Patrol robot based on sensing obstacle recognition

Abstract

The invention relates to the technical field of sensing navigation, in particular to a patrol robot based on sensing obstacle recognition, which comprises a path dynamic priority module, a high-flow avoidance module, a bedside dynamic area module, an approaching risk monitoring module and a path coverage state module. According to the method, node pressure distribution is adjusted and optimized through dynamic priorities, path flexibility is improved, a self-adaptive avoidance area is defined synchronously based on a bed displacement trend and a safety redundancy range, the path congestion risk is reduced, direction deviation is controllable, detour redundancy is reduced, the dynamic avoidance area shrinks or expands along with environment changes, and space adaptability is enhanced. The approach risk level is combined with the angle change rate and the distance decay rate to quantify the potential risk, the coverage overlapping proportion threshold linkage turning freezing mechanism prevents path conflicts, the multi-dimensional decision basis improves the obstacle avoidance response speed and safety, and compared with a static rule, the adaptability to ICU high-frequency environment changes is enhanced, and the method has the advantages of being simple in structure and convenient to operate. A pressure redistribution and risk grading mechanism reduces obstacle avoidance delay.

Description

Technical Field

[0001] This invention relates to the field of perception and navigation technology, and in particular to a mobile clinic robot based on sensor obstacle recognition. Background Technology

[0002] The field of perception-navigation technology encompasses a series of methods and devices for environmental perception, target recognition, and path planning in dynamic and complex environments, utilizing sensors. The core of this technology involves continuously monitoring the surrounding environment using multiple types of sensors. By perceiving information such as obstacles, terrain, and target objects, and combining this with real-time navigation path planning, autonomous movement and obstacle avoidance capabilities are achieved. Perception-navigation technology is widely used in service robots, autonomous driving systems, industrial automation, and medical mobile equipment. Its systematic technical framework typically involves sensor data acquisition, information fusion, environmental modeling, and path execution, supporting high-precision navigation in complex scenarios through the collaborative processing of multimodal information.

[0003] Among them, the sensor-based obstacle recognition-based mobile medical robot is an intelligent device suitable for intensive care environments. Addressing the challenges of dynamic changes, high personnel density, and diverse obstacles in medical spaces, it integrates infrared depth imaging sensors, ultrasonic ranging devices, and millimeter-wave radar to acquire multi-dimensional obstacle information and construct a comprehensive recognition mechanism for both static and dynamic obstacles. This device combines visual processing techniques based on image edge feature extraction with target recognition methods based on wave velocity change inversion to achieve real-time monitoring of the movement paths of medical equipment, beds, and personnel. It also achieves precise movement control of target areas through path avoidance rules built on a grid map. The overall system employs simultaneous localization and environmental mapping to continuously assess the path status within the ICU and adjusts the mobile medical robot's trajectory based on preset logical rules.

[0004] The shortcomings of existing technologies stem from their reliance on fixed rules and static environment modeling, which limits dynamic obstacle avoidance capabilities. While multi-sensor fusion mechanisms can identify obstacle types, they lack a dynamic node evaluation model based on traffic pressure and dwell time, making it difficult to address the path adjustment needs of sudden surges in patient traffic in ICU environments. Trajectory control methods based on preset logic lack dynamic calculation of directional deviation tolerance, easily leading to over-avoidance or path rigidity when obstacle density changes in the target area, increasing equipment operating energy consumption and time costs. Static avoidance zone definitions cannot adapt to real-time changes such as bed movement and equipment repositioning, resulting in a mismatch between the avoidance range and actual spatial requirements, potentially causing safety hazards due to insufficient avoidance zone coverage. Traditional risk monitoring relies solely on a single distance parameter, failing to integrate angle change trends and attenuation rates for risk classification, making it prone to misjudgment in scenarios with frequent personnel movement, causing frequent interruptions in path planning. Path coverage assessment mechanisms lack linkage control between overlap ratio and turning actions; when a new path conflicts with the avoidance zone, the system cannot quickly freeze the turning action, potentially leading to contact risks between equipment and obstacles. For example, during nighttime rounds in the ICU, traditional methods, due to the fixed avoidance areas, cannot respond in a timely manner to the narrowing of passageways caused by the temporary addition of medical equipment, forcing robots to perform long detours and delaying the execution of critical vital sign monitoring tasks. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a mobile clinic robot based on sensor obstacle recognition.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a mobile clinic robot based on sensor-based obstacle recognition includes: The path dynamic priority module is used to monitor the number of times nodes pass through, dwell time and obstacle density in the patrol path grid in real time, dynamically calculate the node passage pressure, and when the pressure increases continuously, transfer the increment to the adjacent low-density area to generate path priority distribution data. The high-traffic avoidance module is used to analyze the difference in obstacle density between the current movement direction and the high-priority area based on the path priority distribution data. If the obstacle density is lower than the current path and the direction deviation is controllable, an avoidance strategy execution state is generated. The bedside dynamic area module is used to capture changes in the coordinates of the bed edge through infrared sensors based on the execution status of the avoidance strategy, identify displacement trends by comparing the trajectory, and determine the dynamic avoidance area by combining the robot body size and safety redundancy range. The approach risk monitoring module is used to calculate the approach angle change trend and distance decay rate based on the dynamic avoidance area and the robot's movement direction, and generate an approach risk level. The path coverage status module is used to assess the overlap range between the new path and the avoidance area based on the approach risk level, the avoidance strategy execution status, and the dynamic avoidance area. If the overlap exceeds a preset ratio, the turning action is frozen, and a path coverage safety status determination result is generated.

[0007] As a further aspect of the present invention, the path priority distribution data includes the node traffic pressure increment transfer result, the adjacent low-density area allocation coefficient, and the real-time priority weight parameter; the avoidance strategy execution status includes the target area obstacle density threshold, the direction deviation tolerance range, and the strategy execution priority sequence; the dynamic avoidance area includes the bed edge coordinate change amount, the displacement trend identification parameter, and the safety redundancy range definition value; the approach risk level includes the approach angle change rate threshold, the distance decay rate critical value, and the risk level determination coefficient; and the path coverage safety status determination result includes the turning action freezing condition, the coverage overlap ratio threshold, and the path optimization verification identifier.

[0008] As a further aspect of the present invention, the path dynamic priority module includes: The node dynamic monitoring submodule is used to detect the number of times a node passes through, the dwell time, and the density of adjacent obstacles in the grid of the patrol path. It matches the three data in real time according to the node number to generate a set of node dynamic indicators. The traffic pressure calculation submodule is used to evaluate each node based on the number of passages and dwell time values ​​in the node dynamic index set, combined with the adjacent obstacle density value. If the node evaluation result is higher than the previous result in three consecutive detection cycles, the node is marked as being in a pressure increase state, and a traffic pressure state is generated. The incremental transfer execution submodule is used to filter nodes with increasing pressure based on the traffic pressure status, call the adjacent obstacle density value in the node dynamic indicator set, adjust the current node pressure increment according to the density difference of adjacent nodes, select the adjacent node with the lowest density as the receiving node, update the path priority allocation, and generate path priority distribution data.

[0009] As a further aspect of the present invention, the specific calculation formula for adjusting the pressure increment of the current node based on the density difference of adjacent nodes is as follows: ; in, Represents the current node Pressure increment value, Represents the current node The number of directly adjacent nodes, Representatives and nodes The adjacent first The obstacle density value of each node. Representing the The path influence weight of each adjacent node Represents the current node The sum of the total lengths of the connecting paths. Represents the current node The initial value of traffic pressure, Represents the current node The average value of the associated path resistance factor.

[0010] As a further aspect of the present invention, the high-traffic avoidance module includes: The path priority parsing submodule is used to obtain the path priority distribution data and movement direction data, integrate the regional path priority values ​​with the heading angle and speed information in the movement direction, and output the path priority level and direction offset. The obstacle density difference calculation submodule is used to call the path priority level and direction offset, integrate the current path obstacle density average and the target area obstacle density average, compare the difference between the two and associate the direction offset with the preset direction deviation range, output the obstacle density difference and direction deviation, and generate the difference evaluation result after fusion. The avoidance strategy triggering submodule is used to call the difference assessment results, compare the preset obstacle density difference range with the direction deviation range, and if the triggering conditions are met, define the steering angle and speed adjustment based on the average obstacle density and direction offset of the target area, and output the avoidance strategy execution status.

[0011] As a further aspect of the present invention, the bedside dynamic area module includes: The edge coordinate acquisition submodule is used to acquire infrared sensor signals at the edge of the bed based on the execution status of the avoidance strategy, extract the coordinate point set under continuous time series, integrate sensor number and timestamp information, and generate a real-time coordinate dataset. The displacement trend analysis submodule is used to call the real-time coordinate dataset, calculate the coordinate difference between adjacent time points, establish the change in movement direction based on the direction and magnitude of the difference, statistically analyze the change law of displacement direction and rate per unit time, and generate displacement trend direction data. The avoidance area calculation submodule is used to input the displacement trend direction data and the robot body size value, superimpose the maximum outer contour distance of the body on the extension line of the robot's current movement direction, and combine it with the preset safety range adjustment value to delineate the dynamic spatial boundary and generate a dynamic avoidance area.

[0012] As a further aspect of the present invention, the approach risk monitoring module includes: The approach angle analysis submodule is used to extract the robot's current movement direction vector based on the position boundary data of the dynamic avoidance area and the robot's current position coordinates, calculate the angle between the current movement direction vector and the normal vector of the avoidance area, and generate the approach angle change trend based on the change of the angle value in continuous time frames. The distance decay calculation submodule is used to call the approach angle change trend and the position coordinate sequence of the robot in continuous frames to obtain the minimum distance value sequence between the current position and the boundary of the dynamic avoidance area. Based on the minimum distance value change rate in continuous time frames, the distance change rate is calculated, and the directionality of the angle change trend is combined for classification and judgment to generate a directional distance decay rate. The risk level generation submodule is used to compare the current rate value with the difference level threshold range based on the directional distance decay rate and the preset approach risk identification threshold, and to make a classification judgment based on the level of the range in which it is located, thereby generating an approach risk level.

[0013] As a further aspect of the present invention, the specific formula for calculating the angle between the current movement direction vector and the normal vector of the avoidance area is as follows: ; in, The angle between the robot's current movement direction vector at time t and the normal vector of the avoidance area represents the angle between them. This represents the displacement vector between the robot's current position and its previous position at time t. This represents the normal vector at the boundary of the avoidance region at time t. This represents the magnitude of the displacement vector. This represents the magnitude of the normal vector. This represents the number of samples used for dynamic weighting adjustment at time t. This represents the instantaneous acceleration value of the robot at the i-th sampling point in the direction of the avoidance boundary. This represents the instantaneous velocity value of the i-th sampling point in the local path tangent direction. This represents the velocity component value of the i-th sampling point in the local path normal direction.

[0014] As a further aspect of the present invention, the path coverage status module includes: The proximity level identification submodule is used to obtain the path point position, obstacle position and velocity vector based on the proximity risk level, divide the path point level interval according to the distance, angle and velocity direction offset value, divide the path point level interval according to the proximity degree threshold, count the proportion of path points of each level, and match it with the set proportion threshold to generate the proximity level coverage. The avoidance state determination submodule is used to call the steering action execution state and dynamic avoidance area definition data according to the approach level coverage, filter the dwell time and direction change angle of the path point, determine the proportion of path segments that meet the freezing conditions, and obtain the action freezing judgment rate. The overlapping area freezing submodule is used to call the overlapping coordinates of the path point and the avoidance area and the turning action sequence according to the action freezing judgment rate, calculate the path segment overlap ratio and compare it with a preset threshold, filter the path segment ratio that meets the freezing conditions, and obtain the path coverage safety status judgment result.

[0015] As a further aspect of the present invention, the specific calculation formula for calling the steering action execution state and the dynamic avoidance area definition data is as follows: ; in, Represents the discrete value of the direction offset. Representing the The angle of change of direction of each path point Representing the The weighting factor of the avoidance zone where each path point is located. Representing the The dwell time at each waypoint Represents the total number of waypoints. This represents the average rate of change of direction of the path segment during the execution of the invoked steering action.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, node pressure distribution is optimized through dynamic priority adjustment to improve path flexibility. Adaptive avoidance zones are defined simultaneously based on bed displacement trends and safety redundancy ranges to reduce path congestion risks. Controllable directional deviations reduce detour redundancy. Dynamic avoidance zones shrink or expand with environmental changes to enhance spatial adaptability. Approach risk levels are combined with angle change rate and distance decay rate to quantify potential risks. Coverage overlap ratio thresholds are linked to a turning freeze mechanism to prevent path conflicts. Multi-dimensional decision-making criteria improve obstacle avoidance response speed and safety. Compared with static rules, this invention enhances adaptability to high-frequency environmental changes in the ICU. Pressure redistribution and risk grading mechanisms reduce obstacle avoidance delays. Attached Figure Description

[0017] Figure 1 The flowchart of the mobile clinic robot based on sensor obstacle recognition of the present invention is shown below; Figure 2 This is a flowchart of a submodule of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] Please see Figure 1 Mobile clinic robots based on sensor-based obstacle recognition include: The path dynamic priority module is used to monitor the number of times nodes pass through, dwell time and density of adjacent obstacles in the patrol path grid in real time, dynamically calculate the passage pressure of nodes, and transfer the increment to the adjacent low-density area when the pressure increases continuously, generating path priority distribution data. The high-traffic avoidance module is used to analyze the difference in obstacle density between the current movement direction and the high-priority area based on path priority distribution data. If the obstacle density in the target area is lower than that of the current path and the directional deviation is controllable, the avoidance strategy execution status is generated. The bedside dynamic area module is used to capture changes in the coordinates of the bed edge through infrared sensors based on the execution status of the avoidance strategy, identify the displacement trend by comparing the trajectory, and determine the dynamic avoidance area by combining the robot body size and safety redundancy range. The approach risk monitoring module is used to calculate the approach angle change trend and distance decay rate based on the positional relationship defined by the dynamic avoidance zone and the robot's movement direction, and to generate an approach risk level. The path coverage status module is used to assess the coverage overlap between the new path and the avoidance area based on the approach risk level, the avoidance strategy execution status, and the dynamic avoidance area definition. If the overlap exceeds a preset ratio, the turning action is frozen, and a path coverage safety status judgment result is generated.

[0021] The path priority distribution data includes the results of incremental transfer of node traffic pressure, the allocation coefficient of adjacent low-density areas, and real-time priority weight parameters. The avoidance strategy execution status includes the target area obstacle density threshold, the directional deviation tolerance range, and the strategy execution priority sequence. The dynamic avoidance area includes the bed edge coordinate change, displacement trend identification parameters, and the safety redundancy range definition value. The approach risk level includes the approach angle change rate threshold, the distance decay rate critical value, and the risk level judgment coefficient. The path coverage safety status judgment results include the turning action freezing condition, the coverage overlap ratio threshold, and the path optimization verification mark.

[0022] The path dynamic priority module includes: The node dynamic monitoring submodule detects the number of times nodes pass through, the dwell time, and the density of adjacent obstacles in the grid of the patrol path. It matches the three data points in real time according to the node number to generate a set of node dynamic indicators. The node dynamic monitoring submodule records the entire process of patrol personnel moving within the path grid through the path tracking system. Each time a person passes a node with a specific number, the passage event is recorded in real time and the cumulative number of passes is counted. For example, if node N1 is passed 5 times on a certain day, it is counted as 5 passes. During each pass, the entry and exit times of the node are recorded, and the length of time spent at that node is calculated. The average of all dwell times is then taken as the dwell time at the node. If the average dwell time at node N1 is 12 seconds, then the dwell time at that node is 12 seconds. Additionally, within a certain radius of the node, such as within 5 meters... The number of surrounding obstacles is collected using laser scanning or camera recognition technology, and the obstacle density per unit area is calculated by combining the area of ​​the scanned range. For example, if 6 obstacles are detected in the area, the corresponding area is about 78 square meters, and the obstacle density is about 0.076 obstacles per square meter. The number of passages, dwell time and obstacle density data are bound to the node number to form a data set, which is updated in real time to form a complete dynamic indicator record of the node. For example, the set items of node N1 can include 5 passages, 12 dwell time and density of 0.076. All nodes in the entire patrol path record indicator parameters in the same way and form a set for subsequent dynamic evaluation.

[0023] The traffic pressure calculation submodule evaluates each node based on the number of passages and dwell time values ​​in the node dynamic indicator set, combined with the adjacent obstacle density value. If the node evaluation result is higher than the previous result in three consecutive detection cycles, the node is marked as being in a pressure increase state, and a traffic pressure state is generated. The traffic pressure calculation submodule assesses traffic pressure based on three data points: node passage frequency, dwell time, and adjacent obstacle density. It comprehensively processes these three data types according to preset parameter weights. For example, passage frequency and dwell time are given higher weights, each accounting for 40%, while obstacle density, due to its smaller fluctuation range, is given only 20%. If a node passes through 5 times, has an average dwell time of 12 seconds, and an obstacle density of 0.076, the current traffic pressure value calculated using the weighted combination is 6.815. The assessment is recorded once every fixed detection cycle, such as collecting pressure values ​​every ten minutes, and continuously tracking for three detection cycles. Each pressure value change is compared. If the pressure value continuously increases within three cycles, it indicates that the node's traffic load is constantly rising. Furthermore, to avoid misjudgment due to small changes, a minimum change threshold is set, such as 0.4, derived from the sum of the average pressure change amplitude and standard deviation of all historical nodes. Only when the pressure continuously increases and exceeds this threshold is the node considered to be in a state of increasing pressure. Subsequently, a traffic pressure status identifier is generated in the system and recorded for path adjustment.

[0024] The incremental transfer execution submodule filters nodes with increasing pressure based on the traffic pressure status, calls the adjacent obstacle density value in the node dynamic indicator set, adjusts the current node's pressure increment based on the density difference between adjacent nodes, selects the adjacent node with the lowest density as the receiving node, updates the path priority allocation, and generates path priority distribution data. The specific calculation formula for adjusting the pressure increment of the current node based on the density difference between adjacent nodes is as follows: ; in, Represents the current node Pressure increment value, Represents the current node The number of directly adjacent nodes, Representatives and nodes The adjacent first The obstacle density value of each node. Representing the The path influence weight of each adjacent node Represents the current node The sum of the total lengths of the connecting paths. Represents the current node The initial value of traffic pressure, Represents the current node The average value of the associated path resistance factor; Detailed explanation of the formula and its calculation derivation: Parameter details It is a key performance indicator, showing the node's performance after considering obstacle density and passage resistance. The degree of change in traffic pressure.

[0025] This is achieved through network topology analysis, ensuring that each node is defined based on the actual number of connections measured.

[0026] Data is acquired through real-time monitoring using environmental sensors, with each node's data accumulated from the on-site sensor system.

[0027] The settings were determined based on historical data analysis, taking into account the frequency and ease of passage.

[0028] This was calculated using actual map data and a GIS system.

[0029] Through long-term monitoring and dynamic analysis, it was determined that this reflected the basic traffic pressure at the node before any adjustments were made.

[0030] This was obtained by averaging statistical data over a period of time and resistance factors over different time periods.

[0031] Case settings

[0032] Assume the existing data is as follows: ,node There are 4 adjacent nodes. The obstacle density value of adjacent nodes. The corresponding path affects the weight. ,node The total length of the connection path. ,node The initial value of the traffic pressure. The average value of the path resistance factor.

[0033] Formula calculation derivation process

[0034] First, calculate the product of the obstacle density value and the path influence weight, then sum them: ; Next, calculate the square root of the path length: ; Then calculate the value of the denominator: ; Next, we calculate the value of the entire expression: ; This result indicates the node The pressure increment is 4.8925, which represents the node's pressure after considering path length, obstacle density, and passage resistance. The traffic pressure increased by approximately 4.8925 units compared to the baseline. This value reflects the increase in traffic pressure at a given time period due to the influence of obstacles from neighboring nodes and path characteristics. Additional traffic pressure needs to be managed to help optimize traffic flow and management strategies between nodes.

[0035] The high-traffic avoidance module includes: The path priority parsing submodule obtains path priority distribution data and movement direction data, integrates regional path priority values ​​with heading angle and speed information in the movement direction, and outputs path priority level and direction offset. When processing path grid information, the path priority parsing submodule first extracts the priority score of each path node. These scores are usually based on factors such as path accessibility, traffic conditions, and historical frequency of use. By reading regional traffic data and path accessibility data files, each path segment is assigned a different score. For example, a main road in an urban area may be scored as 0.82, and a secondary road as 0.56. Combined with the historical trajectory frequency of moving objects, the scores are assigned as 0.75 and 0.30 respectively. In the scoring calculation process, accessibility can be quantified by the reciprocal of traffic flow, traffic conditions can be scored based on the ratio of road width to the number of obstacles, and historical usage frequency is directly extracted from the traffic monitoring platform. The three factors are weighted and summed by a fixed ratio to obtain the path priority value. Movement direction data requires collecting the heading angle and speed of the current object. This data typically comes from navigation systems or inertial navigation modules, such as heading angles of 45 degrees or 60 degrees, and speeds of 30 km / h or 50 km / h. The direction and speed information are then standardized, and the direction values ​​are matched with the path direction. Finally, the path score and direction data are fused according to a certain ratio to derive a comprehensive rating for each path node. The direction offset is calculated by the angle difference between the current orientation and the path direction. For example, if the current orientation is 70 degrees and the path direction is 45 degrees, the difference is 25 degrees. This angle deviation is recorded as a basis for path adjustment, ultimately generating a path priority level and direction offset value as module output.

[0036] The obstacle density difference calculation submodule calls the path priority level and direction offset, integrates the current path obstacle density average and the target area obstacle density average, compares the difference between the two and associates the direction offset with the preset direction deviation range, outputs the obstacle density difference and direction deviation, and generates the difference assessment result after fusion. The obstacle density difference calculation submodule needs to obtain the obstacle distribution density of each grid unit in the current path. The number of obstacles is usually counted by laser ranging equipment or visual recognition system. The number of obstacles detected in each area is divided by the area of ​​the area to obtain the obstacle density value. The average value of the density of multiple grids is taken as the average obstacle density of the current path. The obstacle density of the target area is obtained in the same way. The average obstacle density of the expected area is extracted from the path database or the predicted path model. For example, if the average obstacle density of the current path is 0.3 and that of the target area is 0.7, the difference is 0.4, which means that there are more obstacles in the target area. At the same time, the directional offset value output by the previous module is called. For example, the directional offset is 25 degrees. It is compared with the preset directional offset reference value in the system. If the system specifies a deviation threshold of 15 degrees, then the current directional offset has exceeded the normal range. The obstacle density difference and the directional offset value are combined to form a set of structural features, which characterize the change in complexity of the path from the current area to the target area, and serve as an important basis for subsequent judgment of avoidance requirements.

[0037] The avoidance strategy triggering submodule calls the difference assessment results, compares the preset obstacle density difference range with the directional deviation range, and if the triggering conditions are met, defines the steering angle and speed adjustment based on the average obstacle density and directional offset of the target area, and outputs the avoidance strategy execution status. After receiving the obstacle density difference and directional offset value, the avoidance strategy triggering submodule needs to determine whether both parameters exceed the set range simultaneously. If the obstacle density difference exceeds 0.3 and the directional offset exceeds 20 degrees, the avoidance trigger condition is met. At this point, the obstacle density of the target area is classified into levels. For example, a density between 0.0 and 0.3 is low density, 0.3 to 0.6 is medium density, and above 0.6 is high density. In the current target area with a density of 0.7, it belongs to a high-density area, and the corresponding strategy should be obvious avoidance behavior. The larger the directional offset value, the larger the angle adjustment required. For example, if the offset value is 25 degrees, the system can amplify this angle to 30 degrees for a significant steering adjustment. The speed adjustment part performs proportional deceleration based on the obstacle density level and the current driving speed. For example, if the current speed is 40 km / h, and the adjustment ratio set for a high-density area is 25%, then a deceleration of 10 km / h is required. The final generated avoidance strategy includes whether to activate, the required steering angle, and the speed adjustment range, which is then passed to the path control module for execution instructions.

[0038] The bedside dynamic area module includes: The edge coordinate acquisition submodule acquires infrared sensor signals from the edge of the bed based on the avoidance strategy execution status, extracts the coordinate point set under continuous time series, integrates sensor number and timestamp information, and generates a real-time coordinate dataset. The edge coordinate acquisition submodule deploys an array of infrared sensors at the edge of the bed. Each sensor has an independent number to identify its fixed position in physical space. Upon system startup, each sensor periodically detects the return delay of reflected infrared signals from the front. Based on the delay value, the sensor calculates the two-dimensional coordinates of the detection point using the known emission angle and installation position. The sampling time is set to once every 0.1 seconds, and the current sensor number, the calculated coordinate value, and the timestamp of the acquisition time are recorded together and written to the buffer. For example, if sensor SN3 detects coordinates (102, 248) at time 1680259352.3 seconds, the data record format is (SN3, 1680259352.3, (102, 248)). This process is performed synchronously across multiple sensors, forming a data stream arranged in chronological order. All coordinate data acquired within each time window are processed synchronously in the buffer. If a single sensor detects multiple coordinate points within a short period, the average value is taken to improve stability. The sliding window method continuously updates, allowing the generated coordinate data to continuously reflect the dynamics of the edges. This is suitable for scenarios where the position of the bed edge changes frequently, such as when a patient moves slightly on the bed, causing changes in the outline of the bedding edge.

[0039] The displacement trend analysis submodule calls the real-time coordinate dataset, calculates the coordinate difference between adjacent time points, establishes the change in movement direction based on the direction and magnitude of the difference, statistically analyzes the change pattern of displacement direction and rate per unit time, and generates displacement trend direction data. The displacement trend analysis submodule extracts the coordinate positions of two sensors with the same number between each pair of adjacent time points from the real-time coordinate dataset. By calculating the magnitude and direction of the linear displacement of the two points, the actual movement of the position corresponding to that number is obtained. This analysis is repeated every 0.1 seconds to obtain the displacement of all sensors. If a sensor with a certain number is at position (100, 250) at time t and at position (105, 255) at time t+0.1, it indicates that the position is moving in the upward and right direction, and the displacement distance can be expressed as approximately 7.07 mm using the Euclidean distance between the two points. If the displacement directions of most sensors are consistent across multiple sets of data, the overall movement trend direction can be determined. Therefore, the angle between all displacement directions is calculated in each round of analysis. If the angle between two directions is less than 30 degrees, they can be considered as consistent directions. Then, the direction vectors are weighted and averaged according to the displacement magnitude to obtain the main movement direction and average speed of the current bedside area. The speed is obtained by dividing the distance traveled per unit time by the time interval. If there are many inconsistencies in the direction, the overall direction tends to be neutral or stationary, which is reflected in the data as slow changes in the average direction and small fluctuations in the speed.

[0040] The avoidance area calculation submodule takes displacement trend direction data and robot body size values ​​as input, superimposes the maximum outer contour distance of the body on the extension line of the robot's current movement direction, and delineates the dynamic spatial boundary by combining the preset safety range adjustment value to generate a dynamic avoidance area. After receiving the displacement trend direction and rate, the avoidance area calculation submodule first determines the main direction of movement and extends the distance based on this to construct a dynamic avoidance boundary. Given the robot's dimensions, assuming a length of 600 mm and a width of 400 mm, the maximum diagonal profile length can be calculated geometrically to be approximately 721 mm. A safety margin of 100 mm is then added, resulting in a final extension length of approximately 821 mm. In this direction, starting from the robot's current center point, an avoidance rectangle with a length of 821 mm and a width of 400 mm is constructed forward. Furthermore, considering slight deviations during operation, a 50 mm width is added to the left and right as a redundancy zone, bringing the final avoidance area width to 500 mm. The corner coordinates of this rectangle are calculated based on the center position, orientation angle, and length and width, and are used in subsequent control logic to determine whether obstacles intrude into this area. This dynamic area is continuously updated with the latest displacement trend direction to reflect the changing spatial layout along the robot's path. If the patient's movement at the bedside tends to be in the direction the robot is approaching, the dynamic avoidance area will be expanded and adjusted in a timely manner to avoid missing potential cross-traffic risks in path prediction.

[0041] The approach risk monitoring module includes: The approach angle analysis submodule extracts the robot's current movement direction vector based on the position boundary data of the dynamic avoidance area and the robot's current position coordinates, calculates the angle between the current movement direction vector and the normal vector of the avoidance area, and generates the approach angle change trend based on the change of the angle value in continuous time frames. The specific formula for calculating the angle between the current movement direction vector and the normal vector of the avoidance area is as follows: ; in, The angle between the robot's current movement direction vector at time t and the normal vector of the avoidance area represents the angle between them. This represents the displacement vector between the robot's current position and its previous position at time t. This represents the normal vector at the boundary of the avoidance region at time t. This represents the magnitude of the displacement vector. This represents the magnitude of the normal vector. This represents the number of samples used for dynamic weighting adjustment at time t. This represents the instantaneous acceleration value of the robot at the i-th sampling point in the direction of the avoidance boundary. This represents the instantaneous velocity value of the i-th sampling point in the local path tangent direction. This represents the velocity component value of the i-th sampling point in the local path normal direction; Detailed explanation of the formula and its calculation derivation: First, determine the actual source of each parameter's value: (Displacement vector) and (Normal vector) is obtained through monitoring by position and orientation sensors. Let's assume that at time... , , .

[0042] and It is the magnitude of the vector, calculated as the square root of the sum of the squares of its components. , .

[0043] Indicates time The number of samples used for dynamic weighting adjustment is set through real-time monitoring, and is assumed to be [value missing]. .

[0044] , , For the first The instantaneous acceleration and velocity components in the tangential and normal directions at each sampling point along the local path are obtained through velocity and acceleration sensors, and are denoted as follows: And so on, to obtain the values ​​of other sampling points.

[0045] Now we will proceed with the specific numerical derivation: Calculate the dot product .

[0046] Calculate the vector magnitude , .

[0047] For the weighted average portion, considering four sampling points, calculate the term for each point: .

[0048] for ,calculate ,so .

[0049] Summing and averaging the calculation results of all sampling points, and assuming the sampling number is set to 4, the sum of the calculation results of the other three sampling points is approximately... The average value is .

[0050] Substitute all values ​​into the original formula to calculate the final included angle: ; Calculate using a calculator The angle obtained is approximately .

[0051] This result indicates that at time point The angle between the robot and the normal vector of the avoidance area is This numerical result reflects the robot's navigation accuracy relative to the obstacle avoidance area at its current position and serves as a direct basis for determining whether the direction of travel needs to be adjusted. This value further derives the trend of approach angle changes, thus providing decision support for dynamic path adjustment.

[0052] The distance decay calculation submodule calls the proximity angle change trend and the robot's position coordinate sequence in continuous frames to obtain the minimum distance value sequence between the current position and the boundary of the dynamic avoidance area. It calculates the distance change rate based on the minimum distance value change rate in continuous time frames, and classifies and judges it in combination with the directionality of the angle change trend to generate a directional distance decay rate. The distance decay calculation submodule processes information based on angle change trends and position data. First, it records the robot's center coordinates for each frame in a continuous time frame. Then, it selects the point closest to the robot's position in each frame from the boundary points of the avoidance zone, calculates the planar distance between that point and the boundary point, and constructs a distance sequence frame by frame. It further calculates the change in distance between consecutive frames, which is the difference between the current frame's distance value and the previous frame's distance value. Combining this with the aforementioned directional classification of angle change trends, a continuously decreasing angle is considered "approaching," and if the distance change is negative, it's a "directional approach" type. If the angle increases while the distance increases, it's a "directional moving away." To ensure reliability, angle change trends are divided into three categories: changes greater than a positive threshold are considered rapid moving away; changes less than a negative threshold are considered rapid approach; and changes in between are considered slow changes. For example, if the angle decreases by 15 degrees per frame and the distance decreases by 0.3 meters per frame, it's classified as "directional approach." The distance change is then converted into a velocity value per unit time to generate a directional distance decay rate sequence, which is further used for subsequent risk assessment.

[0053] The risk level generation submodule compares the current rate value with the difference level threshold range based on the directional distance decay rate and the preset approach risk identification threshold, and makes a classification judgment based on the level of the range it is in, thereby generating the approach risk level. The risk level generation submodule classifies risk levels based on the directional distance decay rate obtained in the previous step. It sets a threshold for the rate interval for risk identification, dividing the risk into multiple risk level intervals. For example, the high-risk interval is defined as a range with a significant decrease in negative rate value, the medium-risk interval is a range with a moderate decrease rate, and the low-risk interval is a range with a slight decrease or no change. The current rate value is compared with the threshold of each interval to determine its level, thereby generating the corresponding risk level label. For example, if the directional distance decay rate is 0.6 meters per second, and the high-risk interval is defined as a decrease of more than 0.4 meters per second, then the rate value falls into the high-risk interval, corresponding to the highest level. If the rate is 0.2 meters per second, it falls into the medium-risk interval, corresponding to the medium level. Setting these risk interval thresholds requires reference to parameters such as robot running speed, system reaction time, and braking distance. In the path planning system, simulations can be performed based on actual operating data. The probability of collision can be calculated through the relationship between distance and speed, and then the appropriate interval division can be derived to ensure that the level determination has sufficient responsiveness and adaptability.

[0054] The path coverage status module includes: The proximity level identification submodule obtains the path point location, obstacle location, and velocity vector based on the proximity risk level. According to the distance, angle, and velocity direction offset value, it divides the path point level intervals according to the proximity degree threshold, counts the proportion of path points of each level, and matches them with the set proportion threshold to generate the proximity level coverage. The proximity level recognition submodule first reads the spatial location information of all path points in the current path and simultaneously acquires the relative position and speed data of external obstacles. The position of the path points can be obtained from the vehicle's own path planning system, while the relevant information of the obstacles is collected in real time by forward-facing radar or lidar. For each path point, the relative distance to the obstacle, the angle between the directions of motion, and the velocity direction deviation are calculated. These three parameters together reflect the proximity relationship between the path point and the obstacle. The distance can be measured by measuring the Euclidean distance between the path point and the obstacle in two-dimensional space; the angle can be calculated by calculating the cosine angle between the two direction vectors of motion; and the velocity direction deviation is obtained by the difference in the direction of the two velocity vectors. Based on these basic data, different importance weights are assigned to the three parameters to form a weighted proximity index, which is then divided into preset level ranges, such as low risk (small proximity), medium risk, and high risk. Taking a real-world scenario as an example, if a path point is only 3 meters away from an obstacle, the angle between the directions is 20 degrees, and the velocity direction deviation is 10 degrees, it belongs to the high proximity level after weighted evaluation. The system iterates through all points in the path and counts the proportion of high-level path points. If the proportion is 32%, it is compared with the set risk threshold (e.g., 30%). If it exceeds the threshold, it is determined to be high-proximity level coverage, indicating that the proportion of high-risk areas in the path is relatively large.

[0055] The avoidance status determination submodule calls the steering action execution status and dynamic avoidance area definition data based on the approach level coverage, filters the dwell time and direction change angle of the path points, determines the proportion of path segments that meet the freezing conditions, and obtains the action freezing judgment rate. The specific calculation formula for calling the steering action execution status and dynamic avoidance area definition data is as follows: ; in, Represents the discrete value of the direction offset. Representing the The angle of change of direction of each path point Representing the The weighting factor of the avoidance zone where each path point is located. Representing the The dwell time at each waypoint Represents the total number of waypoints. This represents the average rate of change of direction of the path segment during the execution of the invoked steering action; Parameter definition and acquisition method description: The direction change values ​​of the path points were measured by a combination of an inertial navigation system and an angle sensor. In the experiment, the angle changes of the first to third path points were monitored to be 15°, 20°, and 25°, respectively.

[0056] The weighting is determined by the obstacle density and distance function in the avoidance area. Areas with high obstacle density are assigned a weight of 1.2, while ordinary areas are assigned a weight of 0.8.

[0057] : Dwell time obtained through path point timestamp monitoring equipment, in seconds, corresponding to 3 seconds, 4 seconds, and 5 seconds.

[0058] : The number of path points is counted; the actual number is 3.

[0059] : Real-time accumulation of the rate of change of direction of path segments and analysis Calculate the average value for each path segment. The rate of change of direction in the test sample was recorded as 0.09, 0.10, and 0.11. Calculate the mean: ; Formula calculation process and examples: Calculate the sum of the absolute values ​​of the products of the angle of change of direction and the region weight: ; Calculate the power of the variance term of the stay time: ; ; ; Calculate the denominator: ; Finally, substitute the values ​​into the formula to calculate: ; The results show that the discrete value of the direction offset is 218.03. This value increases significantly, reflecting that the influence of the direction change rate is amplified after removing the constant from the denominator, which improves the sensitivity to the action freeze judgment rate. Further screening of path segments that meet the freeze conditions will tend to be path points with higher direction change rates.

[0060] The overlapping area freezing submodule is used to call the overlapping coordinates of the path point and the avoidance area and the turning action sequence based on the action freezing judgment rate, calculate the path segment overlap ratio and compare it with the preset threshold, filter the path segment ratio that meets the freezing conditions, and obtain the path coverage safety status judgment result. The overlapping area freezing submodule receives the freezing judgment rate data and makes a preliminary judgment based on the set percentage threshold. If the freezing judgment rate is higher than the preset standard, such as exceeding 25%, the system enters the path overlapping area analysis stage. At this time, the system compares the spatial coordinates of the path points with the boundary values ​​of the dynamic avoidance area to determine whether the path points are within the avoidance area. The judgment method may include graphical calculation or spatial mapping logic, such as whether the path point coordinates fall within the polygon boundary constituting the avoidance area. All overlapping path segments are counted to obtain the path overlap ratio. This is then compared with another set threshold. If the overlap ratio exceeds 15%, it is considered that a certain proportion of paths are in the avoidance area, posing a potential risk of path freezing. For example, if 10 out of 50 path segments overlap with the avoidance area, the overlap ratio is 20%, exceeding the set 15% threshold, and the system enters the subsequent screening process. Finally, based on the number of path segments that meet the conditions, their distribution density in the path, and their continuity, the system determines the overall safety status of the path. If there are multiple consecutive overlapping paths and the freezing judgment rate is at a high level, the path is considered to lack execution stability in the current state, and the output path safety status is unsafe.

[0061] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A mobile clinic robot based on sensor-based obstacle recognition, characterized in that: The mobile clinic robot includes: The path dynamic priority module is used to monitor the number of times nodes pass through, dwell time and obstacle density in the patrol path grid in real time, dynamically calculate the node passage pressure, and when the pressure increases continuously, transfer the increment to the adjacent low-density area to generate path priority distribution data. The high-traffic avoidance module is used to analyze the difference in obstacle density between the current movement direction and the high-priority area based on the path priority distribution data. If the obstacle density is lower than the current path and the direction deviation is controllable, an avoidance strategy execution state is generated. The bedside dynamic area module is used to capture changes in the coordinates of the bed edge through infrared sensors based on the execution status of the avoidance strategy, identify displacement trends by comparing the trajectory, and determine the dynamic avoidance area by combining the robot body size and safety redundancy range. The approach risk monitoring module is used to calculate the approach angle change trend and distance decay rate based on the dynamic avoidance area and the robot's movement direction, and generate an approach risk level. The path coverage status module is used to assess the overlap range between the new path and the avoidance area based on the approach risk level, the avoidance strategy execution status, and the dynamic avoidance area. If the overlap exceeds a preset ratio, the turning action is frozen, and a path coverage safety status determination result is generated.

2. The mobile clinic robot based on sensor obstacle recognition according to claim 1, characterized in that: The path priority distribution data includes the node traffic pressure increment transfer result, the adjacent low-density area allocation coefficient, and the real-time priority weight parameter. The avoidance strategy execution status includes the target area obstacle density threshold, the direction deviation tolerance range, and the strategy execution priority sequence. The dynamic avoidance area includes the bed edge coordinate change, displacement trend identification parameter, and safety redundancy range definition value. The approach risk level includes the approach angle change rate threshold, the distance decay rate critical value, and the risk level judgment coefficient. The path coverage safety status judgment result includes the turning action freezing condition, the coverage overlap ratio threshold, and the path optimization verification identifier.

3. The mobile clinic robot based on sensor obstacle recognition according to claim 1, characterized in that: The path dynamic priority module includes: The node dynamic monitoring submodule is used to detect the number of times a node passes through, the dwell time, and the density of adjacent obstacles in the grid of the patrol path. It matches the three data in real time according to the node number to generate a set of node dynamic indicators. The traffic pressure calculation submodule is used to evaluate each node based on the number of passages and dwell time values ​​in the node dynamic index set, combined with the adjacent obstacle density value. If the node evaluation result is higher than the previous result in three consecutive detection cycles, the node is marked as being in a pressure increase state, and a traffic pressure state is generated. The incremental transfer execution submodule is used to filter nodes with increasing pressure based on the traffic pressure status, call the adjacent obstacle density value in the node dynamic indicator set, adjust the current node pressure increment according to the density difference of adjacent nodes, select the adjacent node with the lowest density as the receiving node, update the path priority allocation, and generate path priority distribution data.

4. The mobile clinic robot based on sensor obstacle recognition according to claim 3, characterized in that: The specific calculation formula for adjusting the pressure increment of the current node based on the density difference between adjacent nodes is as follows: ； in, Represents the current node Pressure increment value, Represents the current node The number of directly adjacent nodes, Representatives and nodes The adjacent first The obstacle density value of each node. Representing the The path influence weight of each adjacent node Represents the current node The sum of the total lengths of the connecting paths. Represents the current node The initial value of traffic pressure, Represents the current node The average value of the associated path resistance factor.

5. The mobile clinic robot based on sensor obstacle recognition according to claim 3, characterized in that: The high-traffic avoidance module includes: The path priority parsing submodule is used to obtain the path priority distribution data and movement direction data, integrate the regional path priority values ​​with the heading angle and speed information in the movement direction, and output the path priority level and direction offset. The obstacle density difference calculation submodule is used to call the path priority level and direction offset, integrate the current path obstacle density average and the target area obstacle density average, compare the difference between the two and associate the direction offset with the preset direction deviation range, output the obstacle density difference and direction deviation, and generate the difference evaluation result after fusion. The avoidance strategy triggering submodule is used to call the difference assessment results, compare the preset obstacle density difference range with the direction deviation range, and if the triggering conditions are met, define the steering angle and speed adjustment based on the average obstacle density and direction offset of the target area, and output the avoidance strategy execution status.

6. The mobile clinic robot based on sensor obstacle recognition according to claim 5, characterized in that: The bedside dynamic area module includes: The edge coordinate acquisition submodule is used to acquire infrared sensor signals at the edge of the bed based on the execution status of the avoidance strategy, extract the coordinate point set under continuous time series, integrate sensor number and timestamp information, and generate a real-time coordinate dataset. The displacement trend analysis submodule is used to call the real-time coordinate dataset, calculate the coordinate difference between adjacent time points, establish the change in movement direction based on the direction and magnitude of the difference, statistically analyze the change law of displacement direction and rate per unit time, and generate displacement trend direction data. The avoidance area calculation submodule is used to input the displacement trend direction data and the robot body size value, superimpose the maximum outer contour distance of the body on the extension line of the robot's current movement direction, and combine it with the preset safety range adjustment value to delineate the dynamic spatial boundary and generate a dynamic avoidance area.

7. The mobile clinic robot based on sensor obstacle recognition according to claim 6, characterized in that: The approach risk monitoring module includes: The approach angle analysis submodule is used to extract the robot's current movement direction vector based on the position boundary data of the dynamic avoidance area and the robot's current position coordinates, calculate the angle between the current movement direction vector and the normal vector of the avoidance area, and generate the approach angle change trend based on the change of the angle value in continuous time frames. The distance decay calculation submodule is used to call the approach angle change trend and the position coordinate sequence of the robot in continuous frames to obtain the minimum distance value sequence between the current position and the boundary of the dynamic avoidance area. Based on the minimum distance value change rate in continuous time frames, the distance change rate is calculated, and the directionality of the angle change trend is combined for classification and judgment to generate a directional distance decay rate. The risk level generation submodule is used to compare the current rate value with the difference level threshold range based on the directional distance decay rate and the preset approach risk identification threshold, and to make a classification judgment based on the level of the range in which it is located, thereby generating an approach risk level.

8. The mobile clinic robot based on sensor obstacle recognition according to claim 7, characterized in that: The specific formula for calculating the angle between the current movement direction vector and the normal vector of the avoidance area is as follows: ； in, The angle between the robot's current movement direction vector at time t and the normal vector of the avoidance area represents the angle between them. This represents the displacement vector between the robot's current position and its previous position at time t. This represents the normal vector at the boundary of the avoidance region at time t. This represents the magnitude of the displacement vector. This represents the magnitude of the normal vector. This represents the number of samples used for dynamic weighting adjustment at time t. This represents the instantaneous acceleration value of the robot at the i-th sampling point in the direction of the avoidance boundary. This represents the instantaneous velocity value of the i-th sampling point in the local path tangent direction. This represents the velocity component value of the i-th sampling point in the local path normal direction.

9. The mobile clinic robot based on sensor obstacle recognition according to claim 7, characterized in that: The path coverage status module includes: The proximity level identification submodule is used to obtain the path point position, obstacle position and velocity vector based on the proximity risk level, divide the path point level interval according to the distance, angle and velocity direction offset value, divide the path point level interval according to the proximity degree threshold, count the proportion of path points of each level, and match it with the set proportion threshold to generate the proximity level coverage. The avoidance state determination submodule is used to call the steering action execution state and dynamic avoidance area definition data according to the approach level coverage, filter the dwell time and direction change angle of the path point, determine the proportion of path segments that meet the freezing conditions, and obtain the action freezing judgment rate. The overlapping area freezing submodule is used to call the overlapping coordinates of the path point and the avoidance area and the turning action sequence according to the action freezing judgment rate, calculate the path segment overlap ratio and compare it with a preset threshold, filter the path segment ratio that meets the freezing conditions, and obtain the path coverage safety status judgment result.

10. The mobile clinic robot based on sensor obstacle recognition according to claim 9, characterized in that: The specific calculation formula for calling the steering action execution status and dynamic avoidance area definition data is as follows: ； in, Represents the discrete value of the direction offset. Representing the The angle of change of direction of each path point Representing the The weighting factor of the avoidance zone where each path point is located. Representing the The dwell time at each waypoint Represents the total number of waypoints. This represents the average rate of change of direction of the path segment during the execution of the invoked steering action.

Images