Material grabbing machine UWB high-precision positioning method based on neural network

Abstract

The invention relates to a material grabbing machine UWB high-precision positioning method based on a neural network, and belongs to the technical field of UWB positioning. The method comprises the following steps: receiving a UWB pulse signal transmitted by a vehicle-mounted UWB label of the material grabbing machine in real time through a UWB base station, processing the received pulse signal, inputting the processed pulse signal into a trained neural network model, and calculating through the neural network model to obtain the real-time position of the material grabbing machine. Wherein four UWB base stations are arranged to receive UWB pulse signals at the same time, after the UWB pulse signals are received, the signal intensity of the pulse signals reaching the UWB base stations is obtained, Kalman filtering is conducted on signal intensity data, and then the signal intensity data are input into the neural network model. According to the invention, the neural network model is adopted to process the UWB signal for positioning, the positioning precision can be effectively improved, the positioning precision reaches the cm level, the stability is good, and the difficulty of the existing outdoor positioning technology in the steel scrap scene positioning can be solved.

Description

Technical Field

[0001] The invention belongs to the technical field of UWB positioning, and relates to a UWB high-precision positioning method for a material grabber based on a neural network. Background Art

[0002] With the introduction of the concept of intelligent manufacturing and the advancement of unmanned and intelligent equipment, the demand for intelligence and autonomy of various types of equipment is increasing. In the unmanned scrap steel scene, the accurate positioning of the grabber directly affects the safety and work efficiency of the site. Existing positioning technologies such as GPS require the use of satellites for positioning, and the positioning accuracy is basically in meters. In the scrap steel scene, the accuracy must reach the cm level. In addition, GPS is greatly affected by weather and location, and cannot guarantee positioning accuracy. In addition, geomagnetic and WiFi positioning are easily interfered by other signals and have poor stability.

[0003] Ultra Wide Band (UWB) technology is a wireless carrier communication technology. It does not use a sinusoidal carrier, but uses nanosecond non-sinusoidal narrow pulses to transmit data. It has the advantages of low system complexity, low power spectrum density of transmitted signals, insensitivity to channel fading, low interception capability, and high positioning accuracy. The UWB positioning method is often used for indoor positioning due to its high speed and strong real-time performance. However, it is susceptible to interference and attenuation in outdoor environments, and there are a lot of multipath propagation phenomena (i.e., the signal reaches the receiver through different paths), resulting in poor outdoor positioning accuracy. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a UWB high-precision positioning method for a grabber based on a neural network. In view of the complex and highly random characteristics of the scrap steel scene environment, the grabber is positioned by training a neural network model in the scrap steel scene, thereby reducing the errors caused by multipath propagation, interference and attenuation, and improving the positioning accuracy of the UWB grabber.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A neural network-based UWB high-precision positioning method for a material grabber, the method comprising: receiving a UWB pulse signal emitted by a UWB tag on the material grabber in real time through a UWB base station, inputting the received pulse signal into a trained neural network model after processing, and obtaining the real-time position of the material grabber after calculation through the neural network model.

[0007] Furthermore, for the pulse signal sent by the UWB tag on the material handling machine, four UWB base stations are set to receive the UWB pulse signal simultaneously.

[0008] Furthermore, after the four UWB base stations simultaneously receive the UWB pulse signal, the signal strength of the pulse signal reaching each UWB base station is obtained, and then the signal strength data is Kalman filtered and then input into the neural network model.

[0009] Furthermore, the training process of the neural network model includes: the UWB base station receives the pulse signal emitted by the vehicle-mounted UWB tag, obtains the signal strength of the signal sent to the base station, and uses it as a training sample; initializes various parameters of the neural network; uses the signal strength as the input of the neural network model, and uses the actual position (x, y, z) of the grabber as the expected output of the neural network; iteratively adjusts the parameters of the neural network model, and determines whether the positioning error is less than a set threshold; if the error is less than the set threshold, verifies the model, otherwise continues to train the model; after the verification is passed, determines whether the set number of iterations is reached or whether the loss function converges, if so, outputs the model, otherwise continues to train the model.

[0010] The beneficial effects of the present invention are as follows: the present invention adopts a neural network model to process UWB signals for positioning, which can effectively improve the positioning accuracy, and the positioning accuracy reaches the cm level, and has good stability, which can solve the difficulties of current outdoor positioning technology in positioning in scrap steel scenes; at the same time, high-precision positioning is combined with safety technology (obstacle avoidance, planning, etc.) to reduce the occurrence of accidents, ensure the safety of equipment and personnel, effectively improve the dispatching efficiency and operating efficiency of the grabber, and improve production capacity.

[0011] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0013] Figure 1 This is a schematic diagram of the positioning scene;

[0014] Figure 2 It is the structure diagram of neural network;

[0015] Figure 3 Schematic diagram of the neural network model training process. DETAILED DESCRIPTION

[0016] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0017] In view of the complex and highly random characteristics of the outdoor scrap steel scene environment, the present invention proposes a high-precision positioning method for a UWB grabber based on a neural network. This method positions the grabber by training a neural network model in the scrap steel scene, thereby reducing the errors caused by multipath propagation, interference and attenuation, and improving the positioning accuracy of the UWB grabber.

[0018] like Figure 1 FIG. 1 shows an outdoor scene for an embodiment of the present invention. In the scene, four UWB base stations are set up. The four UWB base stations receive the UWB pulse signal sent by the grabber with the vehicle-mounted UWB tag in real time and obtain the signal strength when the signal is sent to each base station. Then, the signal strength received by each base station is simultaneously input as shown in FIG. Figure 2 In the neural network shown, the real-time positioning coordinates of the grabber are obtained through neural network processing, realizing accurate positioning of the outdoor site of the grabber. Among them, the input data of the four UWB base stations need to be reduced by the Kalman filter algorithm before entering the neural network to keep the signal stable.

[0019] The training process of the neural network is as follows: the pulse signal sent by the vehicle-mounted UWB tag is received by the four UWB base stations in the venue, and the signal strength sent to each base station is obtained and used as the input of the neural network; at the same time, the actual position (x, y, z) of the grabber is collected as the calibration data, that is, the expected output of the neural network, so as to calibrate the neural network. Repeat the above process, collect verification data, and verify whether the neural network training meets the standards.

[0020] Specifically, for the training of neural networks, Figure 3 As shown, the steps include: first, initializing various parameters of the neural network; then preparing training samples, i.e., input data of the neural network, and expected output of the neural network; iteratively adjusting parameter values ​​such as weights during the training process, and judging whether the positioning error is less than a set threshold; if the error is less than the set threshold, verifying the model, otherwise continuing to train the model; after the verification, judging whether the set number of iterations has been reached or whether the loss function has converged, if so, outputting the model, otherwise continuing to train the model.

[0021] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A UWB high-precision positioning method for a material grabber based on a neural network, characterized in that: The UWB pulse signal emitted by the UWB tag on the grabber is received in real time through the UWB base station. The received pulse signal is processed and input into the trained neural network model. The real-time position of the grabber is obtained after calculation by the neural network model.

2. The UWB high-precision positioning method for a material grabber according to claim 1 is characterized in that: Set up 4 UWB base stations to simultaneously receive the UWB pulse signals emitted by the UWB tags of the grabber.

3. The UWB high-precision positioning method for a material grabber according to claim 2 is characterized in that: After the four UWB base stations simultaneously receive the UWB pulse signal, the signal strength of the pulse signal reaching each UWB base station is obtained, and then the signal strength data is subjected to Kalman filtering and then input into the neural network model.

4. The UWB high-precision positioning method for a material grabber according to claim 1, characterized in that: The training process of the neural network model includes: the UWB base station receives the pulse signal emitted by the vehicle-mounted UWB tag, obtains the signal strength of the signal sent to the base station, and uses it as a training sample; initializes various parameters of the neural network; uses the signal strength as the input of the neural network model, and uses the actual position (x, y, z) of the grabber as the expected output of the neural network; iteratively adjusts the parameters of the neural network model, and determines whether the positioning error is less than the set threshold; if the error is less than the set threshold, verifies the model, otherwise continues to train the model; after the verification is passed, determines whether the set number of iterations is reached or whether the loss function converges, if so, outputs the model, otherwise continues to train the model.

Images