An intelligent navigation method for a robot

Abstract

The present application relates to the technical field of robot navigation, and more particularly to an intelligent navigation method for a robot, in which the image of an obstacle in the running track of the robot is acquired in real time, the image is preprocessed, the motion feature is extracted and the motion type is classified, the problem of fuzzy identification of irregular obstacles in the traditional method is solved, and the error decision caused by misjudgment of the obstacle type is avoided, the application adopts differentiated judgment logic for different motion type obstacles: for regular motion obstacles, the passing mode is determined by comprehensively considering the relative motion distance and the difficulty of detouring, so as to avoid excessive conservatism, on the other hand, for irregular motion obstacles, the running uncertainty representation value is introduced to quantify the motion confusion degree, and the real-time relative distance judgment is combined to ensure the safety of detouring, so as to prevent the efficiency from being reduced due to excessive avoidance of low-risk obstacles, and also avoid collision caused by forcibly detouring high-risk obstacles.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot navigation, and in particular to an intelligent navigation method for a robot. BACKGROUND

[0002] In the field of autonomous navigation of robots (such as service robots, delivery robots, etc.), the avoidance of dynamic obstacles (especially irregular motion obstacles such as pedestrians, pets, etc.) is a core technical problem. Most existing algorithms are designed for static obstacles (such as fixed facilities) or regular motion obstacles (such as vehicles traveling on fixed paths), and achieve detouring through preset fixed safety distance or trajectory prediction. However, for irregular motion obstacles (such as pedestrians suddenly changing direction, pets randomly running), their motion direction and speed frequently change, and traditional methods are difficult to accurately predict the trajectory, which can easily lead to collision or overly conservative detouring (such as frequent emergency stops, reducing the efficiency of passing).

[0003] Chinese Patent Application No. CN202411690612.8 discloses an intelligent navigation path planning system for a robot, which includes a controller, a throwing module, a drive wheel, a radio frequency identification module, and a positioning module. The input end of the controller is connected with a laser radar, a camera, a ranging sensor, a cliff sensor, and a collision sensor, which are used to realize the shooting identification of the environment, obstacles, and high drop-offs around the travel route. The output end of the controller is connected with the drive wheel. When there is an obstacle that needs to be re-planned, the RFID tag is thrown onto the ground. In the subsequent inspection process, if the RFID tag is still identified by the device identification module, it is determined that the obstacle still exists, and there is no need to re-plan the route and directly use the backup inspection route. When the obstacle and the RFID tag are cleaned together, the tag cannot be identified, and the original planned route is used, reducing the number of planned routes and improving the inspection efficiency.

[0004] However, the prior art still has the following problems:

[0005] For irregular motion obstacles, their motion direction and speed frequently change, and traditional methods are difficult to accurately predict the trajectory, which can easily lead to collision or overly conservative detouring. SUMMARY

[0006] Therefore, the present application provides an intelligent navigation method for a robot to overcome the problem that for irregular motion obstacles, their motion direction and speed frequently change, and traditional methods are difficult to accurately predict the trajectory, which can easily lead to collision or overly conservative detouring.

[0007] To achieve the above purpose, the present application provides an intelligent navigation method for a robot. It includes:

[0008] In step S1, an obstacle image in a robot running track is acquired in real time, the acquired obstacle image is preprocessed, motion features of the obstacle are extracted, and a motion type of the obstacle is determined based on the motion features of the obstacle.

[0009] In step S2, an obstacle avoidance mode of the robot is determined based on the determined motion type of the obstacle, including:

[0010] The passing mode of the robot is determined according to a relative motion distance between the obstacle and the robot and a detour difficulty of the robot bypassing the obstacle.

[0011] Or, the passing mode of the robot is determined according to a real-time relative distance when the robot reaches the running track of the obstacle with irregular motion and a running uncertainty representation value of the obstacle.

[0012] In step S3, when the passing mode is determined as the detour passing, a motion track of the obstacle with irregular motion is analyzed to determine a motion track range, and a detour track with the shortest detour distance and the smallest detour difficulty is selected based on a robot position.

[0013] In step S4, a passing track is determined according to the determined passing mode.

[0014] Further, in step S1, the motion type of the obstacle is determined based on the motion features of the obstacle, including:

[0015] Based on the preprocessed obstacle image, motion feature data of the obstacle at a plurality of continuous time nodes is extracted at a predetermined interval to obtain a time sequence feature sequence.

[0016] The motion feature data in the time sequence feature sequence is analyzed and core feature data is calculated.

[0017] The core feature data is compared with a preset threshold value respectively.

[0018] Based on the comparison result, it is judged whether there is irregular core feature data, and when there is irregular core feature data, the motion type of the obstacle is determined as irregular motion.

[0019] Further, the core feature data includes a velocity standard deviation, a track motion deviation and a direction change frequency.

[0020] Further, in step S2, the obstacle avoidance mode of the robot is determined based on the determined motion type of the obstacle, including:

[0021] If the motion type of the obstacle is regular motion, the passing mode of the robot is determined according to a relative motion distance between the obstacle and the robot and a detour difficulty of the robot bypassing the obstacle.

[0022] If the motion type of the obstacle is irregular motion, it is determined that the passing mode of the robot is determined according to the real-time relative distance when the robot reaches the running track of the obstacle with irregular motion and the running uncertainty representation value of the obstacle.

[0023] Further, in the step S2, the determination manner of the difficulty of the robot bypassing the obstacle is:

[0024] determining the maximum change of the actual passable space when the robot bypasses the obstacle,

[0025] determining the included angle between the running direction of the obstacle and the current running direction of the robot to obtain the bypass path curvature,

[0026] calculating the ratio of the maximum change of the actual passable space to the preset maximum change to obtain a first sub-bypass difficulty parameter,

[0027] calculating the ratio of the bypass path curvature to the preset bypass path curvature to obtain a second sub-bypass difficulty parameter,

[0028] weighting and summing the first bypass difficulty parameter and the second sub-bypass difficulty parameter to obtain the bypass difficulty.

[0029] Further, in the step S2, the passing mode of the robot is determined according to the relative motion distance between the obstacle and the robot and the bypass difficulty of the robot bypassing the obstacle, which comprises:

[0030] if the relative motion distance is greater than or equal to a preset relative motion distance, it is determined that the robot passes according to the original driving path;

[0031] if the relative motion distance is less than the preset relative motion distance and the bypass difficulty is less than or equal to a preset bypass difficulty, it is determined that the robot bypasses;

[0032] if the relative motion distance is less than the preset relative motion distance and the bypass difficulty is greater than the preset bypass difficulty, it is determined that the robot waits to pass.

[0033] Further, in the step S2, the passing mode of the robot is determined according to the real-time relative distance when the robot reaches the running track of the obstacle with irregular motion and the running uncertainty representation value of the obstacle, which comprises:

[0034] if the real-time relative distance is greater than a preset real-time relative distance, it is determined that the robot passes according to the original driving path;

[0035] if the real-time relative distance is less than the preset real-time relative distance and the running uncertainty representation value of the obstacle is less than or equal to a preset running uncertainty representation value, it is determined that the robot bypasses;

[0036] If the real-time relative distance is less than the preset real-time relative distance and the operation uncertainty characteristic value of the obstacle is greater than the preset operation uncertainty characteristic value, it is determined that the robot waits to pass.

[0037] Further, the determination of the operation uncertainty characteristic value of the obstacle comprises:

[0038] calculating a ratio of a rate of change of direction of the obstacle to a preset rate of change of direction to obtain a first operation uncertainty sub-parameter,

[0039] calculating a ratio of a rate of change of speed of the obstacle to a preset rate of change of speed to obtain a second operation uncertainty sub-parameter,

[0040] weighting and summing the first operation uncertainty sub-parameter and the second operation uncertainty sub-parameter to obtain the operation uncertainty characteristic value.

[0041] Further, the determination of the real-time relative distance comprises:

[0042] real-time acquisition of robot position data and obstacle position data, and calculation of a spatial straight-line distance based on the robot position data and the obstacle position data as an initial real-time relative distance;

[0043] acquisition of obstacle motion direction data and motion speed data, and calculation of a relative distance change amount per unit time;

[0044] correction of the initial real-time relative distance based on the relative distance change amount to obtain the real-time relative distance.

[0045] Further, in the step S3, the determination of the motion trajectory range comprises:

[0046] based on the preprocessed image data in step S1, extraction of a position coordinate sequence of the obstacle to form a discrete motion trajectory point set;

[0047] acquisition of maximum and minimum displacement amounts of each coordinate axis to obtain boundary points;

[0048] connection of each boundary point by a smooth curve to obtain the motion trajectory range;

[0049] based on the robot position, selection of a detour trajectory with the shortest detour distance and the smallest detour difficulty comprises:

[0050] determination of a starting point and an ending point based on the real-time position coordinates of the robot and the target coordinates;

[0051] generation of a plurality of candidate detour trajectories that meet the constraint conditions;

[0052] evaluation of the detour difficulty of the plurality of candidate trajectories, and selection of a candidate trajectory with the smallest detour difficulty as the detour trajectory.

[0053] Compared with the prior art, the present application has the beneficial effects that, in the present application, by preprocessing the image, extracting the motion features (such as the rate of change of direction and the rate of change of speed), and classifying the motion type (regular / irregular motion), the problem of "ambiguous identification of irregular obstacles" in the traditional method is solved, and the error decision caused by misjudgment of the obstacle type is avoided. The present application adopts differentiated judgment logic for different motion type obstacles: for regular motion obstacles, the passing mode is determined by comprehensively determining the relative motion distance and the difficulty of detouring (such as directly detouring when the distance is sufficient and the difficulty of detouring is low, otherwise waiting), avoiding excessive conservatism, on the other hand, for irregular motion obstacles, the "running uncertainty representation value" is introduced to quantify the motion confusion degree, combined with real-time relative distance judgment (such as detouring when the uncertainty value is lower than the preset threshold and the distance is sufficient, otherwise waiting), to ensure the safety of detouring, preventing both excessive avoidance of low-risk obstacles (such as a slow-walking pedestrian) leading to decreased efficiency, and forcibly detouring high-risk obstacles (such as a fast-changing pet) causing a collision.

[0054] Further, in the present application, by "extracting feature data of continuous time nodes at equal intervals" (such as extracting once every 0.5 seconds), a time sequence feature sequence is formed, rather than relying on single frame or short time data, which can capture the "dynamic trend" of obstacle motion (such as a pedestrian gradually changing from regular walking to random pausing), avoiding misjudgment due to instantaneous data (such as a short pause in regular motion being directly classified as irregular), and is especially suitable for scenes with gradually changing motion states (such as a slowly changing direction robot).

[0055] Further, in the present application, it is considered that the relative motion distance can directly reflect the collision risk, and the difficulty of detouring (combined with the change of passable space and path curvature) ensures the feasibility of detouring (directly detouring when the difficulty is low, avoiding unnecessary waiting), so the relative motion distance and the difficulty of detouring are comprehensively analyzed for regular motion obstacles, thereby improving the control accuracy for regular motion obstacles, avoiding excessive conservative treatment of regular motion obstacles, and improving the passing efficiency under the premise of safety. On the other hand, considering that the real-time relative distance ensures the physical safety buffer when the robot reaches the obstacle trajectory (sufficient distance can be straight, without additional operation), the running uncertainty representation value is determined by the rate of change of direction and the rate of change of speed, and then the "confusion degree" of the obstacle motion is measured (when the uncertainty value is low, detouring is possible, and when the uncertainty value is high, waiting is necessary, avoiding collision caused by prediction failure), which balances safety and smoothness of passing.

[0056] Further, in the present application, the fluctuation degree of the space width is converted into a calculable value by the "ratio of the preset maximum change amount to the actual change amount", and the path bending degree is quantified by the "ratio of the actual bending curvature to the preset curvature", avoiding the subjectivity of the traditional "judging difficulty by experience", making the detour difficulty standardized and reproducible, providing a unified basis for the subsequent "straight line / detour / waiting" decision, and comprehensively evaluating the detour difficulty by assigning weights to the two sub-parameters (such as higher weight for space change in narrow scenes and higher weight for curvature in open scenes), focusing on both space stability (avoiding "can pass but space changes frequently leading to collision") and path feasibility (avoiding "space is enough but path is too curved and too sharp leading to loss of control"), achieving balanced control of multi-dimensional risks and reducing the collision risk in the detour process.

[0057] Further, in the present application, when the relative motion distance is greater than the preset value, it is directly determined that "the original path is passable" without additional operation, and when the distance is insufficient, the detour difficulty is compared with the preset difficulty to distinguish between passable and waiting, and when the detour difficulty is small, it indicates that the space around the obstacle is stable (small change amount) and the path is gentle (low curvature), and the robot can safely detour, while when the detour difficulty is large, it indicates that the space changes dramatically or the path is too curved and too sharp, and forced detour may cause collision, thus avoiding the limitations of single index judgment through the hierarchical logic of first looking at the distance and then looking at the difficulty, making the decision more comprehensive and safe.

[0058] Further, in the present application, for regular obstacles with sufficient distance (such as a tracking robot moving at a constant speed far away), the original path is directly passed to avoid unnecessary detour or waiting, improving the efficiency of passing, for obstacles with insufficient distance but low detour difficulty (such as a shelf moving slowly with sufficient space on both sides), the shortest path is detoured, which does not affect the efficiency and avoids task delay caused by waiting, and for obstacles with insufficient distance and high detour difficulty (such as a mechanical arm moving at high speed and with narrow space around), waiting is selected to pass to ensure safety and avoid equipment damage or collision caused by forced detour. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 Flowchart of the intelligent navigation method for robots according to the present application;

[0060] Figure 2 Flowchart for determining the obstacle avoidance mode of the robot;

[0061] Figure 3 Flowchart for determining the passing mode of the regular motion robot;

[0062] Figure 4 Flowchart for determining the passing mode of the irregular motion robot. DETAILED DESCRIPTION

[0063] In order to make the objects and advantages of the present application clearer, the following further describes the present application with reference to the embodiments; it should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0064] It should be noted that the data in the present embodiment are obtained by comprehensively analyzing and evaluating the historical data and the corresponding historical determination results of the present judgment in the past 6 months through the system of the present application. Those skilled in the art can understand that the determination mode of the system of the present application for a single parameter can be to select the value with the highest proportion as the preset standard parameter according to the data distribution, to use weighted summation to obtain the value as the preset standard parameter, to substitute each historical data into a specific formula and to obtain the value by using the formula as the preset standard parameter, or other selection modes, as long as the system of the present application can clearly define different specific conditions in the single determination process through the obtained value.

[0065] The preferred embodiments of the present application are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application and are not intended to limit the protection scope of the present application.

[0066] It should be noted that in the description of the present application, the terms of "upper", "lower", "left", "right", "inner", "outer" and the like indicating the direction or positional relationship are based on the direction or positional relationship shown in the drawings, which is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present application.

[0067] In addition, it should be further noted that in the description of the present application, unless otherwise explicitly specified and limited, the terms of "mounting", "connecting", "connection" should be understood in a broad sense, for example, it can be fixed connection, or detachable connection, or integral connection; it can be mechanical connection, or electrical connection; it can be direct connection, or indirect connection through an intermediate medium, or the communication inside two elements. Those skilled in the art can understand the specific meaning of the above terms in the present application according to the specific circumstances.

[0068] Please refer to Figure 1 Fig. 1 shows a flowchart of the intelligent navigation method for robots of the present application.

[0069] The intelligent navigation method for robots provided in the present embodiment comprises:

[0070] In step S1, an obstacle image in a robot running track is acquired in real time, the acquired obstacle image is preprocessed, a motion feature of the obstacle is extracted, and a motion type of the obstacle is determined based on the motion feature of the obstacle.

[0071] In step S2, an obstacle avoidance mode of the robot is determined based on the determined motion type of the obstacle, and the obstacle avoidance mode of the robot comprises:

[0072] According to the relative motion distance between the obstacle and the robot and the difficulty of the robot bypassing the obstacle, a passing mode of the robot is determined,

[0073] Or, according to the real-time relative distance when the robot reaches the running track of the irregular motion obstacle and the running uncertainty representation value of the obstacle, the passing mode of the robot is determined.

[0074] In step S3, when the passing mode is determined as bypassing, the motion track of the irregular motion obstacle is analyzed to determine the motion track range, and a bypassing track with the shortest bypassing distance and the smallest bypassing difficulty is selected based on the position of the robot.

[0075] In step S4, a passing track is determined according to the determined passing mode.

[0076] Specifically, in the embodiment, the position data of the obstacle in the obstacle image is acquired by a laser radar or a depth camera.

[0077] In the present application, by preprocessing the image, extracting the motion feature (such as the direction change rate and the speed change rate), and classifying the motion type (regular / irregular motion), the problem of "ambiguous identification of irregular obstacles" in the traditional method is solved, and the error decision caused by misjudgment of the obstacle type is avoided. The present application adopts differentiated judgment logic for different motion type obstacles: for regular motion obstacles, the passing mode is determined by comprehensively considering the relative motion distance and the bypassing difficulty (such as directly bypassing when the distance is sufficient and the bypassing difficulty is low, otherwise waiting), which avoids excessive conservatism. On the other hand, for irregular motion obstacles, the "running uncertainty representation value" is introduced to quantify the motion confusion degree, and the real-time relative distance judgment is combined (such as bypassing when the uncertainty value is lower than the preset threshold and the distance is sufficient, otherwise waiting), which ensures the safety of bypassing, prevents excessive avoidance of low-risk obstacles (such as slowly walking pedestrians) to cause efficiency decline, and avoids forcibly bypassing high-risk obstacles (such as quickly changing direction pets) to cause collision.

[0078] Specifically, in step S1, the motion type of the obstacle is determined based on the motion feature of the obstacle, comprising:

[0079] Based on the preprocessed obstacle image, the motion feature data of the obstacle at a plurality of continuous time nodes is extracted at a predetermined interval to obtain a time sequence feature sequence.

[0080] analyze the motion feature data in the time sequence feature sequence and calculate core feature data;

[0081] compare the core feature data with preset threshold values respectively;

[0082] determine whether there is irregular core feature data based on the comparison results, and determine the motion type of the obstacle as irregular motion when there is irregular core feature data, and determine the motion type of the obstacle as regular motion when there is no irregular core feature data.

[0083] Specifically, the core feature data includes a speed standard deviation, a trajectory motion deviation, and a direction change frequency.

[0084] In the present application, the feature data of continuous time nodes is extracted at equal intervals (such as once every 0.5 seconds) to form a time sequence feature sequence, rather than relying on single-frame or short-time data, which can capture the "dynamic trend" of obstacle motion (such as a pedestrian gradually changing from regular walking to random pausing), avoid misjudgment due to instantaneous data (such as a short pause in regular motion being directly classified as irregular), and is especially suitable for scenarios where the motion state gradually changes (such as a robot slowly changing direction).

[0085] Please refer to Figure 2 , which is a determination flow chart for determining the obstacle avoidance mode of a robot.

[0086] Specifically, in step S2, the obstacle avoidance mode of the robot is determined based on the determined motion type of the obstacle, including:

[0087] If the motion type of the obstacle is regular motion, the passing mode of the robot is determined according to the relative motion distance between the obstacle and the robot and the difficulty of the robot bypassing the obstacle.

[0088] If the motion type of the obstacle is irregular motion, the passing mode of the robot is determined according to the real-time relative distance when the robot reaches the running track of the obstacle with irregular motion and the running uncertainty representation value of the obstacle.

[0089] In the present application, the relative motion distance can directly reflect the collision risk, and the detour difficulty (combined with the change of passable space and the path curvature) ensures the feasibility of detour (low difficulty for direct detour, avoiding unnecessary waiting), so the relative motion distance and the detour difficulty are comprehensively analyzed to improve the control accuracy of regular motion obstacles, avoid excessive conservative treatment of regular motion obstacles, improve the passing efficiency under the premise of safety, on the other hand, the real-time relative distance ensures the physical safety buffer when the robot reaches the obstacle track (the distance is sufficient, and the robot can go straight without additional operation), the running uncertainty value is determined by the direction change rate and the speed change rate, and then the "chaotic degree" of the obstacle motion is measured (low uncertainty value for detour, high uncertainty value for waiting, avoiding collision caused by prediction failure), which balances safety and smoothness of passing.

[0090] Specifically, in the step S2, the determination method of the detour difficulty of the robot detouring the obstacle is:

[0091] determining the maximum change amount of the actual passable space when the robot detours the obstacle,

[0092] determining the angle between the running direction of the obstacle and the current running direction of the robot to obtain the path curvature of the detour,

[0093] calculating the ratio of the maximum change amount of the actual passable space to the preset maximum change amount to obtain a first sub-detour difficulty parameter,

[0094] calculating the ratio of the path curvature of the detour to the preset path curvature of the detour to obtain a second sub-detour difficulty parameter,

[0095] weighting and summing the first detour difficulty parameter and the second sub-detour difficulty parameter to obtain the detour difficulty.

[0096] Specifically, in the present embodiment, the preset maximum change amount is determined by the following method: obtaining typical running scene data of a plurality of robots, counting the width change range of the passable space when the historical obstacle passes, and recording the maximum change value in the 95% confidence interval; based on the minimum safe passing width of the robot (its own width + safety margin, such as 0.5 meter robot needs + 0.2 meter safety distance, i.e. 0.7 meter), the maximum change value in the historical data is corrected: if the historical maximum change value exceeds the safe passing ability of the robot, the safe passing ability is taken as the upper limit; if it does not exceed, the historical value is kept and a 10%-20% redundancy is added; the corrected value is taken as the preset maximum change amount and stored in the robot control system, which can be dynamically adjusted according to the scene switching (such as from indoor to outdoor).

[0097] Specifically, in this embodiment, the preset detour path curvature is determined as follows: the robot's minimum turning radius Rmin is extracted (e.g., if the minimum turning radius of a wheeled robot is 1 meter, then its limit curvature K1 = 1 / Rmin = 1 rad / m); based on scene safety requirements, a safety factor α is set (usually 0.6~0.8, balancing flexibility and safety), then the basic curvature threshold K2 = α × K1; if the scene is an open space (e.g., a warehouse) with sparse obstacles, the safety factor can be increased (e.g., 0.8), allowing for a larger curvature (more flexible detour); if the scene is a narrow passage (e.g., a corridor), the safety factor needs to be reduced (e.g., 0.6), limiting the curvature (to avoid the path from going too far and hitting the wall); the corrected curvature value is used as the preset detour path curvature.

[0098] This invention transforms the fluctuation of spatial width into a calculable value by using the "ratio of the preset maximum change to the actual change," and quantifies the path curvature by using the "ratio of the actual detour curvature to the preset curvature." This avoids the subjectivity of traditional "judging difficulty based on experience," making the assessment of detour difficulty standardized and reproducible, and providing a unified basis for subsequent "go straight / detour / wait" decisions. By assigning weights to the two sub-parameters (e.g., spatial change has a higher weight in narrow scenarios, and curvature has a higher weight in open scenarios), the detour difficulty is comprehensively assessed. It considers both spatial stability (avoiding collisions caused by "being able to pass but the space being wide or narrow") and path feasibility (avoiding loss of control caused by "sufficient space but the path being too curved or sharp"). This achieves a balanced control of multi-dimensional risks, reducing the collision risk during the detour process.

[0099] Please see Figure 3 As shown, it is a flowchart for determining the passage mode of a robot with a defined regular movement.

[0100] Specifically, in step S2, the robot's passage method is determined based on the relative movement distance between the obstacle and the robot, as well as the difficulty of the robot bypassing the obstacle, including:

[0101] If the relative motion distance is greater than or equal to the preset relative motion distance, it is determined that the robot travels along the original path.

[0102] If the relative motion distance is less than the preset relative motion distance and the detour difficulty is less than or equal to the preset detour difficulty, then the robot is determined to have detoured successfully.

[0103] If the relative motion distance is less than the preset relative motion distance and the detour difficulty is greater than the preset detour difficulty, then the robot is determined to wait to pass.

[0104] Specifically, in this embodiment, the preset relative motion distance is determined by the following method: preset relative motion distance = shortest braking distance of robot + displacement of obstacle in robot braking time + safety buffer width of robot itself + environmental redundancy, for example: the maximum travel speed of the robot is 1.2 m / s, the shortest braking distance (from 1.2 m / s to stop) is 0.8 m (limited by ground friction and braking system), the width of the robot itself is 0.6 m, the safety buffer width is 0.5 m (to avoid edge collision), assuming that the obstacle is a pedestrian moving at a constant speed, the moving speed is 1.0 m / s, and the relative motion direction with the robot is perpendicular intersection (i.e. the pedestrian crosses the robot path), the robot braking response time is 0.3 s (the delay from detecting the obstacle to starting braking); the indoor environment is relatively narrow, and the sensor (such as camera, laser radar) has a measurement error of ±0.2 m, so the environmental redundancy is 0.3 m; the displacement of the obstacle in the robot braking time = obstacle speed x braking response time = 1.0 m / s x 0.3 s = 0.3 m; preset relative motion distance = 0.8 m (braking distance) + 0.3 m (obstacle displacement) + 0.5 m (safety buffer) + 0.3 m (environmental redundancy) = 1.9 m.

[0105] Specifically, in this embodiment, the preset detour difficulty is determined by the following method: taking an indoor service robot as an example, its running scene is a corridor with a width of 2-3 meters, and the hardware parameters are a minimum turning radius of 0.5 meters and a maximum turning angle of ±30°. According to the corridor width, the maximum safe change of passable space is set to 0.8 meters (to avoid collision due to sudden narrowing of space); according to the minimum turning radius, the preset detour path curvature is calculated to be 0.3 rad / m (corresponding to a turning angle of 20°, which is within the safe range of the robot's turning performance); in the indoor corridor scene, the change of passable space has a greater impact on safety, so the first sub-parameter (space change) weight is 0.6, and the second sub-parameter (path curvature) weight is 0.4; when the first sub-parameter (actual change / preset change) ≤1 and the second sub-parameter (actual curvature / preset curvature) ≤1, the detour difficulty is within the safe range; the critical value of the preset detour difficulty is: (1 x 0.6) + (1 x 0.4) = 1.0, i.e. when the calculated detour difficulty ≤1.0, the robot can safely detour; if >1.0, it is out of the safe range and needs to be waited.

[0106] When the relative motion distance is greater than the preset value, the original path is directly determined to be passable, without additional operation, and when the distance is insufficient, the difficulty of detouring is compared with the preset difficulty to distinguish between detouring and waiting, and when the detouring difficulty is small, it indicates that the space around the obstacle is stable (small change) and the path is flat (low curvature), and the robot can safely detour, and when the detouring difficulty is large, it indicates that the space changes dramatically or the path is too curved, and forced detouring may cause collision, through the hierarchical logic of first looking at the distance and then looking at the difficulty, the limitations of single index judgment are avoided, and the decision is more comprehensive and safe.

[0107] In the present application, for regular obstacles with sufficient distance (such as a tracking robot driving at a constant speed far away), the original path is directly determined to be passable, avoiding unnecessary detouring or waiting, improving the efficiency of passing, for obstacles with insufficient distance but low detouring difficulty (such as a slow-moving shelf with sufficient space on both sides), the shortest path is detoured, without affecting the efficiency, and avoiding task delay caused by waiting, for obstacles with insufficient distance and high detouring difficulty (such as a high-speed reciprocating mechanical arm with narrow space around), waiting is selected to pass, ensuring safety and avoiding equipment damage or collision caused by forced detouring.

[0108] Please refer to Figure 4 The determination flow chart of the passing mode of the robot with irregular motion is shown in the figure.

[0109] Specifically, in step S2, the passing mode of the robot is determined according to the real-time relative distance when the robot reaches the running track of the irregular motion obstacle and the running uncertainty representation value of the obstacle, comprising:

[0110] If the real-time relative distance is greater than the preset real-time relative distance, the robot is determined to pass according to the original driving path;

[0111] If the real-time relative distance is less than the preset real-time relative distance and the running uncertainty representation value of the obstacle is less than or equal to the preset running uncertainty representation value, the robot is determined to detour and pass;

[0112] If the real-time relative distance is less than the preset real-time relative distance and the running uncertainty representation value of the obstacle is greater than the preset running uncertainty representation value, the robot is determined to wait and pass.

[0113] Specifically, in the present embodiment, the preset real-time relative distance is determined by the following method: taking a warehouse logistics robot (running in a warehouse shelf channel, mainly avoiding irregular motion forklifts and personnel) as an example: the robot's own safe braking distance: 5m (speed 3m / s, braking acceleration 2m / s 2 , braking time 1.5s, calculated braking distance = 3x1.5-0.5x2x1.5 2= 3.75m, take the safety redundancy value 5m); irregular obstacles (person) maximum movement range estimation: the maximum lateral / longitudinal movement distance of the person within a short time (2s) is 3m (walking speed 1.5m / s x 2s); robot reaction delay: 0.5s, within which the obstacle can additionally move 0.75m (1.5m / s x 0.5s); preset real-time relative distance = safety braking distance + maximum movement range of the obstacle + displacement within the reaction delay = 5m + 3m + 0.75m = 8.75m.

[0114] Specifically, in this embodiment, the preset running uncertainty characteristic value is determined in the following manner: taking a shopping mall service robot (mainly avoiding irregularly moving pedestrians) as an example: the maximum direction change rate of the pedestrians in the shopping mall is 30° / s (such as suddenly turning around), considering the safety redundancy (+20%), the preset direction change rate threshold is set to 36° / s; the maximum speed change rate of the pedestrians is 0.8m / s 2 (such as sudden acceleration or deceleration), safety redundancy (+20%), preset speed change rate threshold is set to 0.96m / s 2 ; the direction change rate of the pedestrians is more affected by the sudden change, so the direction change rate weight = 0.6, the speed change rate weight = 0.4; the preset running uncertainty characteristic value = (36° / s x 0.6) + (0.96m / s 2 x 0.4) = 21.6 + 0.384 = 21.984.

[0115] Specifically, the determination of the running uncertainty characteristic value of the obstacle comprises:

[0116] calculating the ratio of the direction change rate of the obstacle to the preset direction change rate to obtain a first running uncertainty sub-parameter,

[0117] calculating the ratio of the speed change rate of the obstacle to the preset speed change rate to obtain a second running uncertainty sub-parameter,

[0118] weighting and summing the first running uncertainty sub-parameter and the second running uncertainty sub-parameter to obtain the running uncertainty characteristic value.

[0119] Specifically, in this embodiment, the preset direction change rate is determined in the following manner: taking a pedestrian as an example, the direction change of the pedestrian when walking normally is gentle, and the maximum direction change when suddenly avoiding or turning around is about 45° / s (such as turning from forward to side in 1s, with a deflection of 45°); considering the sensor direction detection error ±3° and the angle that the pedestrian can additionally deflect within the 0.2s reaction delay of the robot (45° / s x 0.2s = 9°), the total redundancy is 12°; the preset direction change rate is calculated: typical value + safety redundancy = 45° / s + 12° / s = 57° / s (rounded to 60° / s).

[0120] Specifically, the determination of the real-time relative distance includes:

[0121] Real-time acquisition of robot position data and obstacle position data, and calculation of the spatial straight-line distance as the initial real-time relative distance based on the robot position data and the obstacle position data;

[0122] Obtaining obstacle motion direction data and motion speed data, and calculating the relative distance change amount per unit time;

[0123] Correcting the initial real-time relative distance based on the relative distance change amount to obtain the real-time relative distance.

[0124] Specifically, in step S3, determining the motion trajectory range includes: based on the image data preprocessed in step S1, extracting the position coordinate sequence of the obstacle to form a discrete motion trajectory point set;

[0125] Obtaining the maximum displacement and the minimum displacement of each coordinate axis to obtain the boundary points;

[0126] Connecting each boundary point with a smooth curve to obtain the motion trajectory range;

[0127] Selecting the shortest and easiest detour trajectory based on the robot position includes:

[0128] Determining the starting point and the ending point based on the real-time position coordinates of the robot and the target coordinates;

[0129] Generating a plurality of candidate detour trajectories that meet the constraint conditions;

[0130] Evaluating the detour difficulty of a plurality of candidate trajectories, and selecting the candidate trajectory with the minimum detour difficulty as the detour trajectory.

[0131] So far, the technical solutions of the present application have been described in conjunction with the preferred embodiments shown in the drawings, but those skilled in the art will readily understand that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to related technical features without departing from the principles of the present application, and the technical solutions after such changes or replacements will fall within the protection scope of the present application.

[0132] The above description is only the preferred embodiments of the present application and is not intended to limit the present application; for those skilled in the art, the present application can have various changes and variations, and any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. An intelligent navigation method for a robot, characterized by, The method comprises the following steps: Step S1, real-time acquisition of obstacle images in the running track of the robot, pre-processing of the acquired obstacle images, extraction of the motion characteristics of the obstacles, and determination of the motion type of the obstacles based on the motion characteristics of the obstacles; Step S2, determination of the obstacle avoidance mode of the robot based on the determined motion type of the obstacles, comprising: determination of the passing mode of the robot according to the relative motion distance between the obstacle and the robot and the difficulty of the robot bypassing the obstacle; or, determination of the passing mode of the robot according to the real-time relative distance when the robot reaches the running track of the irregular motion obstacle and the running uncertainty representation value of the obstacle; Step S3, analysis of the motion track of the irregular motion obstacle to determine the motion track range when the passing mode is determined to be bypassing, and selection of the bypassing track with the shortest bypassing distance and the smallest bypassing difficulty based on the position of the robot; Step S4, determination of the passing track according to the determined passing mode; The determination of the running uncertainty representation value of the obstacle comprises: calculating the ratio of the direction change rate of the obstacle to the preset direction change rate to obtain a first running uncertainty sub-parameter, calculating the ratio of the speed change rate of the obstacle to the preset speed change rate to obtain a second running uncertainty sub-parameter, weighting and summing the first running uncertainty sub-parameter and the second running uncertainty sub-parameter to obtain the running uncertainty representation value. 2.The intelligent navigation method for robots according to claim 1, characterized in that, In step S1, the motion type of the obstacle is determined based on the motion characteristics of the obstacle, comprising: based on the pre-processed obstacle image, extracting the motion characteristic data of the obstacle at a plurality of consecutive time nodes according to a predetermined interval to obtain a time sequence characteristic sequence; analyzing the motion characteristic data in the time sequence characteristic sequence and calculating core characteristic data; comparing the core characteristic data with the preset threshold value respectively; judging whether there is irregular core characteristic data based on the comparison result, and determining the motion type of the obstacle as irregular motion when there is irregular core characteristic data. 3.The intelligent navigation method for robots according to claim 2, wherein, The core characteristic data includes speed standard deviation, track motion deviation, and direction change frequency. 4.The intelligent navigation method for robots according to claim 2, wherein, In step S2, the obstacle avoidance mode of the robot is determined based on the determined motion type of the obstacle, comprising: if the motion type of the obstacle is regular motion, then determine the passing mode of the robot according to the relative motion distance between the obstacle and the robot and the bypassing difficulty of the robot bypassing the obstacle; if the motion type of the obstacle is irregular motion, then determine the passing mode of the robot according to the real-time relative distance when the robot reaches the running track of the irregular motion obstacle and the running uncertainty representation value of the obstacle. 5.The intelligent navigation method for robots according to claim 4, wherein, In step S2, the bypassing difficulty of the robot bypassing the obstacle is determined in the following manner: determining the maximum change of the actual passable space when the robot bypasses the obstacle, determining the included angle between the running direction of the obstacle and the current running direction of the robot to obtain the bypassing path curvature, calculating the ratio of the maximum change of the actual passable space to the preset maximum change to obtain a first sub-bypassing difficulty parameter, calculating the ratio of the bypassing path curvature to the preset bypassing path curvature to obtain a second sub-bypassing difficulty parameter, The first detour difficulty parameter and the second sub-detour difficulty parameter are weighted and summed to obtain a detour difficulty. 6.The intelligent navigation method for robots according to claim 5, wherein, In the step S2, the passing mode of the robot is determined according to the relative motion distance between the robot and the obstacle and the detour difficulty of the robot in detouring the obstacle, and the passing mode of the robot comprises: If the relative motion distance is greater than or equal to a preset relative motion distance, it is determined that the robot passes according to the original driving path; If the relative motion distance is less than the preset relative motion distance and the detour difficulty is less than or equal to a preset detour difficulty, it is determined that the robot detours to pass; If the relative motion distance is less than the preset relative motion distance and the detour difficulty is greater than the preset detour difficulty, it is determined that the robot waits to pass. 7.The intelligent navigation method for robots according to claim 4, wherein, In the step S2, the passing mode of the robot is determined according to the real-time relative distance when the robot reaches the running track of the irregular motion obstacle and the running uncertainty representation value of the obstacle, and the passing mode of the robot comprises: If the real-time relative distance is greater than a preset real-time relative distance, it is determined that the robot passes according to the original driving path; If the real-time relative distance is less than the preset real-time relative distance and the running uncertainty representation value of the obstacle is less than or equal to a preset running uncertainty representation value, it is determined that the robot detours to pass; If the real-time relative distance is less than the preset real-time relative distance and the running uncertainty representation value of the obstacle is greater than the preset running uncertainty representation value, it is determined that the robot waits to pass. 8.The intelligent navigation method for robots according to claim 7, wherein, The determination of the real-time relative distance comprises: Real-time collection of robot position data and obstacle position data, and calculation of a spatial straight line distance as an initial real-time relative distance based on the robot position data and the obstacle position data; Obtaining of obstacle motion direction data and motion speed data, and calculation of a relative distance change amount per unit time; Based on the relative distance change amount, the initial real-time relative distance is corrected to obtain the real-time relative distance. 9.The intelligent navigation method for robots of claim 1, wherein, In the step S3, the determination of the motion track range comprises: Based on the preprocessed image data in the step S1, a position coordinate sequence of the obstacle is extracted to form a discrete motion track point set; Obtaining of maximum and minimum displacement amounts of each coordinate axis to obtain boundary points; Connecting of each boundary point by a smooth curve to obtain the motion track range; Based on the real-time position coordinate of the robot and the target coordinate, the start point and the end point are determined; Generation of a plurality of candidate detour tracks meeting the constraint conditions; Evaluation of the detour difficulty of the plurality of candidate tracks, and selection of a candidate track with the minimum detour difficulty as the detour track. ​

Images