Intelligent vehicle-lamp system for AGV positioning and emergency communication and control method

Abstract

The invention discloses an AGV positioning and emergency communication-oriented intelligent vehicle-lamp system and a control method, and the system comprises an AGV module which comprises a visual module, a calculation unit and an IMU; the vision module is used for collecting an image of specific coding information of the intelligent lamp module, and the calculation unit is used for processing the image collected by the vision module and information obtained by the IMU; the intelligent lamp module comprises an LED lamp array, an emergency communication module and an LED control module; the LED control module controls the brightness of each LED through a digital signal, so that the LED lamp emits visible light with specific coding information, and the specific coding information adopts a digital and dynamic dark bright block coding mode; the emergency communication module is used for communicating with an upper computer; and the upper computer is used for carrying out emergency communication with the intelligent lamp module, sending specific coding information and then controlling the intelligent lamp module. According to the invention, a light coding mode is adopted, and the AGV acquires the instruction information by identifying the light codes, so that the emergency communication and control of the AGV are realized.

Description

Technical Field

[0001] The present invention relates to the field of AGV positioning and emergency communication, and in particular to an intelligent vehicle-light system and a control method for AGV positioning and emergency communication. Background Art

[0002] AGV positioning is one of the key technologies for AGV production operations. Existing AGV positioning and navigation technologies generally include tape navigation, visual navigation, laser navigation, and QR code navigation. Among them, tape navigation and QR code navigation are generally implemented through tapes and QR codes laid on the ground. The AGV's travel trajectory is often fixed, and it is impossible to achieve completely high-degree-of-freedom path planning and control. Vision and laser navigation often require the deployment of additional markers to achieve high-precision positioning and navigation.

[0003] In existing AGV communications, wireless communication protocols are generally used, such as WLAN communication protocols based on the 802.11g standard, Zigbee, etc. However, such wireless communication methods have high requirements for signal base station deployment and communication environment, and communication interruptions are prone to occur.

[0004] There are also some LED coding methods in the prior art, and the commonly used coding methods include QR code, PDF417, etc. However, in the above coding methods, black blocks account for a large proportion. If QR code is used for LED light coding, the overall brightness of the light will decrease, thus affecting the lighting function for personnel.

[0005] Therefore, it is urgent to develop an emergency communication system based on the existing communication system to deal with emergencies such as communication interruption and base station failure. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide an intelligent vehicle-light system and control method for AGV positioning and emergency communication in view of the defects in the prior art.

[0007] The technical solution adopted by the present invention to solve its technical problem is:

[0008] The present invention provides an intelligent vehicle-light system for AGV positioning and emergency communication, the system comprising: an AGV module, an intelligent lighting module and a host computer; wherein:

[0009] The AGV module includes a visual module, a computing unit and an IMU; the visual module is used to collect images of specific coded information of the intelligent lighting module, the IMU is used to obtain the three-axis attitude angle and acceleration information of the AGV module, and the computing unit is used to process the images collected by the visual module and the information obtained by the IMU;

[0010] The intelligent lamp module comprises an LED lamp array, an emergency communication module and an LED control module; the LED lamp array provides a high-frequency flashing light source under the drive of the LED control module, and the LED control module controls the brightness of each LED through a digital signal, so that the LED lamp emits visible light with specific coding information, and the specific coding information adopts a coding method of digital plus dynamic dark and bright blocks; the emergency communication module is used to communicate with the host computer;

[0011] The host computer is used to perform emergency communication with the smart lighting module, send specific coding information, and then control the smart lighting module.

[0012] Furthermore, the visual module of the present invention includes a plurality of high-speed cameras, and the sampling frame rate of the high-speed cameras is greater than the flickering frequency of the LED light array.

[0013] Furthermore, the fixing method of the high-speed camera of the present invention includes:

[0014] When there is a vertically mounted smart lamp module on the ceiling, the high-speed camera is fixed on the AGV module in an upward-looking installation manner;

[0015] When the intelligent lighting module is installed horizontally using a building support column, the high-speed camera is fixed on the AGV module in a forward-looking installation manner.

[0016] Furthermore, in the specific coding information of the present invention, the numbers are Arabic numerals 0-9, and the dynamic dark and bright blocks are 2*2 dynamic dark and bright block codes, that is, there are 4 dark blocks, each dark block is composed of 4 LED lights, and the spacing between adjacent dark blocks is 2 bright blocks. The corner point coordinate data of each dark block is stored, a total of 16, which are defined as:

[0017] F = {((0,0), [x 00 ,y 00 , z 00 ])...((i, j),[x ij ,y ij ,z ij ])...((3,3),[x 33 ,y 33 ,z 33 ])}

[0018] Among them, 3≥i, j≥0, and is an integer, [x ij ,y ij ,z ij ] is the three-dimensional coordinate of the corresponding corner point in the entire scene coordinate system;

[0019] The shape of the dynamic dark and bright block changes with time, and the position of the dark block in the LED light array changes at each moment. n combinations of dark and bright blocks of different shapes are set, and they change at a time interval of △t, so the time interval for dark and bright blocks of the same shape to appear is n*△t.

[0020] Furthermore, the AGV module and the smart lamp module of the present invention are both provided with a power supply module.

[0021] Furthermore, the host computer of the present invention is arranged in a remote server and stores a decoding protocol; the host computer can communicate with the smart lighting module and the AGV module, and is used to send specific coded information to the smart lighting module and then control the smart lighting module; and is used to remotely decode the information collected by the AGV module through the decoding protocol to obtain AGV operation instructions, including AGV scheduling information, AGV stop information, and special information in emergency situations, to realize the operation scheduling of the AGV.

[0022] The present invention provides a smart vehicle-light control method for AGV positioning and emergency communication, comprising the following steps:

[0023] Step a, encoding the smart lighting modules arranged in the scene, determining the unique serial number of each smart lighting module, performing three-dimensional measurement on the checkerboard corner coordinates of the dynamic dark and bright blocks of the specific coded information in the smart lighting module to obtain the three-dimensional coordinates of its corner points; and calibrating the internal parameters of the high-speed camera;

[0024] Step b, during positioning, the high-speed camera carried by the AGV module is used to image the specific coding information in the smart lamp module, and the chessboard image and the specific coding image are obtained after imaging, and the image is recognized by using the improved Kalman filter fusion positioning method based on dynamic coding, and the position constraint generated by dynamic coding is introduced to perform optimization calculation to obtain the position of the AGV module;

[0025] Step c. When there is a need for emergency communication, the host computer issues instructions to the smart lamp module, the LED control module in the smart lamp module analyzes the instructions to generate control instructions, and then controls the brightness of the LED lights in the LED light array to implement specific coding information. The high-speed camera in the AGV module captures the image of the emergency communication instruction, and then decodes the emergency communication instruction to obtain communication information.

[0026] Furthermore, the improved Kalman filter fusion positioning method based on dynamic coding in the method of the present invention specifically includes:

[0027] Step 1: read and decode the digital code in the specific coding information to obtain the coding information, that is, the corresponding SI serial number; use the SI serial number to read the corresponding file to obtain the three-dimensional information of the corner point;

[0028] Step 2: Read the dynamic dark and bright blocks, and extract feature points and descriptors from the checkerboard image: q ={((u 0 ,v 0 ),b 0 )...(u k ,v k ),b k ...(u n ,v n ),b n )}; f q is a set of feature points and descriptors of the current captured image; a feature point extraction and matching algorithm based on a Gaussian probability model is used to predict the position information to obtain the observed value and the predicted value;

[0029] Step 3: Based on the observed and predicted values, the Kalman filter gain is used to obtain the optimal estimate, that is, the position of the AGV.

[0030] Furthermore, the feature point extraction and matching algorithm based on the Gaussian probability model of the present invention specifically includes:

[0031] Step 21, using the FAST-9 feature point extraction method to obtain feature point coordinates;

[0032] Step 22, using SURF and BEBLID algorithms to obtain descriptors of feature points;

[0033] Step 23: Use the descriptors to match the feature points and obtain the distance information between the feature points;

[0034] Step 24: Use the Gaussian model to convert the distance into probability. The formula is:

[0035]

[0036] Where i=1 represents SURF feature space, i=2 represents BEBLID feature space; Θ (i)(j) Represents the distance between the current feature point and the jth feature point in the i-th feature space; Θ (i)min represents the minimum feature point distance in the i-th feature space, σ (i) Represents the variance of the feature point spacing, p j Indicates the similarity probability between the current feature point and the jth feature point, and takes the feature point with the largest probability as the matching result;

[0037] The following constraints are constructed through the three-dimensional data of the feature points stored under each sequence number:

[0038]

[0039] Among them, [u n v n 1] T is the feature point coordinate, K is the camera internal parameter, [X n Y n Z n ] T is the three-dimensional coordinate of the feature point; by calculating the rotation matrix R and the translation vector t, the coordinates of the AGV in the scene coordinate system can be obtained:

[0040]

[0041] in, It indicates the position observation value obtained by calculating the dark and bright blocks at time k by observing them with a high-speed camera;

[0042] At the same time, build the AGV motion model:

[0043] x k =x k-1 -x k-2 +x k-3

[0044]

[0045] Among them, X k is the state x at time k k =[x k y k ] T , using the position information at time k-1, k-2 and k-3, predict the position information at time k and obtain the predicted value

[0046] Furthermore, the method of the present invention for optimizing calculation by introducing position constraints generated by dynamic coding specifically includes:

[0047] At the time interval of n*△t, the AGV observes the code of the same shape, matches the coded images of the same shape taken at different times, and obtains the association between the two images:

[0048] [u k v k 1] T =H[u k-n*Δt v k-n*Δt 1] T

[0049] At the time kn*Δt,

[0050] Then, at time k, through the homography constraint, we have:

[0051] [u k v k 1] T =HH k-n*Δt [XY 1] T

[0052] Breaking it down further:

[0053]

[0054] At time k, the dynamic dark and bright blocks are observed by a high-speed camera, and the AGV position obtained by the homography constraint is calculated as follows:

[0055]

[0056] At the same time, since a small closed loop at the time interval of n*△t is constructed, the historical position information is adjusted by the bundle method, that is, the Bundle Adjustment method is introduced to optimize the historical position information at the k-2 and k-1 moments. The optimized position information is assumed to be The predicted value is improved to:

[0057]

[0058] At the same time, when the AGV is equipped with two or more high-speed cameras, it can perceive different observation and positioning data, and then the Kalman filter fusion algorithm is applied to perform fusion calculation;

[0059] When the AGV can observe the coded information in the lamp, the IMU can also provide positioning reference information. Therefore, the Kalman filter fusion algorithm is used to fuse the positioning information provided by the code and the positioning information observed by the IMU to define the observation matrix:

[0060]

[0061] in, is the observed positioning information, where the parameters correspond to the positioning information calculated by observing the dynamic dark and bright blocks at time K, the positioning information observed by the small closed loop at the n*△t time interval, the positioning information observed by the IMU, and H s2 is the positioning information observed by the second high-speed camera, x k =[x k y k ] T is the position of the AGV.

[0062] The beneficial effects produced by the present invention are:

[0063] 1. The intelligent vehicle-light system for AGV positioning and emergency communication of the present invention, without affecting the traditional lighting function, makes intelligent transformation of lamps and lanterns, and utilizes the form of dynamic coding to make it applicable to high-precision transportation vehicle positioning calculation, and can realize emergency communication and control of AGV through the information contained in the dynamic coding.

[0064] 2. The present invention proposes a method of dynamic dark and light block coding. This method makes intelligent improvements to traditional lamps and can provide more information without affecting the lighting function. In particular, for intelligent vehicles, by reading coded information that is invisible to the human eye, it can perform precise positioning calculations and emergency communications, thereby improving the level of intelligence in industrial production at a low cost and without adding additional equipment.

[0065] 3. The present invention proposes an improved Kalman filter fusion positioning method based on dynamic coding. This method utilizes the dynamic information in the time series formed by dynamic coding to overcome the shortcomings of traditional positioning methods that can only use current data for position calculation, construct a small closed-loop constraint in the time series, and enhance the accuracy of position calculation.

[0066] 4. The present invention proposes a feature point extraction and matching algorithm based on a Gaussian probability model. This method is different from the traditional point-to-point matching method in that the feature point distance dimension is converted into probability through a Gaussian probability model, thereby retaining more matching information and improving the matching accuracy and robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0068] Figure 1 is a diagram of the overall structure of a system according to an embodiment of the present invention;

[0069] Figure 2 It is a typical system layout diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0071] Example 1: System Device

[0072] The present invention provides an intelligent vehicle-light system for AGV positioning and emergency communication, and its hardware components include the following parts:

[0073] The intelligent vehicle-light system consists of an AGV module (vehicle), an intelligent lamp (light) module and a host computer. The AGV module includes a high-speed camera, a computing unit, a power module and a built-in IMU. In the intelligent lamp module, traditional lamps are modified to include an LED array, an emergency communication module, an LED control module and a power module. The host computer is a server, which is used to communicate with the intelligent lamp in an emergency, send coded information, and then control the lamp.

[0074] The AGV is an AGV commonly used in industrial production, and the AGV is equipped with the vision module, computing unit and power module. The vision module is a sensor for the AGV to perceive special codes. The vision module uses a high-speed camera, and its sampling frame rate is greater than the flashing frequency of the strobe light. Under this setting, the high-speed camera can clearly capture the image of the special code without affecting the lighting for personnel. It is understandable that the installation of the high-speed camera needs to be able to capture the special code, so its installation method is: when there is a lamp installed vertically on the ceiling, an upward-looking installation method is adopted; when the lamp is installed horizontally using the building support column, the high-speed camera adopts a forward-looking installation method. The installation method includes but is not limited to the above two methods, as long as it can capture the image of the special code.

[0075] The computing unit is used to receive and process data from the sensor, and after processing, transmit the positioning data to the AGV control system to facilitate subsequent AGV path planning, decision-making and control.

[0076] The power module provides operating energy for the vision module, computing unit and built-in IMU.

[0077] The intelligent lamp module installs the modified lamp in the scene. It is understandable that in the general AGV usage scenario, there are a large number of lighting fixtures, and the function of these lamps is to provide lighting for personnel. The layout of these lamps generally includes vertical installation on the ceiling, or horizontal installation using the support columns of the building. The lamp module of the present invention can be improved on the basis of the existing lighting system, or it can be installed additionally. The installation methods include but are not limited to vertical installation on the ceiling and horizontal installation using the support columns of the building.

[0078] The lamp module is different from traditional lighting lamps, and its internal components include LED lamp array, emergency communication module, LED control module and power module. The LED lamp array and LED control module can provide a flashing light source with a flashing frequency greater than 500Hz. At this frequency, the human eye cannot capture its flashing state, so for personnel, the lamp does not affect its own personnel lighting function. The LED control module uses a digital LED driver chip to accurately control the brightness of each LED through a digital signal. By setting the brightness of each LED lamp in the LED lamp array, the lamp can emit visible light with specific coded information.

[0079] Commonly used encoding methods include QR code, PDF417, etc. However, in the above encoding methods, black blocks account for a large proportion. If the LED light is encoded by QR code, the overall brightness of the light will decrease, which will affect the lighting function of personnel. Therefore, the present invention adopts a specific encoding method of digital + dynamic dark and bright blocks. The digital code uniquely encodes each LED light array, and the dynamic dark and bright block encodes the light. The dynamic dark and bright block, that is, the LED control module, uses a digital LED driver chip to accurately control the brightness of each LED through a digital signal. Among them, the bright block is a normal lighting LED light, and the bright and dark block generates specific coding information. Let the lighting brightness of the bright block be L, and the lighting brightness of the dark block be La, where L has different values ​​for different scenes. For precision operation scenes, the general value is L=2000Lx, and for non-precision operation scenes, such as warehousing, the general value is L=500Lx. The value of a is generally 10%*L~20%*L. In order not to affect conventional personnel lighting, the area occupied by the dark block is generally not more than 20% of the overall light array area.

[0080] The numbers in the special code are Arabic numerals 0 to 9. The dynamic dark and bright block codes in the special code have two functions: one is to provide positioning reference information, and the other is to provide information emergency communication.

[0081] The positioning reference information is provided, that is, through the dynamic dark and light block encoding of a specific shape, so that the AGV performs positioning calculation through the encoding in the image observed at different positions.

[0082] The provision of information emergency communication means realizing emergency communication through short-time flashing of specific digital codes.

[0083] The coding for providing positioning reference information is 2*2 dynamic dark and light block coding, that is, there are 4 dark blocks, each dark block is composed of 4 LED lights, and the distance between adjacent dark blocks is 2 light blocks. The corner point coordinate data of each dark block is stored, a total of 16. It is defined as:

[0084] F = {((0,0), [x 00 ,y 00 , z00 ])...((i, j),[x ij ,y ij ,z ij ])...((3,3),[x 33 ,y 33 ,z 33 ])}

[0085] Where 3≥i, j≥0, and is an integer. ij ,y ij ,z ij ] is the 3D coordinate of the corresponding corner point in the entire scene coordinate system. The above data is stored locally.

[0086] The shape of the dynamic dark and bright block changes over time, and the position of the dark block in the LED light array changes at each moment. Set n combinations of dark and bright blocks of different shapes, and change them at a time interval of △t, then the time interval for dark and bright blocks of the same shape to appear is n*△t.

[0087] The host computer, i.e., the server, is generally in the form of an industrial control computer. It is used to communicate with the smart lamps in an emergency, send coded information, and then control the lamps. It is understandable that the host computer can be integrated with the existing industrial control server to become a part of the existing industrial control server, or it can become a control server alone. Relevant staff can give instructions to the AGV by operating the host computer. The instructions are encoded digitally, and the parsing and decoding protocols are stored on the AGV side, the lamp side, and the server side. It is understandable that the instructions are generally AGV operation instructions, including but not limited to AGV scheduling information, AGV stop information, special information in emergency situations, etc. It is understandable that by setting the coding protocol, specific AGVs can also be scheduled for operation.

[0088] Embodiment 2: Control method

[0089] Step a: First, encode the lamps arranged in the scene. The encoding determines the unique serial number of each lamp. Let S = {s 1 ...s i s n}Where S is the lamp code set, si is the serial number of the i-th lamp, and n is the total number of lamps.

[0090] Accurately map all the special codes installed in the scene, that is, measure the coordinates of the checkerboard corner points in the special code of each lamp in three dimensions, obtain the three-dimensional coordinates of its 16 corner points, and store the three-dimensional coordinates of each special code in an XML file. Encode the serial number of each lamp, and by retrieving the serial number of the lamp, you can get the three-dimensional coordinates of each corner point.

[0091] It is understandable that the coordinate value of each corner point is different. The feature descriptors of the corner points of the chessboard image are stored in the local storage of the computing unit in advance. That is, the data stored in each coding sequence is D i ={F i f i}, where f i ={((u 0 ,v 0 ),b 0 )...(u k ,v k ),b k ...(u n ,v n ),b n )}, where Di represents the data stored in the i-th sequence number, (u k ,v k ),b k Represents the coordinates and descriptor of the kth feature point. The intrinsic parameters of the high-speed camera are calibrated in advance, and the Zhang Zhengyou calibration method can be used to obtain the intrinsic parameters K of the camera. The feature point coordinates are obtained by the FAST-9 feature point extraction method; the descriptors are obtained using the traditional SURF and BEBLID algorithms.

[0092] Step b: During positioning, the special code in the lamp is imaged using the high-speed camera carried by the AGV. After imaging, the checkerboard image and the special code image can be obtained. At the same time, the embodiment of the present invention uses a dynamic coding method, so the high-speed camera can observe the same code shape at an interval of n*△t. The present invention proposes an improved Kalman filter fusion positioning method based on dynamic coding.

[0093] The specific method is as follows:

[0094] Step 1: First, read and decode the digital code in the special code to obtain its coding information, that is, the corresponding SI serial number; use the SI serial number to read the XML file to obtain the three-dimensional information of the corner point;

[0095] Step 2: Then read the dark and bright blocks, and use the image processing algorithm to extract feature points and descriptors from the checkerboard image: q ={((u 0 ,v 0 ),b 0 )...(u k ,v k ),b k ...(u n ,v n ),b n )}; f qIt is the feature point and descriptor set of the current acquired image.

[0096] Based on the traditional ORB and BEBLID image processing algorithms, the present invention proposes a feature point extraction and matching algorithm based on a Gaussian probability model. The steps are as follows:

[0097] Step 21, using the FAST-9 feature point extraction method to obtain feature point coordinates;

[0098] Step 22, using the traditional SURF and BEBLID algorithms to obtain the descriptors of the feature points;

[0099] Step 23: Use the descriptors to match the feature points and obtain the distance information between the feature points. The distance calculated by the SURF descriptor is the Euclidean distance, and the distance calculated by the BEBLID descriptor is the Hamming distance.

[0100] Step 24: Use the Gaussian model to convert distance into probability:

[0101]

[0102] Where i=1 represents the SURF feature space, and i=2 represents the BEBLID feature space. (i)(j) Represents the distance between the current feature point and the jth feature point in the i-th feature space. Θ (i)min represents the minimum feature point distance in the i-th feature space, σ (i) Represents the variance of the feature point spacing, p j It represents the similarity probability between the current feature point and the jth feature point, and the feature point with the largest probability is taken as the matching result.

[0103] Furthermore, the following constraints can be constructed through the three-dimensional data of the feature points stored under each sequence number:

[0104]

[0105] Among them, [u n v n 1] T is the feature point coordinate, K is the camera internal parameter, obtained by prior calibration, [X n Y n Z n ] T is the three-dimensional coordinate of the feature point; through the above calculation, the rotation matrix R and the translation vector t can be obtained. The coordinates of the AGV in the scene coordinate system can be obtained.

[0106]

[0107] in, It represents the position observation value obtained by observing the dark and bright blocks at time k with a high-speed camera and performing calculations.

[0108] At the same time, build the AGV motion model:

[0109] x k =x k-1 -x k-2 +x k-3

[0110]

[0111] X k is the state x at time k k =[x k y k ] T , using the position information at time k-1, k-2 and k-3, the position information at time k can be predicted to obtain the predicted value

[0112] Step 3: After obtaining the observed values ​​and predicted values ​​in the above steps, the optimal estimate can be obtained using the Kalman filter gain:

[0113]

[0114] Where k is the Kalman filter gain.

[0115] Furthermore, the embodiment of the present invention improves on the basis of traditional Kalman filter fusion and proposes an improved Kalman filter fusion positioning method based on dynamic coding, that is, introducing the position constraints generated by dynamic coding for optimization calculation. The specific steps are as follows:

[0116] At the time interval of n*△t, the AGV observes the code of the same shape. By matching the coded images of the same shape taken at different times, the association between the two images can be obtained:

[0117] [u k v k 1] T =H[u k-n*Δt v k-n*Δt 1] T

[0118] Among them, at the time kn*Δt,

[0119] Then, at time k, through the homography constraint, we have:

[0120] [u k v k 1] T =HH k-n*Δt[XY 1] T

[0121] Can be further broken down:

[0122]

[0123] At time k, the dark and bright blocks are observed by a high-speed camera, and the AGV position obtained by the homography constraint can be calculated as:

[0124]

[0125] At the same time, since a small closed loop at the time interval of n*△t is constructed, the historical position information can be adjusted by the bundle method, that is, the Bundle Adjustment method is introduced to optimize the historical position information at the k-2 and k-1 moments. The optimized position information is assumed to be The predicted value can be improved as:

[0126]

[0127] At the same time, when the AGV is equipped with two or more high-speed cameras, it can perceive different observation and positioning data. At this time, fusion algorithms such as Kalman filtering can be applied for fusion calculation.

[0128] Since the layout of lamps is sparse, AGV cannot observe the coded information in the lamps at all times. During this time interval, the positioning of AGV is performed by IMU. The acceleration sensor and gyroscope in IMU can measure the acceleration and angular velocity of AGV, and then calculate the position by acceleration and angular velocity. Since this technology is relatively mature, the patent of this invention will not be repeated.

[0129] It is understandable that when the AGV can observe the coded information in the lamp, the IMU can also provide positioning reference information. Therefore, the Kalman filter fusion algorithm is used to fuse the positioning information provided by the code and the positioning information observed by the IMU. Specifically, the observation matrix can be defined:

[0130]

[0131] in, is the observed positioning information, where the parameters correspond to the positioning information calculated by observing the dynamic dark and bright blocks at time K, the positioning information observed by the small closed loop at the n*△t time interval, the positioning information observed by the IMU, and H s2 is the positioning information observed by the second high-speed camera, x k =[x k y k ] Tis the position of the AGV. Different from the traditional algorithm, by introducing the position constraints formed by dynamic dark and light blocks, more position calculation information is integrated to enhance the accuracy of position calculation.

[0132] Step c: When there is a need for emergency communication, the host computer issues instructions to the smart lamps, and the control module in the smart lamps analyzes the instructions to generate control instructions, and then controls the brightness of the LED lights in the LED array. It is understandable that the LED array has a high flickering frequency, so the frequency of issuing emergency communication instructions is generally 1-5Hz. That is, for an LED array with a typical flickering frequency of 500Hz, when the frequency of issuing emergency communication instructions is 1-5Hz, its specially coded flickering frequency is 499-495Hz. The high-speed camera in the AGV captures the emergency communication instructions, and then decodes the emergency communication instructions to obtain communication information.

[0133] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0134] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. An intelligent vehicle-light system for AGV positioning and emergency communication, characterized in that: The system includes: AGV module, intelligent lighting module and host computer; among which: The AGV module includes a visual module, a computing unit and an IMU; the visual module is used to collect images of specific coded information of the intelligent lighting module, the IMU is used to obtain the three-axis attitude angle and acceleration information of the AGV module, and the computing unit is used to process the images collected by the visual module and the information obtained by the IMU; The intelligent lamp module comprises an LED lamp array, an emergency communication module and an LED control module; the LED lamp array provides a high-frequency flashing light source under the drive of the LED control module, and the LED control module controls the brightness of each LED through a digital signal, so that the LED lamp emits visible light with specific coding information, and the specific coding information adopts a coding method of digital plus dynamic dark and bright blocks; the emergency communication module is used to communicate with the host computer; The host computer is used to perform emergency communication with the smart lighting module, send specific coding information, and then control the smart lighting module.

2. The intelligent vehicle-light system for AGV positioning and emergency communication according to claim 1 is characterized in that: The visual module includes a plurality of high-speed cameras, and the sampling frame rate of the high-speed cameras is greater than the flickering frequency of the LED light array.

3. The intelligent vehicle-light system for AGV positioning and emergency communication according to claim 2 is characterized in that: The fixing method of the high-speed camera includes: When there is a vertically mounted smart lamp module on the ceiling, the high-speed camera is fixed on the AGV module in an upward-looking mounting manner; When the intelligent lighting module is installed horizontally using a building support column, the high-speed camera is fixed on the AGV module in a forward-looking installation manner.

4. The intelligent vehicle-light system for AGV positioning and emergency communication according to claim 1, characterized in that: In the specific coding information, the numbers are Arabic numerals 0-9, and the dynamic dark and light blocks are 2*2 dynamic dark and light block codes, that is, there are 4 dark blocks, each dark block is composed of 4 LED lights, and the distance between adjacent dark blocks is 2 light blocks. The corner point coordinate data of each dark block is stored, a total of 16, which are defined as: F={((0,0),[x 00 ,y 00 ,z 00 ]) ... ((i,j),[x ij ,y ij ,z ij ]) ... ((3,3),[x 33 ,y 33 ,z 33 ])} Among them, 3≥i, j≥0, and is an integer, [x ij ,y ij ,z ij ] is the three-dimensional coordinate of the corresponding corner point in the entire scene coordinate system; The shape of the dynamic dark and bright block changes with time, and the position of the dark block in the LED light array changes at each moment. n combinations of dark and bright blocks of different shapes are set, and they change at a time interval of △t, so the time interval for dark and bright blocks of the same shape to appear is n*△t.

5. The intelligent vehicle-light system for AGV positioning and emergency communication according to claim 1, characterized in that: The AGV module and the smart lamp module are both provided with a power supply module.

6. The intelligent vehicle-light system for AGV positioning and emergency communication according to claim 1, characterized in that: The host computer is arranged in a remote server and stores a decoding protocol; the host computer can communicate with both the smart lamp module and the AGV module, and is used to send specific coded information to the smart lamp module and then control the smart lamp module; and is used to remotely decode the information collected by the AGV module through the decoding protocol to obtain AGV operation instructions, including AGV scheduling information, AGV stop information, and special information in emergency situations, so as to realize the operation scheduling of the AGV.

7. A method for controlling an intelligent vehicle-light for AGV positioning and emergency communication, using the intelligent vehicle-light system for AGV positioning and emergency communication as claimed in any one of claims 1 to 6, characterized in that: The following steps are involved: Step a, encoding the smart lighting modules arranged in the scene, determining the unique serial number of each smart lighting module, performing three-dimensional measurement on the checkerboard corner coordinates of the dynamic dark and bright blocks of the specific coded information in the smart lighting module to obtain the three-dimensional coordinates of its corner points; and calibrating the internal parameters of the high-speed camera; Step b, during positioning, the high-speed camera carried by the AGV module is used to image the specific coding information in the smart lamp module, and the chessboard image and the specific coding image are obtained after imaging, and the image is recognized by using the improved Kalman filter fusion positioning method based on dynamic coding, and the position constraint generated by dynamic coding is introduced to perform optimization calculation to obtain the position of the AGV module; Step c. When there is a need for emergency communication, the host computer issues instructions to the smart lamp module, the LED control module in the smart lamp module analyzes the instructions to generate control instructions, and then controls the brightness of the LED lights in the LED light array to implement specific coding information. The high-speed camera in the AGV module captures the image of the emergency communication instruction, and then decodes the emergency communication instruction to obtain communication information.

8. The intelligent vehicle-light control method for AGV positioning and emergency communication according to claim 7, characterized in that: The improved Kalman filter fusion positioning method based on dynamic coding in this method specifically includes: Step 1: read and decode the digital code in the specific coding information to obtain the coding information, that is, the corresponding SI serial number; use the SI serial number to read the corresponding file to obtain the three-dimensional information of the corner point; Step 2: Read the dynamic dark and bright blocks, and extract feature points and descriptors from the checkerboard image: q ={((u0,v0),b0)...(u k ,v k ),b k ...(u n ,v n ),b n )}; f q is a set of feature points and descriptors of the current captured image; a feature point extraction and matching algorithm based on a Gaussian probability model is used to predict the position information to obtain the observed value and the predicted value; Step 3: Based on the observed and predicted values, the Kalman filter gain is used to obtain the optimal estimate, that is, the position of the AGV.

9. The intelligent vehicle-light control method for AGV positioning and emergency communication according to claim 8, characterized in that: The feature point extraction and matching algorithm based on the Gaussian probability model specifically includes: Step 21, using the FAST-9 feature point extraction method to obtain feature point coordinates; Step 22, using SURF and BEBLID algorithms to obtain descriptors of feature points; Step 23: Use the descriptors to match the feature points and obtain the distance information between the feature points; Step 24: Use the Gaussian model to convert the distance into probability. The formula is: Where i=1 represents SURF feature space, i=2 represents BEBLID feature space; Θ (i)(j) Represents the distance between the current feature point and the jth feature point in the i-th feature space; Θ (i)min represents the minimum feature point distance in the i-th feature space, σ (i) Represents the variance of the feature point spacing, p j Indicates the similarity probability between the current feature point and the jth feature point, and takes the feature point with the largest probability as the matching result; The following constraints are constructed through the three-dimensional data of the feature points stored under each sequence number: Among them, [u n v n 1] T is the feature point coordinate, K is the camera internal parameter, [X n Y n Z n ] T is the three-dimensional coordinate of the feature point; by calculating the rotation matrix R and the translation vector t, the coordinates of the AGV in the scene coordinate system can be obtained: in, It indicates the position observation value obtained by calculating the dark and bright blocks at time k by observing them with a high-speed camera; At the same time, build the AGV motion model: x k =x k-1 -x k-2 +x k-3 Among them, X k is the state x at time k k =[x k y k ] T , using the position information at time k-1, k-2 and k-3, predict the position information at time k and obtain the predicted value 10. The intelligent vehicle-light control method for AGV positioning and emergency communication according to claim 9, characterized in that: The method for introducing the position constraints generated by dynamic coding to perform optimization calculation specifically includes: At the time interval of n*△t, the AGV observes the code of the same shape, matches the coded images of the same shape taken at different times, and obtains the association between the two images: [u k v k 1] T =H[u k-n*Δt v k-n*Δt 1] T At the time kn*Δt, Among them, H k-n*Δt is the homography relationship matrix between the pixel coordinate system and the encoding plane coordinate system at the time kn*Δt; Then, at time k, through the homography constraint, we have: [u k v k 1] T =HH k-n*Δt [X Y 1] T Breaking it down further: H k-n*Δt (i) Denotes the homography matrix H k-n*Δt The i-th column of At time k, the dynamic dark and bright blocks are observed by a high-speed camera, and the AGV position obtained by the homography constraint is calculated as follows: At the same time, since a small closed loop at the time interval of n*△t is constructed, the historical position information is adjusted by the bundle method, that is, the Bundle Adjustment method is introduced to optimize the historical position information at the k-2 and k-1 moments. The optimized position information is assumed to be The predicted value is improved to: At the same time, when the AGV is equipped with two or more high-speed cameras, it can perceive different observation and positioning data, and then the Kalman filter fusion algorithm is applied to perform fusion calculation; When the AGV can observe the coded information in the lamp, the IMU can also provide positioning reference information. Therefore, the Kalman filter fusion algorithm is used to fuse the positioning information provided by the code and the positioning information observed by the IMU to define the observation matrix: in, is the observed positioning information, where the parameters correspond to the positioning information calculated by observing the dynamic dark and bright blocks at time K, the positioning information observed by the small closed loop at the n*△t time interval, the positioning information observed by the IMU, and H s2 is the positioning information observed by the second high-speed camera, x k =[x k y k ] T is the position of the AGV.

Images