Obstacle recognition method and device, unmanned aerial vehicle and storage medium

Abstract

The invention relates to the technical field of unmanned aerial vehicles, in particular to an obstacle recognition method and device, an unmanned aerial vehicle and a storage medium. The obstacle recognition method is applied to the unmanned aerial vehicle and comprises the steps of collecting sensor data of the unmanned aerial vehicle; constructing a three-dimensional grid map of an obstacle avoidance channel in front of the current unmanned aerial vehicle according to a Bayesian model; the space of an obstacle avoidance channel in front of the unmanned aerial vehicle is segmented at equal intervals, and the number of occupied grid points in the space is extracted as feature vectors; loading the trained SVM model, and identifying an obstacle type corresponding to the feature vector; and the obstacle size information is calculated, and an obstacle avoidance path is generated. The method can fully guarantee the obstacle avoidance accuracy while reducing the false alarm rate, and further improves the operation efficiency through guaranteeing the flight safety and continuous operation of the unmanned aerial vehicle sensing system.

Description

Technical Field

[0001] The present invention relates to the field of drone perception technology, and in particular to an obstacle recognition method, device, drone and storage medium. Background Art

[0002] Plant protection drones usually fly in farmlands, orchards, mountains and other environments. These areas may have obstacles such as trees, wires, and poles. In order to effectively avoid collisions and ensure the continuity of operations and the safety of equipment, sensors must be installed in the drone system to perceive the environment and avoid obstacles.

[0003] Radar sensors have outstanding performance and wide application in the field of agricultural drones due to their all-weather working ability, strong penetration, ability to detect objects within a range of tens to hundreds of meters, and high anti-interference to electromagnetic interference, temperature changes, ambient light and other factors. However, it also has some disadvantages and limitations. Radar sensors have low spatial resolution, and it is difficult to accurately identify different types of objects in complex environments. In addition, radar signals are easily affected by multipath effects.

[0004] Chinese invention patent CN117146800A "A map construction method, device, electronic device and storage medium" provides a millimeter-wave radar mapping method for plant protection UAVs. Although the radar sensor noise is suppressed based on the probability update of the Bayesian grid map, many false alarms will still be generated due to the low spatial resolution and poor position accuracy of the radar, making it impossible for the UAV to operate continuously. In addition, due to frequent false alarms of obstacles interrupting the operation, users often turn off the obstacle avoidance function to improve work efficiency, thereby inducing the risk of bombing.

[0005] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the invention and should not be regarded as an acknowledgement or any form of suggestion that the information constitutes the prior art already known to ordinary technicians in this field. Summary of the invention

[0006] The purpose of the present invention is to overcome the deficiencies in the prior art and to provide an obstacle recognition method, device, drone and storage medium to solve the problem of incorrect obstacle recognition in the related art.

[0007] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solutions:

[0008] In a first aspect, the present invention provides an obstacle identification method, which is applied to a drone, and the method comprises:

[0009] Collect sensor data from drones;

[0010] Construct a three-dimensional grid map of the obstacle avoidance channel in front of the current UAV based on the Bayesian model;

[0011] By dividing the obstacle avoidance channel in front of the drone into equal parts, the number of occupied grid points in the space is extracted as a feature vector;

[0012] Load the trained SVM model and identify the obstacle type corresponding to the feature vector;

[0013] Calculate the obstacle size information and generate an obstacle avoidance path.

[0014] Furthermore, the collecting of sensor data of the drone includes:

[0015] The real-time kinematic measurement RTK data, the inertial measurement unit IMU data and the millimeter wave radar data are collected.

[0016] Furthermore, constructing a three-dimensional grid map of the obstacle avoidance channel ahead of the current drone according to the Bayesian model includes:

[0017] Create a three-dimensional grid map, where each grid uses a binary random variable to represent the presence or absence of an obstacle in the corresponding grid;

[0018] Assign an initial prior probability to each grid, combine the information of previous observations and new observations, update the occupancy probability of the grid, and obtain the posterior probability;

[0019] The size between the posterior probability of each grid and the set threshold is judged. If the posterior probability of the grid is greater than or equal to the set threshold, it means that there is an obstacle in the grid, otherwise it means that there is no obstacle in the grid.

[0020] Furthermore, the obstacle avoidance channel in front of the drone refers to a specific spatial range in front of the drone for identifying and avoiding obstacles during flight, and the range has a clear horizontal width, vertical height and detection distance.

[0021] Furthermore, the method of dividing the space of the obstacle avoidance channel in front of the drone by equidistant division and extracting the number of occupied grid points in the space as a feature vector includes:

[0022] Divide the space within the detection distance in front of the drone into N slices with equal distance according to horizontal width;

[0023] Divide each slice into three subspaces at equal distances according to vertical height;

[0024] The number of grid occupancy in each subspace is counted and extracted as the feature vector.

[0025] Furthermore, the method for obtaining the trained SVM model includes:

[0026] Collect obstacle information in different scenarios;

[0027] Convert the collected obstacle information into an obstacle occupancy map;

[0028] Slice the obstacle equally in the horizontal direction, and count the number of occupants of the three vertical spaces (upper, middle, and lower) in each slice to construct a feature vector.

[0029] Select the linear kernel function SVM model, input the feature vector, output the obstacle type corresponding to the feature vector, and obtain the trained SVM model.

[0030] Furthermore, calculating obstacle size information and generating an obstacle avoidance path includes:

[0031] Calculate the minimum enclosing rectangle of the grid occupied by the obstacle;

[0032] The length of the minimum enclosing rectangle in the horizontal direction is obtained as the width of the obstacle, and the length of the minimum enclosing rectangle in the vertical direction is obtained as the height of the obstacle.

[0033] Generate an obstacle avoidance path based on the obstacle type and obstacle height and width information.

[0034] In a second aspect, the present invention provides an obstacle recognition device, comprising:

[0035] Data acquisition module, used to collect sensor data from drones;

[0036] A map construction module is used to construct a three-dimensional grid map of the obstacle avoidance channel in front of the current UAV based on the Bayesian model;

[0037] The feature extraction module is used to divide the space of the obstacle avoidance channel in front of the drone by equidistant division and extract the number of occupied grid points in the space as a feature vector;

[0038] The obstacle recognition module is used to load the trained SVM model and identify the obstacle type corresponding to the feature vector;

[0039] The size calculation module is used to calculate the obstacle size information and generate an obstacle avoidance path.

[0040] In a third aspect, the present invention provides a drone, comprising:

[0041] at least one processor; and

[0042] a memory communicatively connected to the at least one processor; wherein,

[0043] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the obstacle recognition method described in any embodiment of the present invention.

[0044] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the obstacle recognition method described in any embodiment of the present invention when executed.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] The present invention provides an obstacle recognition method applied to unmanned aerial vehicles. By collecting sensor data of the unmanned aerial vehicle, a three-dimensional grid map based on a Bayesian model is constructed. The obstacle avoidance channel in front of the unmanned aerial vehicle is divided into equal intervals, and the number of occupied grid points in the space is extracted as a feature vector. The feature vector is input into a trained SVM model, and the corresponding obstacle type is identified. The obstacle size information is calculated to generate an obstacle avoidance path. The obstacle avoidance accuracy rate can be fully guaranteed while reducing the false alarm rate. By ensuring the flight safety and continuous operation of the unmanned aerial vehicle perception system, the efficiency of plant protection operations is further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a flow chart of an obstacle identification method provided in Embodiment 1 of the present invention;

[0048] Figure 2 This is a structural diagram of the obstacle avoidance channel in front of the drone provided in the first embodiment of the present invention;

[0049] Figure 3 is a structural diagram of an obstacle identification device provided in Embodiment 2 of the present invention;

[0050] Figure 4 4 is a block diagram of a drone provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0051] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0052] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and the like are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", and the like may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0053] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0054] Embodiment 1:

[0055] Figure 1 This is a flow chart of an obstacle identification method provided in Embodiment 1 of the present invention. This embodiment can be applied to a drone for obstacle identification. The method can be executed by an obstacle identification device. The obstacle identification device can be implemented in the form of hardware and / or software. The obstacle identification device can be configured in an electronic device.

[0056] The obstacle recognition method provided in this embodiment is applied to a drone, aiming to improve the drone's ability to recognize obstacles in adverse weather conditions and reduce false alarms, thereby improving the drone's obstacle avoidance capability and flight safety.

[0057] Specifically, the obstacle recognition method includes:

[0058] Step 1: Collect the drone’s sensor data, including:

[0059] The real-time kinematic measurement RTK data, the inertial measurement unit IMU data and the millimeter wave radar data are collected.

[0060] In this embodiment, RTK is used to provide high-precision positioning information, IMU is used to provide the attitude information of the UAV, and millimeter-wave radar is used to detect the front and surrounding environment. Since millimeter-wave radar is not affected by adverse weather conditions such as light, rain and fog, it is used to continuously scan the surrounding environment in working environments such as rainy days, foggy days and poor light to obtain real-time obstacle information.

[0061] For example, RTK (real-time kinematic measurement), IMU (inertial measurement unit) and millimeter-wave radar are installed on a drone, and the drone is placed in an area with a flat ground and open areas on all sides for random flight. The millimeter-wave radar that is not affected by harsh environments is combined with the positioning and attitude information provided by RTK and IMU sensors to construct a Bayesian-based three-dimensional grid map.

[0062] Step 2: Construct a 3D grid map of the obstacle avoidance channel ahead of the current drone based on the Bayesian model, including:

[0063] Create a three-dimensional grid map, where each grid uses a binary random variable to represent the presence or absence of an obstacle in the corresponding grid;

[0064] Assign an initial prior probability to each grid, combine the information of previous observations and new observations, update the occupancy probability of the grid, and obtain the posterior probability;

[0065] The size between the posterior probability of each grid and the set threshold is judged. If the posterior probability of the grid is greater than or equal to the set threshold, it means that there is an obstacle in the grid, otherwise it means that there is no obstacle in the grid.

[0066] In this embodiment, the Bayesian model can be understood as a model updated by Bayesian probability. Bayesian probability refers to a probability inference method based on prior knowledge and new observation data. The posterior probability is calculated by the conditional probability of the prior probability and the newly observed data, thereby updating the probability cognition of the event.

[0067] The Bayesian model can effectively handle the uncertainty in perception and improve the quality of decision-making. By continuously receiving new sensor data and using Bayesian methods to update the understanding of the environment, the system can quickly adapt to the changing environment.

[0068] Exemplarily, three-dimensional point cloud data in the obstacle avoidance channel in front of the drone is acquired in real time through a multi-fusion sensor, and converted into a rasterized probabilistic occupancy map.

[0069] The obstacle avoidance channel in front of the drone refers to a specific spatial range in front of the drone used to identify and avoid obstacles during flight. The range has a clear horizontal width, vertical height and detection distance.

[0070] The obstacle avoidance corridor ahead of the drone can be understood as a space area ahead of the drone's flight path, which is used to detect and identify potential obstacles so that timely obstacle avoidance measures can be taken. This area is usually a three-dimensional spatial volume, and its shape and size depend on the drone's mission requirements, sensor capabilities, and the characteristics of the flight environment.

[0071] See also Figure 2 ,The three key parameters that define the front obstacle avoidance channel include the horizontal width (obs_width), the vertical height (obs_height), and the detection distance (obs_range).

[0072] Horizontal width refers to the horizontal range on both sides of the drone, that is, the width that the drone needs to monitor in the horizontal direction. This width should be wide enough to cover the possible lateral movement range of the drone and ensure that possible obstacles on both sides can be detected.

[0073] Vertical altitude refers to the vertical range from the ground (or the lowest safe flight altitude) to a certain distance above the drone. The choice of altitude should take into account the maximum climbing capability of the drone and the requirements of the flight mission, such as avoiding high wires or other aerial structures.

[0074] The detection distance refers to the maximum distance in front of the drone that can effectively detect obstacles. The longer the detection distance, the more time the drone has to react, but it also increases the computing burden and the requirements for sensor performance, so this parameter needs to be set according to safety requirements.

[0075] In this embodiment, each grid in the three-dimensional grid map contains an occupancy probability value, which reflects the possibility that the position is occupied by an obstacle. For example, each grid uses a binary random variable to indicate whether an obstacle exists in the corresponding grid, for example, 0 indicates idle and 1 indicates occupied.

[0076] Initially, an initial occupancy probability is set for each grid (usually set to an intermediate value of 0.5, indicating an unknown state), that is, in the absence of any information, each grid has an equal probability of being occupied.

[0077] By using multi-fusion sensors to obtain data about the surrounding environment to provide new observations, given information about which areas may be occupied by obstacles. Define a likelihood function to evaluate the likelihood of an observation given a grid state. For example, if the millimeter-wave radar detects an obstacle at a certain location, the probability that the grid corresponding to that location is occupied increases; conversely, if no obstacle is detected, the probability that the location is free increases.

[0078] It should be noted that if the grid is occupied, the probability of the millimeter wave radar returning a signal is higher; conversely, if the grid is empty, the probability of the millimeter wave radar returning a signal is lower. Based on the sensor characteristics, a suitable likelihood function model can be selected, such as a Gaussian distribution model.

[0079] Before receiving new observations, the current occupancy probability of the grid is used as the prior probability, and combined with the new observation information, the occupancy probability of the grid is updated using the Bayesian formula to calculate the posterior probability.

[0080] Then set an occupancy threshold. If the posterior probability of a grid is greater than or equal to the occupancy threshold, the grid is considered to be occupied and its value is set to 1; otherwise, the grid is considered to be idle and its value is set to 0.

[0081] The updated grid values ​​reflect the latest environmental status, and as time goes by and more observation data accumulates, the probability estimates of the grid can be continuously adjusted to ensure the accuracy and real-time nature of the map.

[0082] Step 3: Divide the space of the obstacle avoidance channel in front of the drone into equal parts and extract the number of occupied grid points in the space as a feature vector, including:

[0083] Divide the space within the detection distance in front of the drone into N slices with equal distance according to horizontal width;

[0084] Divide each slice into three subspaces at equal distances according to vertical height;

[0085] The number of grid occupancy in each subspace is counted and extracted as the feature vector.

[0086] In this embodiment, the obstacle avoidance channel in front of the drone is a three-dimensional space with a clear horizontal width, vertical height and detection distance. Select an appropriate number of slices N, and evenly divide the space from the detection distance in front of the drone to the horizontal width into N slices, each slice representing a space within a fixed distance range.

[0087] Each slice is then partitioned vertically, that is, within each slice, it is further divided into three sub-spaces: high altitude, middle altitude, and low altitude according to the vertical height, so as to obtain the distribution of obstacles at different height levels.

[0088] For the three subspaces of high altitude, middle altitude and low altitude in each slice, the number of subspaces marked as "occupied" is counted respectively, and the above statistical results are organized into a feature vector.

[0089] For example, if the number of slices N=5, the final feature vector may be an array with a length of 15, that is, every three consecutive elements correspond to the number of occupied grids of the high-altitude, middle-altitude, and low-altitude subspaces of a slice.

[0090] Step 4: Load the trained SVM model and identify the obstacle type corresponding to the feature vector.

[0091] The method for obtaining the trained SVM model includes:

[0092] Collect obstacle information in different scenarios;

[0093] Convert the collected obstacle information into an obstacle occupancy map;

[0094] Slice the obstacle equally in the horizontal direction, and count the number of occupants of the three vertical spaces (upper, middle, and lower) in each slice to construct a feature vector.

[0095] Select the linear kernel function SVM model, input the feature vector, output the obstacle type corresponding to the feature vector, and obtain the trained SVM model.

[0096] In this embodiment, the SVM model refers to the SVM obstacle feature distribution model, which is a machine learning model based on a support vector machine (SVM) and is specifically used to identify the types of obstacles in the obstacle avoidance channel in front of the drone. The SVM obstacle feature distribution model can classify new feature vectors into predefined obstacle categories by training the features extracted from the sensor data.

[0097] Since the linear kernel performs well in high-dimensional space and has high computational efficiency, this embodiment uses the linear kernel function SVM obstacle feature distribution model. For the drone obstacle avoidance system, when the obstacle feature distribution is relatively simple, the linear kernel function SVM obstacle feature distribution model can provide fast and reliable classification performance to help the drone safely avoid obstacles.

[0098] For example, an operating environment containing different types of obstacles (e.g., trees, wires, and poles) is selected, and a drone equipped with a multi-fusion sensor is used to fly in the selected scene to ensure that all common obstacle types are covered, and the location and category of each obstacle are marked for subsequent supervised learning.

[0099] Obtain the obstacle occupancy map by referring to the above method, cut the obstacle into N slices with equal distance, and count the number of occupied spaces in each slice as the feature vector.

[0100] All collected data are organized into a training set, where each sample includes a feature vector and its corresponding obstacle type label (such as trees, wires, and poles). A linear kernel function SVM model is selected for training, and the trained SVM obstacle feature distribution model is saved.

[0101] In this embodiment, the generated feature vector of the obstacle avoidance channel in front of the current drone is input into the trained linear kernel function SVM obstacle feature distribution model to obtain the identified obstacle type.

[0102] Step 5: Calculate the obstacle size information and generate an obstacle avoidance path, including:

[0103] Calculate the minimum enclosing rectangle of the grid occupied by the obstacle;

[0104] The length of the minimum enclosing rectangle in the horizontal direction is obtained as the width of the obstacle, and the length of the minimum enclosing rectangle in the vertical direction is obtained as the height of the obstacle.

[0105] An obstacle avoidance path is generated based on the calculated obstacle height and width characteristics and the identified obstacle type.

[0106] In this embodiment, the height and width of the obstacle are calculated according to the obstacle type output by the SVM obstacle feature distribution model, and then combined with the current flight state, appropriate obstacle avoidance instructions are generated and sent to the flight control system of the drone.

[0107] The obstacle recognition method provided in this embodiment can ensure the accuracy of obstacle avoidance while reducing the false alarm rate, and further improve the efficiency of plant protection operations by ensuring the flight safety and continuous operation of the drone perception system.

[0108] Embodiment 2:

[0109] Figure 3 This is a schematic diagram of the structure of an obstacle recognition device provided in Embodiment 2 of the present invention.

[0110] like Figure 3 As shown, the device comprises:

[0111] Data acquisition module, used to collect sensor data from drones;

[0112] A map construction module is used to construct a three-dimensional grid map of the obstacle avoidance channel in front of the current UAV based on the Bayesian model;

[0113] The feature extraction module is used to divide the space of the obstacle avoidance channel in front of the drone by equidistant division and extract the number of occupied grid points in the space as the feature vector

[0114] An obstacle recognition module, used to identify the obstacle type corresponding to the feature vector using a SVM model;

[0115] The size calculation module is used to calculate the size information of the obstacle and generate an obstacle avoidance path.

[0116] The obstacle recognition device provided in the embodiment of the present invention can execute the obstacle recognition method provided in the first embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0117] Embodiment three:

[0118] Figure 4 A schematic diagram of an electronic device that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0119] like Figure 4 As shown, the electronic device includes at least one processor and a

[0120] A memory connected to the computer, such as a read-only memory (ROM), a random access memory (RAM), etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) or the computer program loaded from the storage unit to the random access memory (RAM). In the RAM, various programs and data required for the operation of the electronic device can also be stored. The processor, ROM and RAM are connected to each other through a bus. The input / output (I / O) interface is also connected to the bus.

[0121] Multiple components in an electronic device are connected to the I / O interface, including: input units, such as keyboards, mice, etc.; output units, such as various types of displays, speakers, etc.; storage units, such as disks, optical disks, etc.; and communication units, such as network cards, modems, wireless communication transceivers, etc. The communication unit allows the electronic device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunication networks.

[0122] A processor can be a variety of general and / or special processing components with processing and computing capabilities.

[0123] Some examples include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor executes the various methods and processes described above, such as the obstacle recognition method.

[0124] In some embodiments, the obstacle recognition method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device via the ROM and / or the communication unit. When the computer program is loaded into the RAM 43 and executed by the processor, one or more steps of the obstacle recognition method described above can be executed.

[0125] Alternatively, in other embodiments, the processor can be configured to execute the obstacle recognition method by any other suitable means (e.g., by means of firmware).

[0126] The various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs, the one or more computer programs being executable and / or interpretable on a programmable system including at least one programmable processor, the programmable processor being a dedicated or general-purpose programmable processor, receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0127] The computer programs for implementing the methods of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processors of a general-purpose computer, a dedicated computer, or other programmable data processing devices, such that when the computer programs are executed by the processors, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs can be executed entirely on the machine, partially on the machine, as an independent software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0128] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0129] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0130] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0131] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0132] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0133] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An obstacle recognition method, characterized in that: Applied to a drone, the method comprises: Collect sensor data from drones; Construct a three-dimensional grid map of the obstacle avoidance channel in front of the current UAV based on the Bayesian model; By dividing the obstacle avoidance channel in front of the drone into equal parts, the number of occupied grid points in the space is extracted as a feature vector. Load the trained SVM model and identify the obstacle type corresponding to the feature vector; Calculate the obstacle size information and generate an obstacle avoidance path.

2. The method according to claim 1, characterized in that The sensor data collected from the drone includes: The real-time kinematic measurement RTK data, the inertial measurement unit IMU data and the millimeter wave radar data are collected.

3. The method according to claim 1, characterized in that The three-dimensional grid map of the obstacle avoidance channel ahead of the current UAV is constructed according to the Bayesian model, including: Create a three-dimensional grid map, where each grid uses a binary random variable to represent the presence or absence of an obstacle in the corresponding grid; Assign an initial prior probability to each grid, combine the information of previous observations and new observations, update the occupancy probability of the grid, and obtain the posterior probability; The size between the posterior probability of each grid and the set threshold is judged. If the posterior probability of the grid is greater than or equal to the set threshold, it means that there is an obstacle in the grid, otherwise it means that there is no obstacle in the grid.

4. The method according to claim 3, characterized in that The obstacle avoidance channel in front of the drone refers to a specific spatial range in front of the drone used to identify and avoid obstacles during flight. The range has a clear horizontal width, vertical height and detection distance.

5. The method according to claim 4, characterized in that The method of isometrically dividing the space of the obstacle avoidance channel in front of the drone and extracting the number of occupied grid points in the space as a feature vector includes: Divide the space within the detection distance in front of the drone into N slices with equal distance according to horizontal width; Divide each slice into three subspaces at equal distances according to vertical height; The number of grid occupancy in each subspace is counted and extracted as the feature vector.

6. The method according to claim 5, characterized in that The method for obtaining the trained SVM model includes: Collect obstacle information in different scenarios; Convert the collected obstacle information into an obstacle occupancy map; Slice the obstacle equally in the horizontal direction, and count the number of occupants of the three vertical spaces (upper, middle, and lower) in each slice to construct a feature vector. Select the linear kernel function SVM model, input the feature vector, output the obstacle type corresponding to the feature vector, and obtain the trained SVM model.

7. The method according to claim 6, characterized in that The step of calculating obstacle size information and generating an obstacle avoidance path includes: Calculate the minimum enclosing rectangle of the grid occupied by the obstacle; The horizontal length of the minimum enclosing rectangle is taken as the width of the obstacle, and the vertical length of the minimum enclosing rectangle is taken as the height of the obstacle. Generate an obstacle avoidance path based on the obstacle type and obstacle height and width information.

8. An obstacle recognition device, characterized in that: include: Data acquisition module, used to collect sensor data from drones; A map construction module is used to construct a three-dimensional grid map of the obstacle avoidance channel in front of the current UAV based on the Bayesian model; The feature extraction module is used to divide the space of the obstacle avoidance channel in front of the drone by equidistant division and extract the number of occupied grid points in the space as a feature vector; The obstacle recognition module is used to load the trained SVM model and identify the obstacle type corresponding to the feature vector; The size calculation module is used to calculate the obstacle size information and generate an obstacle avoidance path.

9. A drone, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the obstacle recognition method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the obstacle recognition method as described in any one of claims 1 to 7 is implemented.

Images