Method for constructing virtual boundaries for aerial ladder fire trucks

Abstract

This invention discloses a method for constructing a virtual boundary for a fire truck with a ladder. A sensing module collects environmental information around the ladder and sends it to a control module. The control module processes the data and constructs a three-dimensional environment model. A virtual boundary is created in the virtual space by setting a safe distance. Within the virtual boundary, the movement path of the working bucket at the end of the ladder is planned. This invention has the advantages of avoiding ladder collisions, ensuring that the ladder operates within a safe range, and improving the efficiency and safety of fire rescue work.

Description

Technical Field

[0001] This invention relates to the field of aerial fire rescue equipment technology, specifically a method for constructing virtual boundaries for aerial ladder fire trucks. Background Technology

[0002] With the acceleration of urbanization, rescue missions involving high-rise buildings and complex structures place higher demands on firefighting equipment. As a key piece of equipment for handling such missions, the safety and efficiency of the operation of aerial ladder fire trucks directly affect the success or failure of rescue operations.

[0003] Traditional ladder fire truck operation relies on the driver's experience and judgment. In complex and ever-changing operating environments, accidents can easily occur due to the driver's misjudgment or improper operation. For example, the ladder may accidentally hit a wall and be damaged; if the ladder extends too high, the driver's blind spot will increase, making it difficult to accurately locate the fire; dynamic wind loads may cause the ladder to sway, leading to misjudgment by the driver. Summary of the Invention

[0004] The purpose of this invention is to provide a method for constructing virtual boundaries for ladder fire trucks that avoids ladder collisions, ensures ladders operate within a safe range, and improves the efficiency and safety of fire rescue work.

[0005] A method for constructing virtual boundaries for aerial ladder fire trucks, characterized in that:

[0006] Step S1: The sensing module collects environmental information around the ladder and sends it to the control module;

[0007] Step S2: The control module processes the data and constructs a three-dimensional environment model;

[0008] Step S3: Create a virtual boundary in the virtual space by setting a safe distance;

[0009] Step S4: Within the virtual boundary area, plan the movement path of the working bucket at the end of the ladder;

[0010] By creating virtual boundaries, the movement range of the ladder can be accurately controlled, ensuring that the ladder operates within a safe range. This helps firefighters avoid collisions between the ladder and surrounding obstacles when performing high-altitude operations, improving the efficiency and safety of firefighters in complex environments.

[0011] The present invention also includes step S5, which triggers a braking mechanism when the distance between the working bucket at the end of the ladder and the virtual boundary is less than the set safe distance value, thereby controlling the ladder to decelerate or stop; further preventing the ladder from colliding with surrounding obstacles.

[0012] In step S1 of this invention, a lidar is used to acquire high-precision three-dimensional point cloud data, a millimeter-wave radar is used to acquire the target's distance, velocity, and angle information, and a ranging sensor is used to acquire the distance information of nearby obstacles.

[0013] The specific method of step S2 of the present invention is as follows:

[0014] Step S201: Preprocess the collected data;

[0015] Step S202: Feature extraction;

[0016] Step S203: Target detection and classification;

[0017] Step S204: Obstacle trajectory tracking;

[0018] Step S205: Fuse the processed data to create a three-dimensional environment model.

[0019] In step S201 of this invention, Gaussian filtering is used to remove noise, and coordinate transformation is used to convert the data collected in step S1 to the same coordinate system and synchronize the timestamp.

[0020] The specific method for target detection and classification in step S203 is as follows:

[0021] Step S2031: Perform preliminary classification by setting height thresholds and geometry to distinguish different types of obstacles;

[0022] Step S2032: Extract features from the 3D point cloud data and perform accurate classification;

[0023] Step S2033: Further confirm dynamic obstacles by combining the speed information from the millimeter-wave radar;

[0024] In step S204, an extended Kalman filter is used to track the detected obstacles, update the speed and position information, and for dynamic obstacles, continuously track their trajectory and predict future movement paths.

[0025] In step S205, a Kalman filter is used to fuse the data to obtain more accurate obstacle locations and states. A Bayesian network is used to process uncertain and incomplete information to improve system robustness.

[0026] The method for feature extraction of 3D point cloud data acquired by lidar in step S202 of this invention is as follows:

[0027] The K-means clustering algorithm is used to divide the 3D point cloud data into different clusters, each cluster representing a potential obstacle. The geometric features of each cluster are calculated for subsequent classification.

[0028] The RANSAC algorithm is used to detect the ground and distinguish it from obstacles. The distance d from a point to the plane is calculated using plane equations. P To determine whether a point is an obstacle, the plane equation calculation formula is as follows:

[0029] Ax + By + Cz + D = 0;

[0030]

[0031] In the formula, A, B, and C are the normal vectors of the plane; Ax + By + Cz + D = 0 is the fitted plane equation; P(x,y,z) is an unknown point; d P Let P be the distance from point P to the plane;

[0032] If d P If the value is greater than the set threshold, the point is considered an obstacle point.

[0033] The method for feature extraction of target range and angle data acquired by millimeter-wave radar is as follows:

[0034] The Doppler effect is used to calculate the velocity information of a target and distinguish between static and dynamic obstacles. The formula for calculating the Doppler effect is as follows:

[0035]

[0036] In the formula, v target λ is the target's velocity; λ is the wavelength of the radar signal; Δf target This refers to the frequency shift of the target reflected signal, i.e., the Doppler shift.

[0037] In step S3 of this invention, a smooth virtual boundary is established by combining the dynamic buffer method with kinematic constraints. Based on the dimensions of the working bucket at the end of the ladder, the expanded virtual boundary is calculated using the following formula:

[0038]

[0039] In the formula, P obstacle (x o ,y o ,z o S'(P) represents a point on the obstacle; r is the radius of the safety buffer zone of the obstacle point, which can be set to a certain threshold; W, H, and L are the width, height, and length of the ladder truck's bucket; r′ is the radius of the expanded virtual boundary, which is based on r plus the influence of the ladder truck's bucket to ensure that the entire bucket can safely avoid the obstacle; Q(x′,y′,z′) is any point on the virtual boundary; S′(P) is a point on the obstacle. obstacle ,r′) is the expanded virtual boundary.

[0040] In step S4 of this invention, the specific method for path planning is as follows:

[0041] Step S401: Select the spatial location of the target point, and the ladder fire truck obtains the three-dimensional coordinates of the target point;

[0042] Step S402: Using the improved A * Algorithm for path planning of the working bucket at the end of the ladder;

[0043] Step S403: Based on the path points calculated in step S402, use interpolation to generate a smooth trajectory;

[0044] Step S404: Calculate the motion control of the extension, contraction, and rotation of each part of the ladder using inverse kinematics;

[0045] Step S405: During the motion iteration process, optimize the constraints in real time to ensure that the ladder extends and retracts in sequence and does not exceed the range of state variables such as length and angle.

[0046] The specific calculation formula for path planning in step S402 of this invention is as follows:

[0047] f(n) = g(n) + w(n)·h(n);

[0048] w(n)=1+λ·obstacle_density(n);

[0049] h(n)=α·d(n,target)+β·obstacle_cost(n)+γ·kinematic_cost(n);

[0050] In the formula, f(n) is the total cost evaluation function from node n to the target point; g(n) is the actual cost from node n to the target point; h(n) is the estimated cost from node n to the target point, i.e., the heuristic function; d(n,target) is the geometric distance from node n to the target point; obstacle_cost(n) is the influence of obstacles near node n, i.e., the cost of the path passing through this node; kinematic_cost(n) is whether the pose of the ladder end corresponding to node n meets the joint constraints; α, β, γ are weight coefficients, which can be dynamically adjusted; w(n) is the dynamic weight coefficient; obstacle_density(n) is the local obstacle density; and λ is the sensitivity coefficient.

[0051] The specific calculation method in step S404 is as follows:

[0052] Step S4041: The state vector representation of the pose of each joint of the ladder fire truck is as follows:

[0053] q(t)=[θ1(t),l1(t),…,l n [(t),θ2(t),θ3(t)];

[0054] In the formula, θ1(t) represents the horizontal rotation of the turntable; l1(t) represents the length of the first ladder section after independent extension and retraction; l n θ2(t) is the length of the nth ladder section after independent extension and retraction; θ3(t) is the pitch angle of the entire ladder; θ4(t) is the pitch angle of the entire working bucket.

[0055] Step S4042: Use inverse kinematics to calculate the motion control of each joint of the ladder. The calculation process is as follows:

[0056] u(t) = [x(t), y(t), z(t)] T ;

[0057] e(t) = u target -u current ;

[0058]

[0059] Δq=J + ·e(t);

[0060] q new =q current +η·Δq;

[0061] In the formula, u(t) is the position of the end-of-line bucket; x(t) is the x-axis coordinate of the end-of-line bucket at time t; y(t) is the y-axis coordinate of the end-of-line bucket at time t; z(t) is the z-axis coordinate of the end-of-line bucket at time t; e(t) is the error vector; u target The target location of the end-of-pipe work; u current J represents the current position of the end-capsule bucket; J is the Jacobian matrix; J + q is the pseudo-inverse of the Jacobian matrix; Δq is the inverse kinematic iteration; q new The updated joint pose; q current The current joint pose; η is the step length coefficient.

[0062] In step S5 of this invention, a safety distance threshold d is set. safe The specific calculation formula is as follows:

[0063] d safe =v current ·t brake +δ margin ;

[0064] In the formula, v current t represents the current speed at the end of the ladder. brake δ is the maximum braking time required for the ladder to decelerate from its current speed to zero. margin This is a margin, i.e., an additional safety distance;

[0065] Based on the distance d from the working bucket at the end of the ladder to the virtual boundary (p，target) Based on the relationship with the safe distance value, the braking control mode is divided into normal mode, warning mode and emergency braking mode;

[0066] When d (p，target) >2d safe In normal mode, the ladder's end moves according to the path plan.

[0067] When d safe <d (p，target) <2d safe In the early warning mode, the ladder begins to decelerate at the end, and the deceleration of each joint is calculated based on the inverse kinematics iteration.

[0068] When d (p，target) <d safe In emergency braking mode, the end effector stops or reverses, and each joint stops or reverses its movement based on inverse kinematics iteration.

[0069] The beneficial effects of this invention are as follows: by creating a virtual boundary, the movement range of the ladder can be accurately controlled, ensuring that the ladder operates within a safe range, helping firefighters avoid collisions between the ladder and surrounding obstacles when performing high-altitude operations, and improving the work efficiency and safety of firefighters in complex environments. Attached Figure Description

[0070] Figure 1 This is a side view of a ladder fire truck;

[0071] Figure 2 This is a hardware diagram of the virtual wall technology system for aerial ladder fire trucks;

[0072] Figure 3 This is a flowchart of virtual wall technology;

[0073] Figure 4 It is a flowchart of data acquisition and processing;

[0074] Figure 5 This is a flowchart of the ladder terminal path planning;

[0075] Figure 6 This is a side view of the ladder working bucket.

[0076] Attached drawings: Work bucket-101, ladder frame III-102, ladder frame II-103, ladder frame I-104, turntable-105, truck chassis-106;

[0077] LiDAR-201, ranging sensor-202, millimeter-wave sensor-203. Detailed Implementation

[0078] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0079] As attached Figure 1 As shown, the aerial ladder fire truck includes a truck chassis 106 and an upper structure. The upper structure includes a turntable 105 mounted on the truck chassis 106, which can rotate the ladder frame, a ladder frame, and a work bucket. The turntable is provided with ladder frames I 104, II 103, and III 102 arranged sequentially from bottom to top. Ladder frames II 103 and III 102 are telescopic. The work bucket 101 is connected to the end of ladder frame III 102. The structure of the aerial ladder fire truck and the connection relationship between its components are existing technologies and will not be described in detail here. The number of ladder sections is determined according to the vehicle model and design requirements, and is not limited to the appendix to this application. Figure 1 The three sections.

[0080] In this implementation, the aerial ladder fire truck is equipped with a sensing module, which is responsible for collecting information on the position, attitude and surrounding environment of the aerial ladder. The sensing module includes a variety of sensors such as a distance sensor, lidar, millimeter-wave radar, wire sensor, encoder, tilt sensor and weighing sensor.

[0081] The lidar can generate high-precision three-dimensional point cloud maps, perform three-dimensional environment modeling, and identify the contours and distances of the surrounding environment;

[0082] The millimeter-wave radar can detect obstacle information at relatively long distances and maintain good detection performance in adverse weather conditions such as rain, fog, and snow.

[0083] The ranging sensor is used for short-range obstacle detection. When the distance between the working bucket at the end of the ladder and the obstacle exceeds the limit range, emergency measures can be taken.

[0084] As attached Figure 6 As shown, the lidar 201 is installed above the work hopper, the millimeter-wave radar 203 is installed below the work hopper, and the ranging sensor 202 is installed below the work hopper, behind the millimeter-wave radar. In this embodiment, there are two lidars 201, installed on the upper left and upper right of the work hopper respectively, two millimeter-wave radars 203, installed on the lower left and lower right of the work hopper respectively, and one ranging sensor 202. The specific number and installation position of the lidar 201, millimeter-wave radar 203, and ranging sensor 202 are not limited to this and can be set according to actual usage needs.

[0085] The encoder is installed on the turntable and is used to detect the horizontal rotation angle of the ladder.

[0086] The pull-wire sensor is fixed to the base section (fixed section) of the ladder frame, and the end of the pull wire is connected to the outermost section of the ladder frame to measure the total extension and retraction length of the ladder frame;

[0087] The weighing sensor and tilt sensor are installed on the working bucket to jointly collect the status of the ladder working bucket.

[0088] In this embodiment, laser radar and millimeter-wave radar installed on the ladder fire truck are used to monitor the environment around the vehicle in real time. They can detect fixed obstacles such as buildings, utility poles, and trees, as well as pedestrians and other dynamic obstacles, and complex obstacles such as temporary scaffolding. In this embodiment, the ranging sensor is an ultrasonic sensor, which is used to monitor the distance between the working bucket and the wall and obstacles in real time.

[0089] In this embodiment, the perception module works in conjunction with the control module and the human-computer interaction module. The control module uses multi-sensor fusion technology to process the data collected by the perception module, build a three-dimensional environment model, create a virtual boundary, perform path planning, and precisely control the movements of each part of the ladder.

[0090] The human-computer interaction module includes a control panel and host computer software. The host computer software runs in the control panel and can display the movement trajectory of the ladder in real time. It can also select the target position of the ladder's work bucket and monitor the attitude and position of each part of the ladder frame, the work bucket, and the turntable online until the work bucket reaches the target position along the planned path.

[0091] A method for constructing virtual boundaries for aerial ladder fire trucks.

[0092] Step S1: The perception module collects environmental information around the ladder and sends it to the control module. It uses lidar to obtain high-precision three-dimensional point cloud data, millimeter-wave radar to obtain the distance, speed and angle information of the target, and range sensor to obtain the distance information of nearby obstacles.

[0093] Step S2: The control module processes the data to construct an accurate and continuous three-dimensional environment model;

[0094] Step S201: Preprocess the collected data;

[0095] Noise is removed by Gaussian filtering, and the collected data is transformed to the same coordinate system by coordinate transformation, and the timestamp is synchronized.

[0096] Step S202: Feature extraction;

[0097] The feature extraction method for 3D point cloud data acquired by lidar is as follows:

[0098] The K-means clustering algorithm is used to divide the 3D point cloud data into different clusters, each cluster representing a potential obstacle. The geometric features of each cluster (such as height, width, volume, etc.) are calculated for subsequent classification.

[0099] The RANSAC algorithm is used to detect the ground and distinguish it from obstacles. The distance d from a point to the plane is calculated using plane equations. P To determine whether a point is an obstacle, the plane equation calculation formula is as follows:

[0100] Ax + By + Cz + D = 0;

[0101]

[0102] In the formula, A, B, and C are the normal vectors of the plane; Ax + By + Cz + D = 0 is the fitted plane equation; P(x,y,z) is an unknown point; d P Let P be the distance from point P to the plane;

[0103] If d P If the distance is greater than a set threshold (e.g., greater than 0.2 meters; in this embodiment, 0.2 meters is only a distance, and the specific threshold is set according to the actual situation), then the point is considered an obstacle point.

[0104] The method for feature extraction of target range and angle data acquired by millimeter-wave radar is as follows:

[0105] The Doppler effect is used to calculate the velocity information of a target (obtained by extracting the phase difference of the target's reflected signal) to distinguish between static and dynamic obstacles. The Doppler effect calculation formula is as follows:

[0106]

[0107] In the formula, v target λ is the target's velocity; λ is the wavelength of the radar signal; Δf target This refers to the frequency shift of the target reflected signal, i.e., the Doppler frequency shift.

[0108] Step S203: Target detection and classification;

[0109] Step S2031: Perform preliminary classification by setting height thresholds and geometry to distinguish different types of obstacles (such as buildings, trees, utility poles, pedestrians, moving vehicles, etc.);

[0110] Step S2032: Extract features from the 3D point cloud data and perform accurate classification;

[0111] Step S2033: Further confirm dynamic obstacles by combining the speed information from the millimeter-wave radar;

[0112] Step S204: Obstacle trajectory tracking;

[0113] The detected obstacles are tracked using an extended Kalman filter, and the velocity and position information are updated. For dynamic obstacles, their trajectories are continuously tracked and their future movement paths are predicted.

[0114] Step S205: Fuse the processed data to create a three-dimensional environment model;

[0115] By using Kalman filters to fuse data, more accurate obstacle locations and states can be obtained, creating a 3D environment model. Bayesian networks are used to handle uncertain and incomplete information, improving system robustness.

[0116] Step S3: Based on the data processing results, a virtual boundary is created in the virtual space by setting a safe distance. The virtual boundary defines the safe range for the movement of the ladder. The virtual boundary is not limited to static obstacles, but can also automatically adjust its shape and size according to the speed and direction of dynamic obstacles to ensure continuous and effective protection.

[0117] A smooth virtual boundary is established by combining the dynamic buffer method with kinematic constraints. Based on the dimensions of the working bucket at the end of the ladder, the expanded virtual boundary is calculated using the following formula:

[0118]

[0119] In the formula, P obstacle (x o ,y o ,z o S'(P) represents a point on the obstacle; r is the radius of the safety buffer zone of the obstacle point, which can be set to a certain threshold; W, H, and L are the width, height, and length of the ladder truck's bucket; r′ is the radius of the expanded virtual boundary, which is based on r plus the influence of the ladder truck's bucket to ensure that the entire bucket can safely avoid the obstacle; Q(x′,y′,z′) is any point on the virtual boundary; S′(P) is a point on the obstacle. obstacle (r′) represents the expanded virtual boundary;

[0120] Step S4: Within the virtual boundary area, plan the movement path of the ladder end work bucket from the current position to the target position;

[0121] The specific methods for path planning are as follows:

[0122] Step S401: The operator clicks to select the spatial location of the target point on the human-machine interface, and the ladder fire truck obtains the three-dimensional coordinates of the target point;

[0123] Step S402: Using the improved A * The algorithm performs path planning on the working bucket at the end of the ladder, which can efficiently avoid obstacles and find the optimal path. The improved heuristic function takes into account the obstacle density, as well as kinematic constraints such as the range of ladder joint angle changes and extension and retraction limits. The path planning algorithm used in this embodiment also introduces dynamic weight coefficients, which can balance search speed and path quality in complex environments.

[0124] The specific calculation formula for path planning is as follows:

[0125] f(n) = g(n) + w(n)·h(n);

[0126] w(n)=1+λ·obstacle_density(n);

[0127] h(n)=α·d(n,target)+β·obstacle_cost(n)+γ·kinematic_cost(n);

[0128] In the formula, f(n) is the total cost evaluation function from node n to the target point; g(n) is the actual cost from node n to the target point; h(n) is the estimated cost from node n to the target point, i.e., the heuristic function; d(n,target) is the geometric distance from node n to the target point; obstacle_cost(n) is the influence of obstacles near node n, i.e., the cost of the path passing through this node; kinematic_cost(n) is whether the pose of the ladder end corresponding to node n meets the joint constraints; α, β, γ are weight coefficients, which can be dynamically adjusted; w(n) is the dynamic weight coefficient; obstacle_density(n) is the local obstacle density; and λ is the sensitivity coefficient.

[0129] Step S403: Based on the path points calculated in step S402, use interpolation to generate a smooth trajectory;

[0130] Step S404: Inverse kinematics is used to calculate the motion control of the extension and rotation of each part of the ladder. The ladder fire truck uses sensors such as cable sensors, angle encoders, and tilt sensors to detect the position and posture of each joint of the ladder. Each joint includes the working bucket, multi-section ladder frame, turntable, etc.

[0131] The specific calculation method is as follows:

[0132] Step S4041: The state vector representation of the pose of each joint of the ladder fire truck is as follows:

[0133] q(t)=[θ1(t),l1(t),…,l n [(t),θ2(t),θ3(t)];

[0134] In the formula, θ1(t) represents the horizontal rotation of the turntable; l1(t) represents the length of the first ladder section after independent extension and retraction; l n θ2(t) is the length of the nth ladder section after independent extension and retraction; θ3(t) is the pitch angle of the entire ladder; θ4(t) is the pitch angle of the entire working bucket.

[0135] Step S4042: Use inverse kinematics to calculate the motion control of each joint of the ladder. The calculation process is as follows:

[0136] u(t) = [x(t), y(t), z(t)] T ;

[0137] e(t) = u target -u current ;

[0138]

[0139] Δq=J + ·e(t);

[0140] q new =q current +η·Δq;

[0141] In the formula, u(t) is the position of the end-of-line bucket; x(t) is the x-axis coordinate of the end-of-line bucket at time t; y(t) is the y-axis coordinate of the end-of-line bucket at time t; z(t) is the z-axis coordinate of the end-of-line bucket at time t; e(t) is the error vector; u target The target location of the end-of-pipe work; u current J represents the current position of the end-capsule bucket; J is the Jacobian matrix; J + q is the pseudo-inverse of the Jacobian matrix; Δq is the inverse kinematic iteration; q new The updated joint pose; q current The current joint pose; η is the step length coefficient;

[0142] Step S405: During the motion iteration process, optimize the constraints in real time and determine whether the variable constraints are met. If they are met, output the real-time relative pose. If they are not met, execute step S404 to ensure that the ladder extends and retracts in sequence and does not exceed the range of state variables such as length and angle.

[0143] Step S5: When the distance between the working bucket at the end of the ladder and the virtual boundary is less than the set safe distance threshold, the braking mechanism is triggered to quickly generate reverse acceleration to force the ladder to decelerate or stop; further avoiding collisions between the ladder and surrounding obstacles.

[0144] Set a safe distance threshold d safe The specific calculation formula is as follows:

[0145] d safe =v current ·t brake +δ margin ;

[0146] In the formula, v current t represents the current speed at the end of the ladder. brake δ is the maximum braking time required for the ladder to decelerate from its current speed to zero.margin This is a margin, i.e., an additional safety distance;

[0147] Based on the distance d from the working bucket at the end of the ladder to the virtual boundary (p，target) Based on the relationship with the safe distance value, the braking control mode is divided into normal mode, warning mode and emergency braking mode;

[0148] When d (p，target) >2d safe In normal mode, the ladder's end moves according to the path plan.

[0149] When d safe <d (p，target) <2d safe In the early warning mode, the ladder begins to decelerate at the end, and the deceleration of each joint is calculated based on the inverse kinematics iteration.

[0150] When d (p，target) <d safe In emergency braking mode, the end effector stops or reverses, and each joint stops or reverses its movement based on inverse kinematics iteration.

[0151] With the cooperation of the perception module, control module, and human-computer interaction module, the virtual boundary construction method can be mounted on aerial ladder fire trucks to assist operators in efficiently completing fire rescue operations. By creating virtual boundaries, the movement range of the aerial ladder can be accurately controlled to ensure that the aerial ladder operates within a safe range. This helps firefighters avoid collisions between the aerial ladder and surrounding obstacles when performing high-altitude operations, thus improving the efficiency and safety of firefighters in complex environments.

Claims

1. A method for constructing virtual boundary of aerial ladder fire truck, characterized in that: Step S1: the sensing module collects the environment information around the aerial ladder and sends it to the control module; Step S2: the control module processes the data and constructs a three-dimensional environment model; Step S201: preprocessing the collected data; Step S202: feature extraction; Step S203: target detection and classification; The specific method of target detection and classification in step S203 is: Step S2031: preliminary classification by setting height threshold and geometric shape to distinguish different types of obstacles; Step S2032: extracting features from three-dimensional point cloud data for accurate classification; Step S2033: further confirming dynamic obstacles by combining speed information of millimeter wave radar; Step S204: obstacle trajectory tracking; Step S205: fusing the processed data to create a three-dimensional environment model; Step S3: creating a virtual boundary in the virtual space by setting a safe distance; Step S4: planning the movement path of the aerial ladder's end bucket within the virtual boundary.

2. The virtual boundary construction method for a turntable ladder fire apparatus according to claim 1, wherein: Step S5: when the distance between the aerial ladder's end bucket and the virtual boundary is less than the set safe distance value, triggering the braking mechanism to control the aerial ladder to slow down or stop.

3. The virtual boundary construction method of a turntable ladder fire apparatus according to claim 1 or 2, characterized in that: In step S1, laser radar is used to obtain high-precision three-dimensional point cloud data, millimeter wave radar is used to obtain distance, speed and angle information of the target, and distance sensor is used to obtain distance information of the near-distance obstacle. 4.The method according to claim 1 or 2, characterized in that: In step S201, Gaussian filter is used for denoising, and the data collected in step S1 is converted to the same coordinate system by coordinate transformation and synchronized with time stamp; In step S204, extended Kalman filter is used to track the detected obstacles to update speed and position information, and for dynamic obstacles, their trajectories are continuously tracked and their future movement paths are predicted; In step S205, Kalman filter is used to fuse the data, and Bayesian network is used to process uncertain and incomplete information.

5. The virtual boundary construction method of a turntable ladder fire apparatus according to claim 1 or 2, wherein: In step S202, the feature extraction method for three-dimensional point cloud data collected by laser radar is: K-means clustering algorithm is used to divide the three-dimensional point cloud data into different clusters, each cluster represents a potential obstacle, and the geometric features of each cluster are calculated for subsequent classification; The ground is detected using a RANSAC algorithm and distinguished from obstacles, and a plane equation is used to calculate the distance d of a point to a plane P , and determine whether it is an obstacle point. The plane equation calculation formula is: Ax+By+Cz+D=0; In the formula, A, B, C are normal vectors of the plane; Ax+By+Cz+D=0 is the fitted plane equation; (x, y, z) is the three-dimensional coordinate of the point cloud; P(x P ,y P ,z P ) is the three-dimensional coordinate of an unknown point P; d P is the distance from the point P to the plane; If d P If the distance is greater than a set threshold, the point is considered an obstacle point. For the target distance and angle data obtained by millimeter wave radar, the feature extraction method is: Doppler effect is used to calculate the speed information of the target to distinguish static and dynamic obstacles, and the Doppler effect calculation formula is: In the formula, v target λ is the target's velocity; λ is the wavelength of the radar signal; Δf target This refers to the frequency shift of the target reflected signal, i.e., the Doppler shift. 6.The method according to claim 1 or 2, characterized in that: In step S3, dynamic buffer zone method is combined with kinematic constraints to establish a smooth virtual boundary, and according to the size of the aerial ladder's end bucket, the extended virtual boundary is calculated, and the calculation formula is: where P obstacle (x o ,y o ,z o ) is a point on the obstacle; r is a safety buffer radius of the obstacle point, which can be set as a certain threshold; W, H, and L are the width, height, and length of the aerial ladder working bucket; r' is the extended virtual boundary radius; Q(x', y', z') is an arbitrary point on the virtual boundary; and S'(P obstacle ,r') is the extended virtual boundary.

7. The virtual boundary construction method of a turntable ladder fire apparatus according to claim 1 or 2, wherein: In step S4, the specific method of path planning is: Step S401: selecting the spatial position of the target point, and the aerial ladder fire truck obtaining the three-dimensional coordinates of the target point; Step S402: Use the improved A * algorithm to plan the path for the ladder tipper end bucket; Step S403: According to the path point calculated in step S402, a smooth trajectory is generated using an interpolation method; Step S404: The motion control of the extension and rotation of each part of the aerial ladder is calculated by inverse kinematics; Step S405: In the motion iteration process, the constraints are optimized in real time to ensure that the ladder frame extends and retracts in sequence.

8. The virtual boundary construction method of a turntable ladder fire apparatus according to claim 7, wherein: The specific calculation formula of path planning in step S402 is: f(n)=g(n)+w(n)·h(n); w(n)=1+λ·obstacle_density(n); h(n)=α·d(n,target)+β·obstacle_cost(n)+γ·kinematic_cost(n); In the formula, f(n) is the total cost evaluation function of node n to the target point; g(n) is the actual cost of node n to the target point; h(n) is the estimated cost of node n to the target point, i.e. the heuristic function; d(n,target) is the geometric distance from node n to the target point; obstacle_cost(n) is the influence of obstacles near node n, i.e. the cost of the path passing through the node; kinematic_cost(n) is whether the end pose of the aerial ladder corresponding to node n meets the joint limit; α, β, γ are weight coefficients, which can be dynamically adjusted; w(n) is a dynamic weight coefficient; obstacle_density(n) is the local obstacle density, and λ is the sensitivity coefficient; The specific calculation method in step S404 is: Step S4041: The state vector of each joint pose of the aerial ladder fire truck is represented as: q(t) = [θ1(t), l1(t),..., l n (t), θ2(t), θ3(t)]; In the formula, θ1(t) is the horizontal rotation of the rotary table; l1(t) is the length of the first ladder section after independent extension; l n (t) is the length of the nth ladder section after independent extension; θ2(t) is the pitch angle of the whole ladder frame; and θ3(t) is the pitch angle of the whole working bucket. Step S4042: The motion control of each joint of the aerial ladder is calculated by inverse kinematics, and the calculation process is: u(t) = [x(t), y(t), z(t)] T ; e(t) = u target - u current ; Δq = J + • e(t); q new = q current + η · Δq; In the formula, u(t) is the end working bucket position; x(t) is the x-axis coordinate value of the end working bucket at t moment; y(t) is the y-axis coordinate value of the end working bucket at t moment; z(t) is the z-axis coordinate value of the end working bucket at t moment; e(t) is the error vector; u target is the end working bucket target position; u current is the current position of the end working bucket; J is the Jacobian matrix; J + is the pseudo-inverse of the Jacobian matrix; Δq is the inverse kinematics iteration; q new is the updated joint pose state; q current is the current joint pose state; η is the step length coefficient.

9. The virtual boundary construction method of a turntable ladder fire apparatus according to claim 1 or 2 or 8, wherein: The safety distance threshold value d is set in step S5 safe The specific calculation formula is as follows: d safe = v current · t brake + δ margin ; where v current is the current speed of the tip bucket of the ladder; t brake is the maximum braking time required for the ladder to decelerate from the current speed to zero; δ margin is the margin, i.e. the additional safety distance; According to the distance value d of the ladder tip working bucket to the virtual boundary (p，target) The relationship with the safety distance value divides the brake control mode into normal mode, pre-warning mode and emergency brake mode; When d (p，target) > 2d safe is normal mode, the ladder tip moves according to path planning; When d safe <d (p，target) <2d safe is the pre-warning mode, the ladder truck terminal starts to decelerate, and each joint movement deceleration is calculated according to the inverse kinematics iteration; When d (p，target) <d safe When in emergency braking mode, the end stops or reverses motion, and each joint stops or reverses motion according to inverse kinematics iteration.

Images