to form a binocular system to image a non-cooperative target at different distances and obtain a stereo image pair of the target at the same time. The common points are obtained through binocular vision matching, and the parallax and three-dimensional information of the feature points are calculated, as shown in Figure 2 The measurement cameras are placed in parallel, and the positions of the cameras are adjusted so that the optical axes are parallel forward, and the imaging planes of the left camera and the right camera are located on the same plane. M is a point in the world coordinate system, and m is the projection point of the point on the imaging plane of the camera l , m r is the projection point of the point on the imaging plane of the right camera, T is the baseline length between the two cameras, and Z is the distance from M to the camera, i.e. the depth of field. l and O r are the optical centers of the left and right cameras, f l and f r are the focal lengths of the left and right cameras, and f l = f r = f. According to the similar triangle principle, the following formula is obtained:

[0082]

[0083] The depth distance of M to the camera can be derived from the above formula:

[0084]

[0085] In the formula, m l -m r is the parallax value d, f and T are obtained by calibration. This is the basic formula of binocular stereo vision, which reveals the relationship that the depth information and the parallax value are inversely proportional;

[0086] In the dynamic parameter identification platform, the basic principle of the space station manipulator end position and posture solution is that there are four target points on the space station manipulator end, and the coordinates of the four target points in the manipulator end coordinate system are known (ground precise measurement). Three of the four target points are used to create a new target coordinate system in the end coordinate system, and the transformation matrix of the manipulator end coordinate system in the new target coordinate system is calculated. Three target point data of the four target points on the end observed by the camera are used to create a new target coordinate system in the core cabin body geometric coordinate system, and the transformation matrix of the new target coordinate system relative to the core cabin body geometric coordinate system is calculated. Through the above two transformation matrices, the transformation matrix of the manipulator end coordinate system relative to the core cabin body geometric coordinate system is solved, and the coordinates of the end coordinate system in the core cabin body geometric coordinate system are solved.

[0087] In order to simplify the symbol description, the coordinates described in this paper are in the reference coordinate system, i.e. the core cabin body geometric coordinate system O0, unless otherwise specified.

[0088] The specific calculation method of the space station manipulator end position and posture solution is as follows:

[0089] The structure diagram of the manipulator end and its target is shown in Figure 3 The layout diagram of the manipulator optical feature mark is shown in Figure 4 The space station manipulator end position calculation process is as follows:

[0090] The coordinates of the end target points A, B, C and D in the end coordinate system E' are obtained by ground precise measurement;

[0091] Based on the coordinates of the target points A, B, C and D in the end coordinate system E', a coordinate system O1 is created at the target point A, and the coordinate conversion relationship between the coordinate systems E' and O1 is calculated

[0092] The coordinates of the end target points A, B, C, and D in the reference coordinate system (core cabin body geometric coordinate system, denoted as O0) are obtained by the camera. Based on the coordinates of the target points A, B, C, and D in O0, the coordinate conversion matrix between the coordinate system O1 and O0 is calculated

[0093] According to the coordinate conversion relationship between the coordinate system E' and O1 The coordinate conversion matrix between the coordinate system O1 and O0 is obtained The coordinates of the end origin E' in the reference coordinate system O0 can be calculated That is, the position matrix of the end coordinate system of the robot in the reference coordinate system.

[0094] The position calculation flowchart of the end coordinate system (origin) of the robot in the reference coordinate system is shown in Figure 5

[0095] The coordinates of the four end target points A, B, C, and D in the end coordinate system E' are obtained by ground precise measurement. Three points A, B, and C (or A, B, and D) are used to construct a coordinate system O1 with A as the origin, as shown in Figure 1 , Figure 2 The specific construction method is as follows, assuming:

[0096]

[0097] The plane ABC normal vector is:

[0098]

[0099] From

[0100]

[0101] Perpendicular to Define The three vectors are the direction vectors of the XYZ axes of the newly constructed coordinate system O1, that is, the representation of the newly constructed coordinate system O1 in the end coordinate system E and the coordinate conversion relationship between the two coordinate systems are completed Thus, the representation of in the coordinate system O1 is obtained:

[0102] To obtain the pose representation of the newly constructed coordinate system O1 in the reference coordinate system O0 and the conversion relationship, the XYZ axis direction vector of the O1 coordinate system, that is, the vector in O0 needs to be measured. According to the same steps as above, the coordinates of A, B, C, and D in O0 are measured by the camera, and the vector ​​The coordinates in O0, i.e. the representation of the newly established coordinate system O1 in the reference coordinate system O0 and the coordinate conversion relationship between the two coordinate systems Thus, the coordinates of the end coordinate system origin E' in the reference coordinate system O0 are obtained The representation in the coordinate system O0 is:

[0103]

[0104] The coordinates of the end coordinate system origin E' in the reference coordinate system O0, i.e. the vector The coordinates in the reference coordinate system O0 are:

[0105]

[0106] is a measured quantity, has been calculated in the foregoing.

[0107] In summary, the coordinates of the end coordinate system origin E' in the reference coordinate system O0, i.e. the position of the end coordinate system in the reference coordinate system, are obtained, denoted as

[0108] Space station manipulator end pose calculation

[0109] Given the coordinates of the end target point in the end coordinate system E' (ground precision measurement) and the reference coordinate system O0 (camera measurement), the rotation matrix of the end coordinate system E relative to the reference coordinate system O0 is calculated through the coordinate conversion relationship The specific calculation process is as follows:

[0110] Let the position vectors of A, B, and C in the end coordinate system E' be:

[0111]

[0112] The coordinates of A, B, and C in the reference coordinate system O0 can be measured by the camera:

[0113]

[0114] From the coordinates of A, B, and C in the end coordinate system E', there is a matrix I:

[0115]

[0116] Combining the coordinates of A, B, and C in the reference coordinate system O0 and the position of the end E in the core cabin body geometric coordinate system (x 0E , y 0E , z 0E ), there is a matrix J:

[0117]

[0118] Let is the rotation matrix of the end coordinate system E relative to the reference coordinate system O0, combined with I, J:

[0119]

[0120] Solve is:

[0121]

[0122] That is, the relative attitude matrix between the end coordinate system and the reference coordinate system.

[0123] The attitude calculation flowchart of the end coordinate system in the reference coordinate system is as Figure 6 shown;

[0124] Specifically, the measurement accuracy of the binocular camera on the target is better than 5mm in three axes, and the measurement accuracy of the near-field camera is better than 5mm in three axes. The measurement accuracy of the far-field camera is better than 2mm in three axes. The original image collected by the camera is compressed and output. The maximum compressed output image of each probe of the left camera and the right camera is not greater than 210K bytes per frame. The refresh frequency of the output image is 40f / s. The data download time is less than 0.5 hours.

[0125] The binocular camera visual measurement process is as follows:

[0126] Real-time acquisition of high-quality images in the camera field of view: in the space environment, during the movement of the cabin body, real-time and continuous imaging can be realized, and the image has high resolution and clarity to meet the needs of visible light-based visual monitoring and measurement; the collected image data is compressed and encoded and output in real time, and is downloaded to the ground through the data transmission system;

[0127] Complete the rapid and reliable detection of visual markers with known size, shape, distribution and number in the camera field of view; output the three-dimensional information of the visual marker: output the relative pose relationship between the visual marker and the camera;

[0128] Adjust the camera imaging parameters to obtain the best quality image: adjust the camera exposure time, frame frequency, light measurement mode and other parameters to ensure the best imaging quality in the space light environment; the adaptability of the camera to weak light conditions can be enhanced by controlling the light source to obtain the best quality image;

[0129] The compressed image data is sent to the information processing unit through the Ethernet port, and the information processing unit stores the received compressed image; after the on-orbit measurement is completed, the information processing unit can download the image data in the specified storage area according to the ground instruction.

[0130] In order to maximize the use of the measurement field of view of the binocular monitoring measurement sensor under various working conditions, the method of using the common field of view of the far field and near field probes is adopted, and the mechanical arm is monitored as a whole field of view.

[0131] After the camera is powered on, it will enter standby mode, only a few modules of the camera FPGA work, the APS chip is in standby mode, and the camera FPGA waits for instructions. At this time, the camera power consumption is low, and the system power consumption is minimized. After receiving the instructions of the GNC, the instructions are judged to determine whether the camera is working in manual exposure mode or automatic exposure mode, so as to start the camera shooting, image return and other operations.

[0132] According to the characteristics of the binocular monitoring measurement sensor camera in orbit application, the camera can first perform automatic exposure before the test starts. The obtained image is downloaded, and the camera exposure parameters are confirmed through ground comparison. At this time, the exposure parameters of the left and right cameras need to be unified to prevent the image matching and measurement accuracy from being affected due to inconsistent exposure time. After confirmation, the exposure parameter upload is completed by the ground to perform manual exposure.

[0133] The mechanical arm feature flag provides a motion measurement target for the binocular camera. In this scheme, the feature flag adopts the form of a target, and the installation of the target ball on the mechanical arm adopts the method of sharing the base with the handrail. It is ensured that each target is within the motion envelope without affecting the motion of the mechanical arm. In addition, due to the possibility that the horizontal base of the wrist handrail is not the same height, there are radial holes distributed on the inclined surface and arc surface, so the target will avoid being distributed on the inclined surface and arc surface with chamfer. Based on the consideration of human-machine efficiency and spatial structure, this scheme adopts a spherical target uniformly distributed on the handrail of the mechanical arm. As shown in Figure 4 (b) shows the target ball on a single handrail. The high-reflective target ball adopts surface treatment processes such as corner cube prism, micro-prism, glass microbeads, and high-reflective white paint to improve the light reflectivity of the target ball, increase the brightness and contrast of the target ball in the binocular camera, and ensure the measurement accuracy of the binocular camera.

[0134] The motion of the mechanical arm is controlled within the field of view of the binocular camera to complete the positioning accuracy, repeatability, brake distance, vibration mode, and damping ratio of the end of the mechanical arm. The binocular camera obtains the motion information of the whole arm of the mechanical arm through the measurement of the target ball, and realizes the identification of the on-orbit parameters of the mechanical arm through ground image processing to obtain the actual on-orbit parameters of the mechanical arm. The response characteristics of the mechanical arm in the on-orbit space environment are accurately understood.

[0135] Based on the differences in parameter measurement methods, the scores to be measured are divided into two categories: kinematic parameters and dynamic parameters. Kinematic parameters include kinematic parameters, end-effector pose transformation curve, end-effector velocity transformation curve, end-effector pose accuracy, settling time, maximum velocity, etc. The kinematic parameter identification process is as follows: Figure 5 As shown; the dynamic parameters include joint torsional stiffness, joint damping, joint friction, joint clearance, robotic arm output force, frequency, maximum end effector force, and maximum end effector torque.

[0136] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solutions of the present invention by utilizing the methods and techniques disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

[0137] The contents not described in detail in this specification are common knowledge to those skilled in the art.

Claims

1. A binocular vision-based on-orbit parameter identification system for a space robotic arm, used to identify the on-orbit kinematic and dynamic parameters of the robotic arm, characterized in that: It includes an information processing unit, a GNC controller, a measurement and control subsystem, a robotic arm motion planning and control subsystem, and a ground information processing subsystem, among which: Information processing unit: Receives command signals sent by the GNC controller, controls the binocular camera to measure the target motion information, and compresses and processes the measurement image information sent by the binocular camera before transmitting it to the measurement and control subsystem; the command signals include camera exposure index and sampling frequency information; GNC Controller: Based on ground measurement and control information, it coordinates with the control information processing unit and the robotic arm motion planning and control subsystem to control the movement of the robotic arm through the robotic arm motion planning and control subsystem, and measures the movement of the robotic arm through the information processing unit. The telemetry and control subsystem receives image information sent by the information processing unit and communicates with the ground information processing subsystem within the telemetry and control arc to transmit the image information down. Robotic arm motion planning and control subsystem: controls the motion trajectory of the robotic arm according to the preset robotic arm path, so as to realize the movement of the robotic arm within the effective range of the information processing unit; Ground information processing system: It acquires the spatial motion information sequence of measurement points on the robotic arm from image information, and obtains the robotic arm motion information and robotic arm dynamic model after data processing based on the spatial motion information sequence, and determines the actual dynamic characteristic parameters of the robotic arm on track; The ground information processing system includes a binocular image ground resolution platform and a dynamic parameter identification platform, wherein: Binocular image ground resolution platform: Acquires image information and obtains spatial motion information sequences of measurement points on the robotic arm; Dynamic parameter identification platform: Based on the spatial motion information sequence, determine the actual end pose information of the robotic arm on the track, compare it with the known robotic arm dynamic model, and obtain the actual dynamic characteristic parameters of the robotic arm on the track. In the aforementioned dynamic parameter identification platform, using the known coordinates of four target points at the end of the robotic arm and the data obtained from observing three arbitrarily selected target points using a binocular camera, a new target coordinate system is created under the end-effector coordinate system. The transformation matrix from the end-effector coordinate system to the target coordinate system is calculated. Using the data obtained from observing three arbitrarily selected target points using a binocular camera, a new target coordinate system is created under the core module body geometric coordinate system. The transformation matrix from the core module body geometric coordinate system to the target coordinate system is calculated. Based on the two transformation matrices obtained, the transformation matrix of the end-effector coordinate system relative to the core module body geometric coordinate system is calculated. Based on the final transformation matrix, the actual end-effector pose information of the target at the end of the robotic arm is transformed and calculated to determine the actual dynamic characteristic parameters of the selected target points in the end-effector coordinate system and the core module body geometric coordinate system, including the vector matrix of the target points in the end-effector coordinate system and the core module body geometric coordinate system, and the relative attitude matrix of the end-effector coordinate system and the core module body geometric coordinate system. In the dynamic parameter identification platform, target points A, B, C, and D at the end of the robotic arm are selected and their coordinates in the end-effector coordinate system E' are determined. A new target coordinate system O1 is established at target point A, and the coordinate transformation matrix from the end-effector coordinate system E' to the target coordinate system O1 is calculated. ; Determine the coordinates of the terminal target points A, B, C, and D in the core module's geometric coordinate system O0, and calculate the coordinate transformation matrix from the target coordinate system O1 to the core module's geometric coordinate system O0. ; Based on the coordinate transformation matrix Coordinate transformation matrix Solve the coordinates of the origin E' of the robotic arm's end effector in the geometric coordinate system O0 of the core module. ; When target points A, B, and C are selected, determine the position vector and coordinate matrix I of the selected target points in the end coordinate system E', and measure the position vector of the selected target points in the core module body geometric coordinate system O0. Based on the calculated coordinates of target points A, B, and C in the geometric coordinate system O0 of the core module, Determine the position vector matrix J; The relative attitude matrix of the end-effector coordinate system with respect to the core module's geometric coordinate system O0 is calculated based on the coordinate matrix I and the position vector matrix J. .

2. The on-orbit parameter identification system for a space robotic arm based on binocular vision according to claim 1, characterized in that: It also includes a watchdog module, which monitors the core voltage of the information processing unit during operation. When the core voltage is lower than the normal operating voltage, the information processing unit is reset and restarted, an error code is generated and sent to the GNC controller.

3. The on-orbit parameter identification system for a space robotic arm based on binocular vision according to claim 1, characterized in that: The information processing unit includes an instruction receiving unit, an image synthesis unit, an exposure control unit, and an image compression unit, wherein: Command receiving unit: Receives control commands from the GNC controller to control the stereo camera and sends exposure control commands to adjust the exposure mode of the stereo camera, including active exposure mode and automatic exposure mode; Image synthesis unit: Receives the measurement image information and measurement noise data output by the binocular camera at the current moment, and synthesizes the current moment data and the compressed measurement image information into a single clear photo; Exposure control unit: Receives control commands from the GNC controller and adjusts the exposure time and exposure gain of the binocular camera; Image compression unit: Receives a single clear photograph sent by the image synthesis unit, performs grayscale correction, compresses the image, and transmits the processed image information to the measurement and control subsystem.

4. The on-orbit parameter identification system for a space robotic arm based on binocular vision according to claim 3, characterized in that: The GNC controller receives ground remote control commands, extracts ground measurement and control information, switches the binocular camera on / off, powers it on, collects and transports data, and calls the robotic arm motion planning and control subsystem to control the on-orbit motion of the robotic arm, including robotic arm movement, stopping, braking, forward movement, and backward movement.

5. The on-orbit parameter identification system for a space robotic arm based on binocular vision according to claim 4, characterized in that: In the binocular image ground solution platform, binocular cameras are used to image non-cooperative targets at different distances to obtain simultaneous stereo image pairs of the targets. Common points are obtained through visual matching of the binocular cameras. The disparity of the binocular cameras and the three-dimensional information of all robotic arm measurement points are calculated to obtain the spatial motion information sequence of the measurement points on the robotic arm.

6. The on-orbit parameter identification system for a space robotic arm based on binocular vision according to claim 5, characterized in that: In the aforementioned binocular image ground resolution platform, the method for measuring the parallax of the binocular camera and the three-dimensional information of all robotic arm measurement points is as follows: To ensure the imaging planes of the left and right cameras of the stereo camera are on the same plane, establish a world coordinate system. Select any point M in the world coordinate system and determine its projection point on the imaging planes of the left and right cameras. , ; The position parameters of the binocular camera are measured by measuring the camera, and the camera position is adjusted so that the optical axis is parallel to the front. Calculate the depth distance from M to the measuring camera and the parallax values ​​of the left and right cameras; Calculate the three-dimensional position of point M in space based on the relative positions and focal lengths of the left and right cameras; Obtain the three-dimensional spatial position of the target point on the robotic arm at any time, and determine the spatial motion information sequence of the measurement point on the robotic arm.

7. The on-orbit parameter identification system for a space robotic arm based on binocular vision according to claim 6, characterized in that: In the binocular camera, the near-field camera has a three-axis accuracy better than 5mm, the far-field camera has a three-axis accuracy better than 2mm, and the maximum output compressed image is no more than 210KB; the refresh rate of the output image is 40f / s; and the image information download time is less than 0.5 hours.

Images