Mobile platform hull cleaning robot positioning method based on self-adaptive federated filter

Abstract

The invention discloses a mobile platform hull cleaning robot positioning method based on a self-adaptive federated filter. The method comprises the following steps: S1, representing the position, speed and attitude information of a cleaning robot by taking an inertial navigation system as a reference; s2, the inertial navigation system outputs a three-dimensional attitude angle, three-axis position information and three-axis speed information; s3, outputting three-dimensional attitude angle information by the electronic compass; s4, acquiring three-dimensional pose information by a speedometer; s5, if the cleaning robot is on the water, a GPS module obtains the current position and speed information; if the robot is underwater, a depth meter collects current depth information; s6, completing fusion of data information of an inertial navigation system, an electronic compass, a speedometer, a depth meter and a GPS module by adopting a federated filter; and S7, obtaining pose and speed information of the cleaning robot relative to the ship body. The method has the advantages of high positioning precision, no error divergence along with time, low operation complexity, small data volume and good real-time performance by adopting a secondary filtering structure of a main filter and a sub-filter of a federated filter.

Description

Technical Field

[0001] The invention relates to the technical field of cleaning robots, and in particular to a positioning method for a moving platform hull cleaning robot based on an adaptive federal filter. Background Art

[0002] In the field of global trade and transportation, ships play a pivotal role. Due to their strong carrying capacity, low freight rates, and strong adaptability, they account for nearly 2 / 3 of the total global freight trade. Because the bottom of the ship is soaked in water for a long time, a large amount of seaweed, barnacles and other debris are often attached. Studies have shown that when the coverage rate of ship bottom fouling reaches 10%, a ship with a speed of 15 knots needs to increase the additional power by 36% to maintain the speed; when the coverage rate is 48%, the maximum additional power required to maintain the speed is 57%. This has caused a huge waste of resources. Therefore, in order to ensure shipping efficiency and save resources, it is very important to clean the ship regularly.

[0003] Compared with the manual cleaning method using divers, using a dedicated hull cleaning robot to clean the hull can not only greatly improve the efficiency of hull cleaning, but also can perform cleaning operations whether the ship is sailing or stopped, while also ensuring the personal safety of relevant personnel. In order to realize the recognition of the position of the hull cleaning robot in the ship and the subsequent recovery of the robot, high-precision positioning of the cleaning robot relative to the hull has become a top priority.

[0004] At present, the navigation technology of underwater robots is mainly divided into inertial navigation, acoustic navigation, and dead reckoning. Among them, inertial navigation is an autonomous navigation system based on the principle of Newtonian mechanics. It can work without relying on any external information, so it has high autonomy and reliability. However, due to the errors of the inertial sensor itself, and the errors will accumulate over time, and the high-precision inertial sensor has a complex structure and is expensive, so the single inertial navigation is only suitable for high-precision navigation in a short time and small range. Acoustic navigation uses the propagation of sound in the water to complete the navigation and positioning function. It needs to be realized with the help of a sound generator and a receiver, so both need to be arranged in advance. Its common classifications are long baseline (LBL), short baseline (SBL), and ultra-short baseline (USBL). This type of navigation method is generally expensive, bulky, and difficult to maintain. It is suitable for large-scale long-distance positioning of large underwater robots. Dead reckoning is to obtain the robot's running speed, posture and other motion information with the help of electronic compass, Doppler velocimeter (DVL) and other equipment, and combine the motion time to calculate the robot's motion distance through integration and other methods, so as to complete the iterative update of the robot's position. This method, like inertial navigation, will have the problem of error accumulation over time, but it has the advantages of low cost, simple algorithm structure, small amount of calculation, etc. Therefore, the present invention is derived from this. Summary of the invention

[0005] In view of at least one of the above-mentioned technical problems, the present invention aims to provide a method for positioning a hull cleaning robot on a moving platform based on an adaptive federated filter, which is suitable for small hull cleaning robots with low cost, low power consumption and easy maintenance. It can provide high-precision positioning relative to the hull (which may be in motion) for the hull cleaning robot, assist the operator to understand the position of the robot relative to the hull in real time, and also facilitate the subsequent recovery of the robot.

[0006] The technical solution of the present invention is:

[0007] The object of the present invention is to provide a positioning method for a moving platform hull cleaning robot based on an adaptive federated filter, comprising:

[0008] S1. A ship navigation system composed of an inertial navigation system and a GPS is arranged at the deployment point of the cleaning robot on the hull. The ship navigation system obtains the position and velocity information of the hull relative to the navigation coordinate system in real time. The hull coordinate system and the navigation coordinate system both take the deployment point as the origin. The positive x-axis of the hull coordinate system points to the bow, the positive y-axis points to the starboard side of the hull, and the positive z-axis points to the top of the hull. The x-axis of the navigation coordinate system points to the geographic north, the y-axis points to the geographic east, and the z-axis points to the sky. The coordinate systems of the sensors of the cleaning robot are defined, and the position, velocity and attitude information of the cleaning robot are represented with the inertial navigation system as a reference.

[0009] S2, the inertial navigation system obtains and outputs the three-dimensional attitude angle, three-axis position information and three-axis speed information of the cleaning robot;

[0010] S3, the electronic compass obtains and outputs the three-dimensional posture angle information of the cleaning robot;

[0011] S4, the odometer obtains the three-dimensional position information of the cleaning robot;

[0012] S5. If the cleaning robot is above water, the depth meter stops working, and the GPS module works and obtains the current position and speed information of the cleaning robot; if the cleaning robot is underwater, the GPS module stops working, and the depth meter collects the current depth information of the cleaning robot;

[0013] S6. Use federal filter to complete the fusion of data information of inertial navigation system, electronic compass, odometer, depth meter and GPS module;

[0014] S7. The federated filter outputs the optimal estimate of the position and velocity information of the cleaning robot relative to the navigation coordinate system, and combines it with the information of the hull coordinate system relative to the navigation coordinate system given by the host computer navigation system to obtain the position and velocity information of the cleaning robot relative to the hull through coordinate transformation.

[0015] Preferably, the odometer is installed on two driving wheels of the cleaning robot. In step S4, the specific steps of the odometer obtaining the three-dimensional posture information of the cleaning robot are as follows:

[0016] Assume that the radius of the driving wheel of the cleaning robot is R, the distance between the left and right driving wheels is L, the forward speed is v, and the speeds of the left and right driving wheels are v respectively. L , v R , the angular velocity of rotation is ω, assuming that the cleaning robot only moves forward and rotates in the heading direction;

[0017] Assume that the position of the cleaning robot in the navigation coordinate system at time k-1 is p k-1 =[x k-1 ,y k-1 ,z k-1 ] T ,attitude

[0018] The direction cosine matrix of the cleaning robot from time k-1 to the current time k for:

[0019]

[0020] Assume the direction cosine matrix of the cleaning robot's own coordinate system and navigation coordinate system at time k-1 Then the direction cosine matrix from the current k-th moment to the navigation coordinate system is for:

[0021]

[0022] Speed ​​of cleaning robot Angular velocity

[0023] The speed and angular velocity vector of the cleaning robot in the navigation coordinate system are:

[0024]

[0025] in,

[0026] The cleaning robot's current position p in the navigation coordinate system at time k k and postureΩ k It is expressed as:

[0027] p k =p k-1 +vnΔt;

[0028] Ω k =Ω k-1 +ω n Δt.

[0029] Preferably, the federal filter includes a main filter and a first sub-filter, a second sub-filter and a third sub-filter. In step S6, the specific steps of using the federal filter to complete the fusion of data information of the inertial navigation system, the electronic compass, the odometer, the depth meter and the GPS module are as follows:

[0030] S61, the first sub-filter completes the fusion of the attitude angle information of the inertial navigation system and the attitude angle information of the electronic compass, and transmits the fused information to the main filter;

[0031] S62, the second sub-filter completes the fusion of the attitude angle and position information of the inertial navigation system and the corresponding information of the odometer, and transmits the fused information to the main filter;

[0032] S63, the third sub-filter completes the fusion of the information of the inertial navigation system and the corresponding information of the depth meter or GPS, and transmits the fused information to the main filter;

[0033] S64, the main filter completes the fusion of information and obtains the optimal error estimate of the federated filter, and then adds the optimal error estimate to the result calculated by the inertial navigation system to output the optimal estimate of the current position, posture and speed of the cleaning robot.

[0034] Preferably, in step S63, when the cleaning robot is underwater, the third sub-filter completes the fusion of the position information of the inertial navigation system and the corresponding information of the depth meter, and transmits the fused information to the main filter;

[0035] When the cleaning robot is on the water, the third sub-filter completes the fusion of the position information and speed information of the inertial navigation system with the corresponding information of the GPS, and transmits the fused information to the main filter.

[0036] Preferably, in step S64, the main filter feeds back the optimal error estimate to each sub-filter at the same time, so that both the main filter and the sub-filter can compensate for their own errors using the optimal error estimate and start the next iteration with the optimal result.

[0037] Preferably, in step S6, the federal filter completes the step of fusing the data information of the inertial navigation system, electronic compass, odometer, depth meter, and GPS by introducing a forgetting factor λ, and combining the error change trend between the output of each sub-filter and the optimal estimate in the previous n moments, to adaptively adjust the information allocation coefficient β of each sub-filter, wherein the size of n is autonomously adjusted in combination with the computational complexity and filtering accuracy.

[0038] Preferably, the forgetting factor λ is introduced, and the specific steps of adaptively adjusting the information allocation coefficient β of each sub-filter in combination with the error variation trend between the output of each sub-filter and the optimal estimate in the previous n moments are as follows:

[0039] Assume that the current moment is k, and calculate the error percentage of each output quantity of each sub-filter in the previous n moments:

[0040]

[0041] Among them, X ki is the output vector of sub-filter i at time k, is the optimal estimation vector of all output components of the federated filter corresponding to sub-filter i at time k, where i is the sequence number of the sub-filter;

[0042] set up is the error percentage of the jth component of the ith sub-filter at time k, and the error percentage vector of the ith sub-filter at time k is recorded as

[0043]

[0044] Among them, s i is the number of output components of sub-filter i, s i The dimension is equal to the observation vector, and the value of the forgetting factor λ ranges from 0 to 1;

[0045] The error output of sub-filter i is:

[0046]

[0047] definition To find the vector The maximum value of all components;

[0048] Calculate the information distribution coefficient of sub-filter i at time k

[0049]

[0050] Wherein, w is the total number of all sub-filters of the federated filter, and i is the sequence number of the sub-filter.

[0051] Preferably, in step S64, an inertial navigation system is first selected as a common positioning reference system of the federated filter, and a state space equation of the inertial navigation system is established, and the state equation of the inertial navigation system is directly defined as the state equation of the main filter. The specific steps are as follows:

[0052] S641, select state variables Eighteen-dimensional vector, where δφ e ,δφ n ,δφ u is the attitude angle error, δv e ,δv n ,δv uis the velocity error, δx, δy, δz are the position errors, ε bx ,ε by ,ε bz is the gyroscope constant drift, ε rx ,ε ry ,ε rz Gyroscope first-order Markov process, Accelerometer constant drift;

[0053] Assume that the state equation of the inertial navigation system in the discrete time domain is:

[0054] Z k =Φ k|k-1 X k-1 +Γ k-1 W k-1 ;

[0055] Φ k|k-1 is the state transfer matrix, Γ k-1 is the system noise driving matrix, W k-1 is the system process noise, and its covariance matrix is ​​Q k-1 , subscripts k and k-1 represent sampling moments, with k being the current moment and k-1 being the previous moment;

[0056] S642: Complete the information distribution between the main filter and each sub-filter in the federated filter as follows:

[0057]

[0058]

[0059] where β k-1 is the information allocation factor at time k-1, β k-1 >0, l represents the total number of sub-filters, where The subscript i of represents the subfilter i, The subscript m indicates the main filter, P k-1 is the state covariance matrix at time k-1, where represents the state covariance matrix of sub-filter i, represents the state covariance matrix of the main filter, Q k-1 is the process noise covariance matrix at time k-1, where represents the process noise covariance matrix of sub-filter i, represents the process noise covariance matrix of the main filter; the subscript g represents the optimal estimate of the main filter output at time k-1;

[0060] S643, local filtering of each sub-filter, and calculation of the measurement equation of each sub-filter:

[0061]

[0062] in, is the observation matrix, is the observation noise, and its covariance matrix is ​​set to i is the serial number of the sub-filter;

[0063] S644, then complete the status update of the main and sub filters:

[0064] The status of each sub-filter is updated:

[0065]

[0066] Status update of the main filter:

[0067]

[0068] S645, completing the fusion of the main filter and each sub-filter information:

[0069]

[0070] S646, calculating the information allocation coefficients of the main filter and each sub-filter:

[0071]

[0072] S647, the federated filter optimal estimation completes the compensation of the inertial navigation system output, obtains the optimal output of the cleaning robot's posture and speed information, and at the same time estimates the optimal error Optimal estimation of the state covariance matrix Information distribution coefficient Feedback to the corresponding main sub-filter.

[0073] Preferably, the cleaning robot comprises:

[0074] The housing comprises a shell body which is hollow inside to form a mounting cavity and has an open top, and a top plate which covers the open top of the shell body, wherein the mounting cavity comprises a first cavity body at the front side and a second cavity body at the rear side;

[0075] A sealed cabin, which is arranged at the second cavity and has a sealed cavity therein;

[0076] Two driving wheels are arranged oppositely at two sides of the front end of the shell and the inner ends of the two driving wheels extend into the first cavity respectively;

[0077] A GPS module, comprising a GPS sensor disposed in the sealed cavity and a GPS antenna connected to the GPS sensor and partially disposed in the first cavity and partially extending to the outer side of the upper end of the top plate;

[0078] Two depth gauges are arranged at intervals on the outer side of the front end of the housing;

[0079] A mounting frame, which is mounted on the outer side of the front end of the housing with a gap formed therebetween;

[0080] A visual sensor is mounted at the middle of the front end of the mounting frame and partially extends to the outside of the front end of the mounting frame and partially is placed in the interval;

[0081] A cleaning roller brush, which rotates horizontally and can be lifted vertically, is arranged at the lower outer side of the front end of the housing and has a baffle fixed to the mounting frame at the upper front end thereof;

[0082] A main control module, which is arranged in the sealed cavity;

[0083] An inertial measurement device, which is arranged at one side of the sealed cavity and close to the front side of the sealed cavity;

[0084] Two odometers, respectively mounted on one end of the two driving wheels extending into the first cavity;

[0085] An electronic compass, which is arranged at the rear side of the sealed cavity and on the same side as the inertial navigation system;

[0086] A wheel motor driver, which is arranged at the rear side of the sealed cavity;

[0087] The visual sensor, inertial navigation system, odometer, electronic compass and wheel motor driver are all electrically connected to the main control module.

[0088] Preferably, the cleaning robot further comprises:

[0089] Ultrasonic obstacle avoidance sensors are arranged on the front end face, two side faces and the rear ends of the convex plates protruding outwards on the two outer sides of the front end of the shell and used for respectively sealing the drive motors corresponding to the two drive wheels;

[0090] Two water leakage sensors are disposed on the top plate and the bottom plate in the sealed cabin opposite to each other, and the two water leakage sensors are located between the upper and lower parts of the main control module;

[0091] A backup power supply, which is arranged at one side of the sealed cavity and between the inertial navigation system and the electronic compass;

[0092] A universal wheel, which is arranged at the rear end of the housing;

[0093] The lighting device is arranged on the mounting frame and distributed on both sides of the visual sensor.

[0094] Compared with the prior art, the advantages of the present invention are:

[0095] The present invention is based on the positioning method of the dynamic platform hull cleaning robot of the adaptive federal filter, compared with other positioning schemes using fiber optic gyroscopes and Doppler velocimeters, these devices have high positioning accuracy, but are expensive, large in size and high in power consumption. In comparison, the method has low cost, small in size, low in power consumption, convenient maintenance, high positioning accuracy, and the error does not diverge over time. At the same time, the secondary filtering structure of the main sub-filter of the federal filter is adopted, the computational complexity is low, the data volume is small, the requirements for the master control performance of the lower machine are low, and the real-time performance of the system is also good. At the same time, the adaptive federal filter can autonomously control the opening and closing of each sub-filter according to the actual situation as needed, which also makes it possible that even if a certain sensor cannot work due to a fault, the system can still work normally under the condition of sacrificing a certain accuracy, thereby improving the robustness of the system. The adaptive federal filter based on the forgetting factor adjustment not only utilizes the optimal estimation output of the filter, but also combines the previous historical information, increases the source of information, and improves the accuracy of adjusting the information allocation coefficient. A variety of sensors are used for information fusion, so the accuracy requirements for a single sensor are relatively low, and the advantages of a variety of sensors can be combined at the same time to make up for the positioning error generated by a single sensor, effectively improving the positioning accuracy. The type and number of sensors for fusion can be switched freely according to the environment in which the robot is located. GPS cannot be used underwater, but the data of the depth meter can be added. On the water surface, the depth meter loses its function, so the depth meter can be turned off and GPS can be turned on. This free switching mode not only enhances the robustness of the system, but also effectively reduces the total power consumption of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0097] Figure 1 It is a structural schematic diagram of a moving platform hull cleaning robot based on an adaptive federated filter according to an embodiment of the present invention;

[0098] Figure 2 It is a schematic structural diagram of a dynamic platform hull cleaning machine based on an adaptive federated filter in an embodiment of the present invention, in which a top plate and a sealed hatch cover are omitted;

[0099] Figure 3 It is a schematic diagram of the hardware structure flow of a dynamic platform hull cleaning robot based on an adaptive federated filter according to an embodiment of the present invention;

[0100] Figure 4 This is a navigation and positioning flow chart of a method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to an embodiment of the present invention;

[0101] Figure 5 A navigation filter switching flow chart of a positioning method for a moving platform hull cleaning robot based on an adaptive federated filter according to an embodiment of the present invention;

[0102] Figure 6 This is a filtering flow chart of a federal filter (underwater) of a method for positioning a moving platform hull cleaning robot based on an adaptive federal filter according to an embodiment of the present invention;

[0103] Figure 7 The present invention is a filtering flow chart of a federal filter (above water) of a method for positioning a moving platform hull cleaning robot based on an adaptive federal filter according to an embodiment of the present invention.

[0104] Among them: 10, outer shell; 11, shell; 12, top plate; 13, first cavity; 14, sealed cabin; 15, convex plate; 16, mounting frame; 20, driving wheel; 21, universal wheel; 30, cleaning roller brush; 31, baffle; 40, inertial measurement device; 50, GPS module; 51, GPS sensor; 52, GPS antenna; 60, odometer; 70, depth gauge; 80, electronic compass; 90, main control module; 100, visual sensor; 110, ultrasonic obstacle avoidance sensor; 120, AC-DC power supply; 130, water leakage sensor; 140, umbilical cable; 150, power carrier module; 160, lighting device; 170, wheel motor driver; 180, backup power supply. DETAILED DESCRIPTION

[0105] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0106] See also Figure 1 to Figure 2 The dynamic platform hull cleaning robot based on the adaptive federal filter of the embodiment of the present invention includes a shell 10, a sealed cabin 14, a driving wheel 20, a cleaning roller brush 30, a wheel motor driver 170, a main control module 90, an AC-DC power supply 120, a power carrier module and main functional modules such as an inertial measurement device 40, an electronic compass 80, an odometer 60, a depth gauge 70, a GPS module 50 and other auxiliary modules such as a visual sensor 100, an ultrasonic obstacle avoidance sensor 110, a water leakage sensor 130, a backup power supply 180, and an umbilical cable 140 (used for power transmission, communication and load-bearing of the robot above and below the water).

[0107] The cleaning robot of the embodiment of the present invention is a two-wheel differential structure, two odometers 60 are respectively installed on two driving wheels 20, ultrasonic obstacle avoidance sensors 110 are installed around the cleaning robot, a cleaning roller brush 30, a camera (visual sensor 100), a searchlight (illumination device 160) and a depth meter 70 are installed in the front, and a GPS antenna 52 and an umbilical cable 140 are installed above the cleaning robot. The umbilical cable 140 is used to complete the transmission of power, control signals and video signals of the cleaning robot. The inertial measurement device 40, the electronic compass 80 and the main control module 90 are placed in the sealed cabin 14.

[0108] More specifically, the housing 10 includes a shell 11 which is hollow inside to form a mounting cavity and has an open top, and a top plate 12 which is covered on the open top of the shell 11, and the mounting cavity includes a first cavity 13 on the front side and a second cavity on the rear side. The sealed cabin 14 is arranged at the second cavity and has a sealed cavity therein. The two driving wheels 20 are arranged relatively on both sides of the front end of the shell 11, and the inner ends of the two driving wheels 20 extend into the first cavity 13 respectively. The GPS module 50 includes a GPS sensor 51 arranged in the sealed cavity and a GPS antenna 52 which is connected to the GPS sensor 51 and is partially placed in the first cavity 13 and partially extends to the outer side of the upper end of the top plate 12. Two depth gauges 70 are arranged at intervals on the outer side of the front end of the shell 11, and are used to assist in positioning after the robot leaves the water surface, that is, when it is on the water. The mounting frame 16 is roughly U-shaped, mounted on the outer side of the front end of the shell 11, and a gap is formed between the two, and is used for the installation of the visual sensor 100, the lighting device 160, and the ultrasonic obstacle avoidance sensor 110. The visual sensor 100 is installed in the middle position of the front end of the mounting frame 16 and partially extends to the outside of the front end of the mounting frame 16, and partially is placed in the interval and does not contact the front outer end surface of the shell 11, and the visual sensor 100 avoids the depth gauge 70 on the front outer side of the shell 11, and is used to capture images of the environment in which the cleaning robot is located. The cleaning roller brush 30 is horizontally rotatable and can be vertically lifted and lowered at the lower front outer side of the shell 11, and a baffle 31 fixed to the mounting frame 16 is provided at its upper front. The rotating shaft of the cleaning roller brush 30 is connected to the suspension on both sides of the front end of the shell 11, so that the cleaning roller brush 30 can be lifted and lowered according to the thickness of the dirt at the cleaning position and elastically pressed against the dirt to clean the dirt. The main control module 90 is arranged in the sealed cavity (roughly in the middle and front end position). The main control module 90 includes two main control circuit boards, one for image processing, that is, Figure 3 The main control 2 shown is used to control Figure 3The main control 1 shown. The inertial measurement device 40 is arranged on one side of the sealed cavity and close to the front side of the sealed cavity, and can complete the measurement of the cleaning robot's own attitude angle and three-axis acceleration. The two odometers 60 are respectively installed on one end of the two driving wheels 20 extending into the first cavity 13, and are used to complete the collection and measurement of the moving speed and angular velocity of the cleaning robot. The electronic compass 80 is arranged on the rear side of the sealed cavity and is on the same side as the inertial measurement device 40. It can complete the measurement of the three-dimensional attitude angle of the cleaning robot according to the geomagnetic field, and its measurement error will not accumulate and diverge over time. The wheel motor driver 170 is arranged on the rear side of the sealed cavity, and is used to drive the wheels to rotate and drive the cleaning robot to walk. The visual sensor 100, the inertial measurement device 40, the odometer 60, the electronic compass 80 and the wheel motor driver 170 are all electrically connected to the main control circuit board. More specifically, as Figure 3 As shown, the inertial measurement device 40, the odometer 60, the electronic compass 80 and the wheel motor driver 170 are connected to the main control 1 via serial communication, and the visual sensor 100 is electrically connected to the main control 2. The ultrasonic obstacle avoidance sensor 110 is arranged on the front end face of the shell 11, the two side faces and the rear end of the convex plates 15 protruding outward on the two outer sides of the front end of the shell 11 for respectively sealing the drive motors corresponding to the two drive wheels 20, so as to prevent the cleaning robot from colliding with foreign objects while walking. The ultrasonic obstacle avoidance sensor 110 is also connected to the main control 1 via serial communication. The two water leakage sensors 130 are relatively arranged on the top plate 12 and the bottom plate in the sealed cabin 14, and the two water leakage sensors 130 are located between the upper and lower parts of the main control circuit board for detecting the sealing. The AC-DC power supply 120 is arranged on one side of the sealed cavity, that is, Figure 2 The backup power supply 180 is located at the middle position on the left side of the sealed cavity. Figure 2 The power carrier module is located on one side of the sealed cavity, i.e., between the inertial measurement device 40 and the electronic compass 80, and is used to provide power to various electrical components in the cleaning robot. Figure 2 The upper left side is shown and is electrically connected to the main control 1. The space of the sealed cabin 14 is reasonably utilized to seal each sensor system and other electrical components in the sealed cavity. The universal wheel 21 is arranged at the rear end of the housing 10 to assist the driving wheel 20 so that the cleaning robot can rotate freely and improve the flexibility of the cleaning robot. The lighting device 160 is arranged on the mounting frame 16 and distributed on both sides of the visual sensor 100 to provide lighting for the cleaning robot, so that the cleaning robot can walk in a darker environment and take clear images.

[0109] like Figure 2As shown, the ship navigation system can transmit the coordinates of the ship relative to the navigation coordinate system to the cleaning robot in real time through the communication cable. In this way, combined with the coordinates of the cleaning robot itself relative to the navigation coordinate system, the coordinates of the cleaning robot relative to the ship in the navigation coordinate system can be obtained. After a coordinate transformation, the coordinates of the cleaning robot relative to the ship in the hull coordinate system can be obtained. By arranging two sets of navigation systems, even if the ship is moving, the position information of the cleaning robot relative to the carrier ship can still be obtained.

[0110] See also Figures 3 to 7 , the positioning method of the cleaning robot according to the embodiment of the present invention comprises:

[0111] S1. A ship navigation system composed of an inertial navigation system and a GPS is arranged at the deployment point of the cleaning robot on the hull. The ship navigation system obtains the position and velocity information of the hull relative to the navigation coordinate system in real time. The hull coordinate system and the navigation coordinate system both take the deployment point as the origin. The positive x-axis of the hull coordinate system points to the bow, the positive y-axis points to the starboard side of the hull, and the positive z-axis points to the top of the hull. The x-axis of the navigation coordinate system points to the geographic north, the y-axis points to the geographic east, and the z-axis points to the sky. The coordinate systems of the sensors of the cleaning robot are defined, and the position, velocity and attitude information of the cleaning robot are represented with the inertial navigation system as a reference.

[0112] S2, the inertial navigation system obtains and outputs the three-dimensional attitude angle, three-axis position information and three-axis speed information of the cleaning robot;

[0113] S3, the electronic compass obtains and outputs the three-dimensional posture angle information of the cleaning robot;

[0114] S4, the odometer obtains the three-dimensional position information of the cleaning robot;

[0115] S5. If the cleaning robot is above water, the depth meter stops working, and the GPS module works and obtains the current position and speed information of the cleaning robot; if the cleaning robot is underwater, the GPS module stops working, and the depth meter collects the current depth information of the cleaning robot;

[0116] S6. Use federal filter to complete the fusion of data information of inertial navigation system, electronic compass, odometer, depth meter and GPS module;

[0117] S7. The federated filter outputs the optimal estimate of the position and velocity information of the cleaning robot relative to the navigation coordinate system, and combines it with the information of the hull coordinate system relative to the navigation coordinate system given by the host computer navigation system to obtain the position and velocity information of the cleaning robot relative to the hull through coordinate transformation.

[0118] Preferably, the odometer is installed on two driving wheels of the cleaning robot. In step S4, the specific steps of the odometer obtaining the three-dimensional posture information of the cleaning robot are as follows:

[0119] Assume that the radius of the driving wheel of the cleaning robot is R, the distance between the left and right driving wheels is L, the forward speed is v, and the speeds of the left and right driving wheels are v respectively. L , v R , the angular velocity of rotation is ω, assuming that the cleaning robot only moves forward and rotates in the heading direction;

[0120] Assume that the position of the cleaning robot in the navigation coordinate system at time k-1 is p k-1 =[x k-1 ,y k-1 ,z k-1 ] T ,attitude

[0121] The rotation matrix of the cleaning robot from its own coordinate system at time k-1 to its own coordinate system at the current time k is the direction cosine matrix for:

[0122]

[0123] Assume the direction cosine matrix of the cleaning robot's own coordinate system and navigation coordinate system at time k-1 Then the direction cosine matrix from the current k-th moment to the navigation coordinate system is for:

[0124]

[0125] Speed ​​of cleaning robot Angular velocity

[0126] The speed and angular velocity vector of the cleaning robot in the navigation coordinate system are:

[0127]

[0128] in,

[0129] The cleaning robot's current position p in the navigation coordinate system at time k k and postureΩ k It is expressed as (it should be noted that when the sampling time is short enough and the speed and angular velocity of the cleaning robot change slowly, the following discretization formula can be used to approximate it):

[0130] p k =p k-1 +vnΔt;

[0131] Ω k =Ω k-1 +ω n Δt.

[0132] It should be noted that, in the above direction matrix, b (abbreviation of body) represents the body coordinate system, and n (abbreviation of navigation) represents the navigation coordinate system.

[0133] Preferably, the federal filter includes a main filter and a first sub-filter (corresponding to sub-filter 1 in the figure), a second sub-filter (corresponding to sub-filter 2 in the figure) and a third sub-filter (corresponding to sub-filter 3 in the figure). In step S4, the specific steps for the federal filter to complete the fusion of data information of the inertial navigation system, the electronic compass, the odometer, the depth meter and the GPS are as follows:

[0134] S61, the first sub-filter completes the fusion of the attitude angle information of the inertial navigation system and the attitude angle information of the electronic compass, and transmits the fused information to the main filter;

[0135] S62, the second sub-filter completes the fusion of the attitude angle and position information of the inertial navigation system and the corresponding information of the odometer, and transmits the fused information to the main filter;

[0136] S63. The third sub-filter completes the fusion of the information of the inertial navigation system and the corresponding information of the depth meter or GPS, and passes the fused information to the main filter; more specifically, when the cleaning robot is underwater, the third sub-filter completes the fusion of the position information of the inertial navigation system and the corresponding information of the depth meter, and passes the fused information to the main filter; when the cleaning robot is above water, the third sub-filter completes the fusion of the position information and speed information of the inertial navigation system and the corresponding information of the GPS, and passes the fused information to the main filter.

[0137] S64, the main filter completes the fusion of information, obtains the optimal error estimate of the federated filter, and then adds the optimal error estimate to the result calculated by the inertial navigation system to output the optimal estimate of the current position, posture and speed of the cleaning robot. Preferably, the main filter feeds back the optimal error estimate to each sub-filter at the same time, so that the main and sub-filters can use the optimal error estimate to compensate for their own errors and start the next iteration with the optimal result.

[0138] For the federated filtering algorithm, the optimization and adaptive selection of the information allocation coefficient β is the key to improving the performance of the federated filter. Therefore, this design proposes an adaptive federated filtering algorithm based on forgetting factor adjustment. By introducing the forgetting factor λ and combining the error change trend between the output of each sub-filter and the system optimal estimate in the previous n moments, the information allocation coefficient β of each sub-filter is adaptively adjusted. The rationality and robustness of β allocation are improved by introducing information in the time dimension.

[0139] Specifically, in step S6, the federal filter completes the step of fusing the data information of the inertial navigation system, electronic compass, odometer, depth meter, and GPS by introducing a forgetting factor λ, and combining the error change trend between the output of each sub-filter and the optimal estimate in the previous n moments, and adaptively adjusting the information allocation coefficient β of each sub-filter, wherein the size of n is selected based on the performance of the main control and the real-time requirements of the navigation solution, and is autonomously adjusted in combination with the computational complexity and filtering accuracy. In this application, n can generally be selected as 10-20.

[0140] More specifically, the forgetting factor λ is introduced, and the specific steps of adaptively adjusting the information allocation coefficient β of each sub-filter are as follows, combining the error change trend between the output of each sub-filter and the optimal estimate in the previous n moments:

[0141] Assume that the current moment is k, and calculate the error percentage of each output quantity of each sub-filter in the previous n moments:

[0142]

[0143] Among them, X ki is the output vector of sub-filter i at time k, is the optimal estimation vector of all output components of sub-filter i corresponding to the federated filter at time k, where i is the sequence number of the sub-filter; in the embodiment of the present invention, the number of sub-filters is three, that is, the value of i is 1, 2, and 3.

[0144] Because the federated filter generally outputs information such as position, attitude, and speed, and a sub-filter generally only outputs one or several of them. For example, the sub-filter system corresponding to the electronic compass only outputs attitude information. Therefore, all outputs of the federated filter are generally not used when calculating the output error percentage.

[0145] The forgetting factor λ is introduced to adjust the importance of the error percentage at different times to the error percentage of the final output of the sub-filter. The closer to the current moment, the heavier the weight should be. is the error percentage of the jth component of the ith sub-filter at time k, and the error percentage vector of the ith sub-filter at time k is recorded as

[0146]

[0147] Among them, s i is the number of output components of sub-filter i, s i The dimension is equal to the observation vector, and the value of the forgetting factor λ ranges from 0 to 1;

[0148] The error output of sub-filter i is:

[0149]

[0150] definition To find the vector The maximum value of all components.

[0151] By calculating the error percentage between the sub-filter output and the optimal estimate output in the previous n moments, the degree of deviation of the sub-filter output from the optimal value is measured, so as to judge the accuracy of the sub-filter itself. The larger the error percentage of the sub-filter, the poorer its accuracy, and the smaller its information allocation coefficient should be, that is, the size of the sub-filter error deviation is inversely proportional to the coefficient of system information allocation.

[0152] Calculate the information distribution coefficient of sub-filter i at time k

[0153]

[0154] Wherein, w is the total number of all sub-filters of the federated filter, and i is the sequence number of the sub-filter.

[0155] In the embodiment of the present invention, the value range of the forgetting factor λ is between 0 and 1. The closer it is to 1, the less obvious the forgetting effect is. When λ is 1, it indicates that the error percentage at each moment is of the same importance in the final error percentage of the sub-filter. Conversely, the closer it is to 0, the more significant the forgetting effect is. When λ is 0, the information allocation coefficient becomes a constant value allocated at the initial moment of the system. In the embodiment of the present invention, the forgetting factor λ is preferably 0.98.

[0156] Preferably, in step S64, an inertial navigation system is first selected as a common positioning reference system of the federated filter, and a state space equation of the inertial navigation system is established, and the state equation of the inertial navigation system is directly defined as the state equation of the main filter. The specific steps are as follows:

[0157] S641, select state variables Eighteen-dimensional vector, where δφ e ,δφ n ,δφ u is the attitude angle error, δv e ,δv n ,δvu is the velocity error, δx, δy, δz are the position errors, ε bx ,ε by ,ε bz is the gyroscope constant drift, ε rx ,ε ry ,ε rz Gyroscope first-order Markov process, Accelerometer constant drift;

[0158] Assume that the state equation of the inertial navigation system in the discrete time domain is:

[0159] X k =Φ k|k-1 X k-1 +Γ k-1 W k-1 ;

[0160] Φ k|k-1 is the state transfer matrix, Γ k-1 is the system noise driving matrix, W k-1 is the system process noise, and its covariance matrix is ​​Q k-1 , subscripts k and k-1 represent sampling moments, with k being the current moment and k-1 being the previous moment;

[0161] S642: Complete the information distribution between the main filter and each sub-filter in the federated filter as follows:

[0162]

[0163] where β k-1 is the information allocation factor at time k-1, β k-1 >0, l represents the total number of sub-filters. In the embodiment of the present invention, l=3, where The subscript i of represents the subfilter i, The subscript m indicates the main filter, P k-1 is the state covariance matrix at time k-1, where represents the state covariance matrix of sub-filter i, represents the state covariance matrix of the main filter, Q k-1 is the process noise covariance matrix at time k-1, where represents the process noise covariance matrix of sub-filter i, represents the process noise covariance matrix of the main filter; the subscript g represents the optimal estimate of the main filter output at time k-1;

[0164] S643, local filtering of each sub-filter, calculating the measurement equation of each sub-filter i:

[0165]

[0166] in, is the observation matrix, is the observation noise, and its covariance matrix is ​​set to i is the serial number of the sub-filter;

[0167] It should be noted that the state equations of the main and sub-filters are both state equations of the inertial navigation system, among which only the sub-filters have measurement equations, and the main filter is responsible for time update and information fusion.

[0168] Specifically, the measurement equation of sub-filter 1 is, selecting the error of attitude angle and position output of odometer and inertial navigation system as the observed quantity:

[0169]

[0170] is the observation matrix, is the observation noise, and its covariance matrix is ​​set to

[0171] The measurement equation of sub-filter 1 is, the error of attitude angle and position output of odometer and inertial navigation system is selected as the observed quantity:

[0172]

[0173] is the observation matrix, is the observation noise, and its covariance matrix is ​​set to

[0174] If the cleaning robot is underwater, sub-filter 3, i.e., the third filter, is a fusion of the depth information of the depth meter and the z-direction output information of the inertial navigation system, specifically:

[0175] The measurement equation of the (underwater) sub-filter 3 is to select the error of the position output of the depth meter and the inertial navigation system in the z-axis direction as the observation quantity:

[0176]

[0177] is the observation matrix, is the observation noise, and its covariance matrix is ​​set to

[0178] If it is on water, sub-filter 3 is the fusion of GPS position and speed information and inertial navigation corresponding information, specifically:

[0179] The measurement equation of the (aquatic) sub-filter 3 is to select the error of the position and velocity output of the GPS and inertial navigation system as the observed quantity:

[0180]

[0181] is the observation matrix, is the observation noise, and its covariance matrix is ​​set to

[0182] S644, then complete the status update of the main and sub filters:

[0183] The status of each sub-filter is updated:

[0184]

[0185]

[0186] Status update of the main filter:

[0187]

[0188] S645, completing the fusion of the main filter and each sub-filter information:

[0189]

[0190] S646, calculating the information allocation coefficients of the main filter and each sub-filter:

[0191]

[0192] S647, the federated filter optimal estimation completes the compensation of the inertial navigation system output, obtains the optimal output of the cleaning robot's posture and speed information, and at the same time estimates the optimal error Optimal estimation of the state covariance matrix Information distribution coefficient Feedback to the corresponding main sub-filter.

[0193] Compared with other positioning solutions that use fiber optic gyroscopes and Doppler velocimeters, these devices have high positioning accuracy, but are expensive, large in size and high in power consumption. In comparison, this method has low cost, small size, low power consumption, easy maintenance, high positioning accuracy, and the error does not diverge over time. At the same time, the secondary filtering structure of the main sub-filter of the federal filter is adopted, which has low computational complexity, small data volume, low requirements on the master control performance of the lower computer, and also makes the system real-time. At the same time, the adaptive federal filter can autonomously control the opening and closing of each sub-filter according to needs and actual conditions. This also means that even if a sensor fails to work due to a fault, the system can still work normally at the expense of a certain degree of accuracy, thereby improving the robustness of the system.

[0194] The adaptive federated filter based on forgetting factor adjustment not only utilizes the optimal estimation output of the filter, but also combines the previous historical information, thereby increasing the source of information and improving the accuracy of adjusting the information distribution coefficient.

[0195] This solution uses multiple sensors for information fusion, so the accuracy requirements for a single sensor are relatively low. At the same time, it can combine the advantages of multiple sensors to compensate for the positioning error caused by a single sensor, effectively improving the positioning accuracy.

[0196] This solution can freely switch the type and number of sensors for fusion according to the environment in which the robot is located. GPS cannot be used underwater, but the data of the depth meter can be added. On the water surface, the depth meter loses its function, so the depth meter can be turned off and GPS can be turned on. This free switching mode not only enhances the robustness of the system, but also effectively reduces the total power consumption of the system.

[0197] This solution adopts the form of two-stage filtering. Therefore, even if a sensor fails and cannot work normally, the system can still operate normally by simply cutting off its path in the main filter. Each sub-filter operates independently and does not interfere with each other. At the same time, it effectively reduces the amount of data and the complexity of calculations, reduces the pressure on the system master control, and improves the real-time performance of the system.

[0198] This solution introduces a dual positioning system, that is, there are positioning systems on the hull and the cleaning robot, so that the robot can still accurately locate the relative position of the carrier when the hull is in motion.

[0199] It should be understood that the above specific embodiments of the present invention are only used to illustrate or explain the principles of the present invention, and do not constitute a limitation of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc. made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. In addition, the appended claims of the present invention are intended to cover all changes and modifications that fall within the scope and boundaries of the appended claims, or the equivalent forms of such scope and boundaries.

Claims

1. A method for positioning a moving platform hull cleaning robot based on an adaptive federated filter, characterized in that: include: S1. A ship navigation system composed of an inertial navigation system and a GPS is arranged at the deployment point of the cleaning robot on the hull. The ship navigation system obtains the position and velocity information of the hull relative to the navigation coordinate system in real time. The hull coordinate system and the navigation coordinate system both take the deployment point as the origin. The positive x-axis of the hull coordinate system points to the bow, the positive y-axis points to the starboard side of the hull, and the positive z-axis points to the top of the hull. The x-axis of the navigation coordinate system points to the geographic north, the y-axis points to the geographic east, and the z-axis points to the sky. The coordinate systems of the sensors of the cleaning robot are defined, and the position, velocity and attitude information of the cleaning robot are represented with the inertial navigation system as a reference. S2, the inertial navigation system obtains and outputs the three-dimensional attitude angle, three-axis position information and three-axis speed information of the cleaning robot; S3, the electronic compass obtains and outputs the three-dimensional posture angle information of the cleaning robot; S4, the odometer obtains the three-dimensional position information of the cleaning robot; S5. If the cleaning robot is above water, the depth meter stops working, and the GPS module works and obtains the current position and speed information of the cleaning robot; if the cleaning robot is underwater, the GPS module stops working, and the depth meter collects the current depth information of the cleaning robot; S6. Use federal filter to complete the fusion of data information of inertial navigation system, electronic compass, odometer, depth meter and GPS module; S7. The federated filter outputs the optimal estimate of the position and velocity information of the cleaning robot relative to the navigation coordinate system, and combines it with the information of the hull coordinate system relative to the navigation coordinate system given by the host computer navigation system to obtain the position and velocity information of the cleaning robot relative to the hull through coordinate transformation.

2. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to claim 1, characterized in that: The odometer is installed on the two driving wheels of the cleaning robot. In step S4, the specific steps of the odometer obtaining the three-dimensional posture information of the cleaning robot are as follows: Assume that the radius of the driving wheel of the cleaning robot is R, the distance between the left and right driving wheels is L, the forward speed is v, and the speeds of the left and right driving wheels are v respectively. L , v R , the angular velocity of rotation is ω, assuming that the cleaning robot only moves forward and rotates in the heading direction; Assume that the position of the cleaning robot in the navigation coordinate system at time k-1 is p k-1 =[x k-1 ,y k-1 ,z k-1 ] T ,attitude The direction cosine matrix of the cleaning robot from time k-1 to the current time k for: Assume the direction cosine matrix of the cleaning robot's own coordinate system and navigation coordinate system at time k-1 Then the direction cosine matrix from the current k-th moment to the navigation coordinate system is for: Speed ​​of cleaning robot Angular velocity The speed and angular velocity vector of the cleaning robot in the navigation coordinate system are: in, The cleaning robot's current position p in the navigation coordinate system at time k k and postureΩ k It is expressed as: p k =p k-1 +v n Δt; Oh k =Oh k-1 +oh n Δt.

3. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to claim 1, characterized in that: The federal filter includes a main filter and a first sub-filter, a second sub-filter and a third sub-filter. In step S6, the specific steps of using the federal filter to complete the fusion of data information of the inertial navigation system, the electronic compass, the odometer, the depth meter and the GPS module are as follows: S61, the first sub-filter completes the fusion of the attitude angle information of the inertial navigation system and the attitude angle information of the electronic compass, and transmits the fused information to the main filter; S62, the second sub-filter completes the fusion of the attitude angle and position information of the inertial navigation system and the corresponding information of the odometer, and transmits the fused information to the main filter; S63, the third sub-filter completes the fusion of the information of the inertial navigation system and the corresponding information of the depth meter or GPS, and transmits the fused information to the main filter; S64, the main filter completes the fusion of information and obtains the optimal error estimate of the federal filter, and then adds the optimal error estimate to the result calculated by the inertial navigation system to output the optimal estimate of the current position, posture and speed of the cleaning robot.

4. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to claim 3, characterized in that: In step S63, when the cleaning robot is underwater, the third sub-filter completes the fusion of the position information of the inertial navigation system and the corresponding information of the depth meter, and transmits the fused information to the main filter; When the cleaning robot is on the water, the third sub-filter completes the fusion of the position information and speed information of the inertial navigation system with the corresponding information of the GPS, and transmits the fused information to the main filter.

5. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to claim 3, characterized in that: In step S64, the main filter feeds back the optimal error estimate to each sub-filter at the same time, so that both the main filter and the sub-filter can use the optimal error estimate to compensate for their own errors and start the next iteration with the optimal result.

6. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to claim 3, characterized in that: In step S6, the federal filter completes the step of integrating the data information of the inertial navigation system, electronic compass, odometer, depth meter, and GPS by introducing the forgetting factor λ and combining the error change trend between the output of each sub-filter and the optimal estimate in the previous n moments to adaptively adjust the information allocation coefficient β of each sub-filter, where the size of n can be adjusted independently in combination with the computational complexity and filtering accuracy.

7. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to claim 6, characterized in that: The specific steps of introducing the forgetting factor λ and combining the error change trend between the output of each sub-filter and the optimal estimate in the previous n moments to adaptively adjust the information allocation coefficient β of each sub-filter are as follows: Assume that the current moment is k, and calculate the error percentage of each output quantity of each sub-filter in the previous n moments: Among them, X ki is the output vector of sub-filter i at time k, is the optimal estimation vector of all output components of the federated filter corresponding to sub-filter i at time k, where i is the sequence number of the sub-filter; set up is the error percentage of the jth component of the ith sub-filter at time k, and the error percentage vector of the ith sub-filter at time k is recorded as Among them, s i is the number of output components of sub-filter i, s i The dimension is equal to the observation vector, and the value of the forgetting factor λ ranges from 0 to 1; The error output of sub-filter i is: definition To find the vector The maximum value of all components; Calculate the information distribution coefficient of sub-filter i at time k Wherein, w is the total number of all sub-filters of the federated filter, and i is the sequence number of the sub-filter.

8. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to claim 3, characterized in that: In step S64, the inertial navigation system is first selected as the common positioning reference system of the federated filter, and the state space equation of the inertial navigation system is established, and it is directly defined as the state equation of the main filter. The specific steps are as follows: S641, select state variables Eighteen-dimensional vector, where δφ e ,δφ n ,δφ u is the attitude angle error, δv e ,δv n ,δv n is the velocity error, δx, δy, δz are the position errors, ε bx ,ε by ,ε bz is the gyroscope constant drift, ε rx ,ε ry ,ε rz Gyroscope first-order Markov process, Accelerometer constant drift; Assume that the state equation of the inertial navigation system in the discrete time domain is: X k =Φ k|k-1 X k-1 +C k-1 W k-1 ; Φ k|k-1 is the state transfer matrix, Γ k-1 is the system noise driving matrix, W k-1 is the system process noise, and its covariance matrix is ​​Q k-1 , subscripts k and k-1 represent sampling moments, with k being the current moment and k-1 being the previous moment; S642: Complete the information distribution between the main filter and each sub-filter in the federated filter as follows: where β k-1 is the information allocation factor at time k-1, β k-1 >0, l represents the total number of sub-filters, where The subscript i of represents the subfilter i, The subscript m indicates the main filter, P k-1 is the state covariance matrix at time k-1, where represents the state covariance matrix of sub-filter i, represents the state covariance matrix of the main filter, Q k-1 is the process noise covariance matrix at time k-1, where represents the process noise covariance matrix of sub-filter i, represents the process noise covariance matrix of the main filter; the subscript g represents the optimal estimate of the main filter output at time k-1; S643, local filtering of each sub-filter, and calculation of the measurement equation of each sub-filter: in, is the observation matrix, is the observation noise, and its covariance matrix is ​​set to i is the serial number of the sub-filter; S644, then complete the status update of the main and sub filters: The status of each sub-filter is updated: Status update of the main filter: S645, completing the fusion of the main filter and each sub-filter information: S646, calculating the information allocation coefficients of the main filter and each sub-filter: S647, the federated filter optimal estimation completes the compensation of the inertial navigation system output, obtains the optimal output of the cleaning robot's posture and speed information, and at the same time estimates the optimal error Optimal estimation of the state covariance matrix Information distribution coefficient Feedback to the corresponding main sub-filter.

9. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to any one of claims 1 to 8, characterized in that: The cleaning robot comprises: The housing comprises a shell body which is hollow inside to form a mounting cavity and has an open top, and a top plate which covers the open top of the shell body, wherein the mounting cavity comprises a first cavity body at the front side and a second cavity body at the rear side; A sealed cabin, which is arranged at the second cavity and has a sealed cavity therein; Two driving wheels are arranged oppositely at two sides of the front end of the shell and the inner ends of the two driving wheels extend into the first cavity respectively; A GPS module, comprising a GPS sensor disposed in the sealed cavity and a GPS antenna connected to the GPS sensor and partially disposed in the first cavity and partially extending to the outer side of the upper end of the top plate; Two depth gauges are arranged at intervals on the outer side of the front end of the housing; A mounting frame, which is mounted on the outer side of the front end of the housing with a gap formed therebetween; A visual sensor is mounted at the middle of the front end of the mounting frame and partially extends to the outside of the front end of the mounting frame and partially is placed in the interval; A cleaning roller brush, which rotates horizontally and can be lifted vertically, is arranged at the lower outer side of the front end of the housing and has a baffle fixed to the mounting frame at the upper front end thereof; A main control module, which is arranged in the sealed cavity; An inertial measurement device, which is arranged at one side of the sealed cavity and close to the front side of the sealed cavity; Two odometers, respectively mounted on one end of the two driving wheels extending into the first cavity; An electronic compass, which is arranged at the rear side of the sealed cavity and on the same side as the inertial navigation system; A wheel motor driver, which is arranged at the rear side of the sealed cavity; The visual sensor, inertial navigation system, odometer, electronic compass and wheel motor driver are all electrically connected to the main control module.

10. The method for positioning a moving platform hull cleaning robot based on an adaptive federated filter according to claim 9, characterized in that: The cleaning robot also includes: Ultrasonic obstacle avoidance sensors are arranged on the front end face, two side faces and the rear ends of the convex plates protruding outwards on the two outer sides of the front end of the shell and used for respectively sealing the drive motors corresponding to the two drive wheels; Two water leakage sensors are arranged on the top plate and the bottom plate in the sealed cabin, and the two water leakage sensors are located on the main control Between the upper and lower parts of the module; A backup power supply, which is arranged at one side of the sealed cavity and between the inertial navigation system and the electronic compass; A universal wheel, which is arranged at the rear end of the housing; The lighting device is arranged on the mounting frame and distributed on both sides of the visual sensor.

Images