Self-adaptive obstacle crossing control method for electrically-driven column type wheel-foot robot based on advancing path recognition

Abstract

The invention relates to the technical field of robot control, in particular to an electric drive column type wheel-foot robot self-adaptive obstacle crossing control method based on advancing path recognition, and the method comprises the following steps: S1, environment perception and terrain map generation: generating a terrain map; s2, mode switching and judgment: according to the topographic map, in combination with a preset wheel-foot mode switching threshold value, judging that the current obstacle type is a climbing platform, a slope ridge or a crossing ditch; s3, setting each mode parameter; s4, dynamic stability checking and correction preprocessing: according to the current gravity center position of the robot, the obstacle crossing stability is evaluated, and if the overturning risk exists, setting is automatically adjusted. According to the method, typical obstacle types such as platform climbing, ditch crossing and slope ridge can be accurately identified, and compared with a traditional method which only depends on a single sensor or manually presets a threshold value, the method has higher environmental adaptability and obstacle crossing scene judgment precision.

Description

Technical Field

[0001] The present invention relates to the technical field of robot control, and particularly to an adaptive obstacle-crossing control method for an electric-driven columnar wheel-legged robot based on traveling path recognition. Background Art

[0002] Due to the combination of the high-efficiency movement of wheels and the strong obstacle-crossing ability of legs, wheel-legged hybrid robots are widely used in reconnaissance, search and rescue, and survey tasks in complex terrain environments. However, in actual operation, the following technical limitations generally exist in existing wheel-legged robots: On the one hand, traditional obstacle-crossing control methods mostly rely on preset fixed strategies and cannot perform dynamic recognition and mode switching according to the actual distribution characteristics of different types of obstacles in the path (such as steps, grooves or slopes). Especially in unstructured terrains, the recognition accuracy of obstacle types is insufficient, which easily leads to the failure of obstacle-crossing actions or the mis-triggering of control instructions.

[0003] On the other hand, obstacle-crossing parameters generally adopt fixed or manual adjustment methods, such as setting a unified foot-end lifting height, front-wheel opening angle, etc. There is a lack of real-time analysis of key physical properties such as terrain slope and surface roughness, resulting in a mismatch between control parameters and the environment, and affecting the success rate of obstacle crossing and operation stability.

[0004] In addition, there are deficiencies in the passive response of attitude stability control. When the robot has an overturning risk such as a center-of-mass offset or an unbalanced support structure during the obstacle-crossing process, traditional systems usually rely on global replanning or delay response mechanisms and are difficult to perform local attitude correction in a timely manner, increasing the operation risk. Summary of the Invention

[0005] The present invention provides an adaptive obstacle-crossing control method for an electric-driven columnar wheel-legged robot based on traveling path recognition, which is an adaptive obstacle-crossing control method combining multi-source environment perception, path obstacle recognition, and dynamic control parameter adjustment to improve the intelligent response ability and obstacle-crossing safety of the wheel-legged robot in complex paths.

[0006] An adaptive obstacle-crossing control method for an electric-driven columnar wheel-legged robot based on traveling path recognition includes the following steps: S1: Environment perception and terrain map generation: Real-time collect three-dimensional terrain feature parameters of the traveling path through a front-view lidar and a foot-end pressure sensor, including obstacle height distribution, surface roughness, and slope change, and dynamically generate a terrain map in combination with visual recognition to perceive the height of the front step or the width of the groove, for determining whether to enter the control mode of climbing a platform, crossing a ditch, or a slope; S2: Mode switching and determination: According to the terrain map, in combination with a preset wheel-leg mode switching threshold, determine the current obstacle type as climbing a platform, a slope, or crossing a ditch; S3: Setting of each mode parameter: In the slope - step mode, the dynamic friction coefficient between each foot end and the slope - step ground is calculated based on the slope and surface roughness, and the safety threshold of the foot - end pressure is set in combination with the target action mode: In the platform - climbing mode, a wheel - foot lifting and propulsion command is generated based on the terrain map, the hydraulic cylinder is controlled, the front - wheel support height is preferentially lifted, and the wheel - feet are lifted and extended in sequence to climb the platform; In the ditch - crossing mode, a wheel - foot rotation and propulsion command is generated based on the terrain map, the front - wheel opening angle is extended to more than 50 degrees, and the wheel - feet rotate and advance in sequence to cross the obstacle; S4: Dynamic stability check and correction pre - processing: According to the current center - of - gravity position of the robot, the over - obstacle stability is evaluated. If there is a risk of tipping, it is automatically adjusted and set.

[0007] Optionally, the specific steps of S1 include: S11, perform three - dimensional point - cloud modeling on the traveling path through the front - view lidar, extract the height distribution of obstacles and slope changes, and update the terrain contour in real - time through the point - cloud registration algorithm; S12: Collect the contact - force signals between each wheel - foot and the ground through the foot - end pressure sensor, calculate the surface roughness based on the contact - force variance, and determine that the current area is a high - roughness terrain when the contact - force variance exceeds the preset roughness threshold; S13: Identify the geometric features of the front - step or trench through the binocular vision sensor, perform sub - pixel - level measurement on the step height or trench width, and segment the step / trench boundary based on the edge - detection algorithm.

[0008] Optionally, S1 further includes fusing multi - source data such as the obstacle height distribution, slope - change parameters extracted by the lidar, the surface roughness calculated by the foot - end pressure sensor, and the step height or trench width measured by the vision sensor to generate a three - dimensional grid map including elevation, roughness, and obstacle type as the terrain map.

[0009] Optionally, in S2: According to the comparison results of the ratio of the obstacle height in the terrain map to the maximum lifting height of the robot's wheel - feet, the ratio of the trench width to the maximum span of the wheel - feet, and the slope to the wheel - foot tipping threshold, the platform - climbing, ditch - crossing, or slope - step mode is dynamically triggered.

[0010] Optionally, in S2: If it is identified as a "slope - step type" obstacle, start the normal - passing process; If it is identified as a "step type" obstacle, start the platform - climbing action process; If it is identified as a "trench type" obstacle, start the ditch - crossing action process.

[0011] Optionally, the parameter setting in the slope - step mode includes: According to the slope angle and surface roughness level in the terrain map, based on a preset friction coefficient mapping table, calculate the real-time friction coefficient between each foot end and the slope ground ; Based on the friction coefficient and the current load weight of the robot, calculate the minimum anti-slip pressure threshold of each wheel-foot end through the static equilibrium equation , denote the minimum anti-slip pressure threshold of the i-th wheel-foot end, and set the pressure safety threshold of the wheel-foot end

[0012] Optionally, the parameter setting in the climbing platform mode includes: According to the step height H in the terrain map, generate a wheel-foot lifting and advancing instruction, and control the hydraulic cylinder according to , is the lifting margin, lift the front wheel support height, and at the same time, based on the projection position of the robot's center of mass, plan the lifting timing of the rear wheel-feet, so that the wheel-feet climb over the step in the action sequence of "front wheel lift → front wheel extend → rear wheel follow-up"

[0013] Optionally, the parameter setting in the ditch-crossing mode includes: According to the groove width W in the terrain map, generate a wheel-foot rotation and advancing instruction, control the expansion of the front wheel angle, and calculate the rotation angle and advancing speed of each wheel-foot based on the inverse kinematics module, so that the wheel-feet perform the crossing actions of "front wheel rotate and cut into the groove edge → middle and rear wheels synchronously advance → front wheel reset" in sequence

[0014] Optionally, S4 includes constructing a support polygon projection boundary according to the current position coordinates and contact states of each wheel-foot, and obtaining the center of gravity position coordinates in real time through the robot's center of mass sensor; if the center of gravity position coordinates exceed the support polygon projection boundary, it is determined that there is a risk of tipping

[0015] Optionally, in S4, after determining that there is a risk of tipping, insert a buffer support section in the preset action sequence, control at least a pair of diagonal wheel-feet to touch the ground synchronously and increase the foot end pressure until the center of gravity returns within the support polygon projection boundary

[0016] Advantages of the present invention: In the present invention, by fusing lidar three-dimensional point cloud, foot end pressure signal and binocular vision image, a multi-source three-dimensional terrain map with elevation, slope, roughness and geometric structure information is constructed, which can accurately identify typical obstacle types such as climbing platforms, ditch-crossing and slope banks. Compared with the traditional method that only relies on a single sensor or manually preset thresholds, it has higher environmental adaptability and obstacle-crossing scenario judgment accuracy

[0017] In the present invention, by comparing the ratio of obstacle geometric features with the robot threshold, corresponding action modes are automatically triggered, and exclusive parameters are set for different modes: such as dynamically estimating the friction coefficient based on the slope-roughness mapping table, deriving the anti-slip pressure threshold according to the static force model, calculating the opening angle and the forward extension distance, etc. This mechanism ensures the precise correspondence between the control strategy and the terrain features, and improves the stability of obstacle crossing execution and the energy consumption control efficiency.

[0018] In the present invention, a support polygon projection construction and torque balance evaluation mechanism is introduced. Combining real-time monitoring of the centroid coordinates, when the risk of tipping is detected, a buffer support section is automatically inserted or the hydraulic pressure distribution is adjusted to ensure the stable control of the overall posture of the robot during the obstacle crossing process. This strategy enhances the redundant fault tolerance ability in multi-legged collaborative actions and reduces the probability of obstacle crossing failure caused by terrain mutations or action interferences. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only those of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0020] Figure 1 It is a schematic flow chart of the control method according to an embodiment of the present invention; Figure 2 It is a schematic diagram of mode switching and determination according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] The present invention will be described in detail below in conjunction with the drawings and specific embodiments. At the same time, it should be noted here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments. For some well-known technologies, those skilled in the art can also adopt other alternative methods for implementation; and the drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0022] It should be pointed out that in the specification, it is mentioned that "an embodiment", "embodiments", "exemplary embodiments", "some embodiments", etc. indicate that the described embodiments may include specific features, structures or characteristics, but not necessarily every embodiment includes such specific features, structures or characteristics. In addition, when combining embodiments to describe specific features, structures or characteristics, implementing such features, structures or characteristics in combination with other embodiments (whether explicitly described or not) should be within the knowledge scope of those skilled in the relevant art.

[0023] Generally, terms can be understood at least in part from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or property in the singular sense, or can be used to describe a combination of features, structures, or properties in the plural sense. Additionally, the term "based on" can be understood to not necessarily be intended to convey an exclusive set of factors, but rather can alternatively, depending at least in part on the context, allow for the existence of other factors that are not necessarily explicitly described.

[0024] As Figure 1 - Figure 2 shown, an adaptive obstacle-crossing control method for an electric drive column-wheel-legged robot based on travel path recognition includes the following steps: S1: Environmental perception and terrain map generation: Real-time collect three-dimensional terrain feature parameters of the travel path through a forward-looking lidar and a foot-end pressure sensor, including obstacle height distribution, surface roughness, and slope change, and combine visual recognition to perceive the height of the front step or the width of the groove, and dynamically generate a terrain map for determining whether to enter the platform climbing, ditch crossing, or slope control mode; S2: Mode switching and determination: According to the terrain map, combined with a preset wheel-foot mode switching threshold, determine the current obstacle type as platform climbing, slope, or ditch crossing; the wheel-foot mode switching threshold includes the ratio threshold of the obstacle height to the maximum lifting height of the robot's wheel feet, the ratio threshold of the groove width to the maximum span of the wheel feet, and the wheel-foot tipping threshold; S3: Setting of each mode parameter: In the slope mode, calculate the dynamic friction coefficient between each foot end and the slope ground based on the slope and surface roughness, and combine the target action mode to set the safety threshold of the foot end pressure: In the platform climbing mode, generate a wheel-foot lifting and propulsion instruction based on the terrain map, control the hydraulic cylinder, and preferentially lift the front wheel support height to make the wheel feet lift and extend in sequence for platform climbing; In the ditch crossing mode, generate a wheel-foot rotation and propulsion instruction based on the terrain map, expand the front wheel angle to more than 50 degrees, and the wheel feet rotate and advance in sequence to cross the obstacle; S4: Dynamic stability check and correction preprocessing: Evaluate the obstacle-crossing stability according to the current center of gravity position of the robot, and automatically adjust the setting if there is a tipping risk.

[0025] S1 specifically includes: S11, use a forward-looking lidar to perform three-dimensional point cloud modeling on the travel path at a scanning frequency, extract the height distribution of the obstacles and the slope change parameters , and use the iterative closest point registration method (ICP) to register adjacent frame point clouds to update the terrain contour in real time: Represents the point cloud obtained by the lidar at the current moment, represents the point cloud at the previous moment, represents the Iterative Closest Point algorithm, represents the registered point cloud, which is used to construct the continuous terrain contour.

[0026] S12. Collect the contact force signals between each wheel foot and the ground through the foot-end pressure sensor , and calculate its variance within a certain time window , which is used as the basis for judging surface roughness: ; When the following condition is met: , it is determined that the current position is a terrain with high roughness. Among them, represents the contact force of the i-th wheel foot at time t, represents the average value of the contact force, N represents the number of sampling points, represents the roughness judgment threshold, and its value range is 15 - 25 N².

[0027] S13. Identify the contour of the obstacle in front through the binocular vision sensor, and use the edge detection and sub-pixel precise positioning algorithm to extract the boundaries of steps or grooves, and calculate their geometric features: Step height is calculated from the depth difference of image stereo matching; Groove width is obtained from the difference in the left and right edge positions.

[0028] , ; Among them, are the depth values of the upper and lower edges of the step in the image depth map, is the horizontal position of the groove edge in the image coordinate system.

[0029] S14. Perform multi-source fusion on the from the lidar, from the foot-end pressure sensor, from visual recognition, and adopt a weighted fusion mechanism to construct a three-dimensional grid terrain map : , where each grid contains the elevation, slope, roughness and obstacle geometric attributes of the local terrain.

[0030] In S2, according to the ratio of each parameter in the terrain map to the robot's ability boundary, judge the obstacle type and dynamically trigger the control mode: If: Trigger the climbing platform mode; If: Trigger the crossing groove mode; If: The slope and step mode is triggered; wherein, is the maximum lifting height of the wheel-foot, is the maximum span of the wheel-foot, is the maximum anti-overturning safety slope threshold of the robot, set to 25° - 28°, is the control mode trigger threshold, is set to 0.8, that is, the climbing mode is triggered when the step height reaches 80% of the maximum lifting capacity, is set to 0.7, that is, the trench crossing mode is triggered when the trench width reaches 70% of the maximum span.

[0031] S3 specifically includes: Slope and step mode parameter setting:

[0032] S31, according to the slope angle and surface roughness level in the terrain map, query and dynamically obtain the real-time friction coefficient between the foot end and the ground from the preset friction coefficient mapping table: ; wherein, is the friction coefficient between the foot end and the ground, represents the mapping table from slope and roughness to friction coefficient, is the contact surface roughness level.

[0033] S32, based on the obtained friction coefficient and the current load mass M of the robot, calculate the minimum anti-slip pressure threshold of each foot end according to the static equilibrium model: ; Set the actual foot end pressure safety threshold as: ; wherein, represents the minimum anti-slip pressure of the i-th wheel-foot, is the corresponding safety foot end support pressure threshold, M is the total mass of the robot, g is the acceleration due to gravity, n is the number of stressed wheel-feet, is the current slope of the slope surface, is the safety margin coefficient.

[0034] The above-mentioned preset friction coefficient mapping table is used to estimate the friction coefficient according to the slope angle and surface roughness level , and its content structure is as follows: Input parameter discretization: Slope angle : Divided into six gears of 5°, 10°, 15°, 20°, 25°, 30°; Surface roughness : Divided into three grades according to the variance of contact force: Low roughness N ; Medium roughness ([[]] N ; High roughness N ; Table 1 Preset friction coefficient mapping table

[0035] The specific derivation of the minimum anti-slip pressure threshold of each foot end calculated by the static equilibrium model is as follows: a Basic settings: The terrain is an inclined plane with a slope of ; The total mass of the robot is M, and the acceleration due to gravity is g; The number of foot ends is n, and the force is evenly distributed; The friction coefficient is (obtained from the aforementioned mapping table); b Force analysis: To prevent sliding, the pressure provided by each wheel foot should satisfy the balance between the frictional force and the downward sliding force: , represents the magnitude of the frictional force required to prevent sliding for the i-th wheel foot in the slope mode; Therefore, the minimum anti-slip pressure is: ; And set the upper limit of the safety pressure: ; If the slope increases or the roughness decreases (resulting in decreases), then automatically increases to avoid the wheel feet from sliding down or being overloaded, ensuring stability and execution efficiency in the slope mode. Parameter setting for the climbing platform mode:

[0036] S33, according to the step height H extracted from the terrain map, generate the wheel foot lifting and propulsion commands, and set the lifting height to: ; At the same time, based on the projection position of the robot's center of mass plan the timing of the rear wheel lifting and propulsion, and form the following action sequence: 1. The front wheel is lifted to ; 2. The front wheel extends forward to the step surface; 3. The rear wheel follows and climbs.

[0037] Among them, H is the actual step height, is the lifting margin, is the target lifting height, is the lifting margin ratio, is the centroid position of the robot, obtained by the robot centroid sensor. Cross-gutter mode parameter setting:

[0038] S34. Calculate the required front-wheel opening angle according to the gutter width W in the terrain map Satisfy: , and ; and call the inverse kinematics module to calculate the rotation angle of each wheel foot and the propulsion speed , generating the following phased actions: 1. The front wheels rotate and cut into the edge of the gutter; 2. The middle and rear wheels advance synchronously; 3. The front wheels reset to restore the default posture.

[0039] Among them, W is the gutter width, L is the wheel foot length, is the front-wheel opening angle, , is the rotation angle of the i-th wheel foot, is the propulsion speed of the i-th wheel foot, is the minimum opening angle for the crossing action (about 50°).

[0040] The inverse kinematics module is specifically used to convert control intentions (such as wheel foot position, opening angle, propulsion speed) into actual executable drive instructions. The core includes forward kinematics modeling, Jacobian matrix construction, and inverse solution calculation.

[0041] 1. Establish a wheel foot structure model Each wheel foot model includes: Revolute joint (for opening angle expansion); Prismatic or slide rail joint (for forward extension / retraction); The position of the end effector of the support arm is the end pose .

[0042] 2. Forward kinematics expression: Calculate the end position through the joint angles : ; 3. Solve the inverse solution: Let the target position be , actually: ; If it is a redundant system, use the pseudo-inverse solution method: ; Among them, is the Jacobian matrix, is the Moore-Penrose pseudo-inverse, is the end desired speed, is the commanded speed for each joint.

[0043] 4. Implementation steps: Generate the desired end-effector trajectory according to the path planning; Calculate the Jacobian and its pseudo-inverse in real time; Output the corresponding joint angles and speeds to control the actuator to achieve precise motion.

[0044] S4 is specifically as follows: S41, Construction of the projection boundary of the support polygon: According to the position coordinates of the i-th wheel-foot at the current moment and the contact status flag , where 1 indicates contact and 0 indicates suspension, construct the set of effective support foot tips: ; Construct the boundary of the support polygon through the convex hull algorithm , and obtain the current center-of-gravity coordinates of the robot .

[0045] S42, Posture stability determination: If , it is determined that there is a risk of tipping over currently.

[0046] S43, If it is detected that the center of gravity is out of bounds, execute the buffer support correction: Insert a buffer support segment into the current action sequence, and control a pair of diagonal wheel-feet to touch the ground synchronously and apply enhanced pressure: , until it satisfies: , where is the buffer support strength coefficient, is the support force of the i-th wheel-foot, is the safety foot-tip support pressure threshold.

[0047] During the obstacle-crossing process of the robot, to evaluate its posture stability, it is necessary to determine the boundary of its support area based on the positions of the wheel-feet in contact with the ground currently. The specific method is as follows: Contact foot-tip screening: First, according to the status of the foot-tip contact sensor, screen out all the wheel-feet in contact with the ground currently, and extract the set of coordinate points on the ground projection plane (x-y plane); Convex hull boundary construction: Use the above-mentioned foot-tip coordinate points as the input, and perform boundary fitting using the two-dimensional convex hull algorithm. The convex hull algorithm finds a smallest closed polygon such that all contact points are located on the boundary of the polygon or inside it, and the boundary lines do not intersect themselves.

[0048] Boundary polygon formation: The obtained convex hull point set is sorted in clockwise or counterclockwise order to form the vertex sequence of the support polygon, thereby defining the geometric boundary of the wheel-foot support surface on the ground.

[0049] Centroid position judgment: By comparing with the projection position of the current centroid of the robot on the ground, it is judged whether the centroid falls within the range of the support polygon, which is used as the basis for judging the tipping risk.

[0050] The present invention covers any alternatives, modifications, equivalent methods and solutions made within the spirit and scope of the present invention. In order to enable the public to have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, and those skilled in the art can fully understand the present invention without these detailed descriptions. In addition, well-known methods, processes, procedures, components and circuits are not described in detail in order to avoid unnecessary confusion to the essence of the present invention.

[0051] The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. An adaptive obstacle-crossing control method for an electric drive columnar wheel-legged robot based on traveling path recognition, characterized in that, It includes the following steps: S1: Environmental perception and terrain map generation: Real-time collection of three-dimensional terrain feature parameters of the traveling path through a front-view lidar and foot-end pressure sensors, including obstacle height distribution, surface roughness, and slope changes, and combining visual recognition to perceive the height of the front step or the width of the groove, dynamically generating a terrain map for determining whether to enter the climbing platform, crossing the ditch, or slope control mode; S2: Mode switching and determination: According to the terrain map, combined with the preset wheel-foot mode switching threshold, determine the current obstacle type as a climbing platform, slope, or crossing the ditch; S3: Setting of each mode parameter: In the slope mode, calculate the dynamic friction coefficient between each foot-end and the slope ground based on the slope and surface roughness, and set the foot-end pressure safety threshold in combination with the target action mode: In the climbing platform mode, generate a wheel-foot lifting and propulsion command based on the terrain map, control the hydraulic cylinder, and preferentially lift the front-wheel support height to make the wheel-feet lift and extend in sequence for climbing the platform; In the crossing the ditch mode, generate a wheel-foot rotation and propulsion command based on the terrain map, expand the front-wheel angle, and the wheel-feet rotate and propel in sequence to cross the obstacle; S4: Dynamic stability check and correction preprocessing: Evaluate the obstacle-crossing stability according to the current center-of-gravity position of the robot, and automatically adjust the setting if there is a risk of tipping.

2. The adaptive obstacle-crossing control method for an electric drive columnar wheel-legged robot based on travel path recognition according to claim 1, wherein, The specific content of S1 includes: S11, Perform three-dimensional point cloud modeling on the traveling path through a front-view lidar, extract the obstacle height distribution and slope changes, and update the terrain contour in real time through the point cloud registration algorithm; S12: Collect the contact force signals between each wheel-foot and the ground through the foot-end pressure sensors, calculate the surface roughness based on the contact force variance, and determine that the current area is a high-roughness terrain when the contact force variance exceeds the preset roughness threshold; S13: Identify the geometric features of the front step or groove through a binocular vision sensor, perform sub-pixel-level measurement of the step height or groove width, and segment the step / groove boundary based on the edge detection algorithm.

3. The adaptive obstacle-crossing control method for an electric drive column-wheel-legged robot based on travel path recognition according to claim 2, wherein, S1 also includes multi-source data fusion of the obstacle height distribution, slope change parameters extracted by the lidar, the surface roughness calculated by the foot-end pressure sensors, and the step height or groove width measured by the vision sensors to generate a three-dimensional grid map including elevation, roughness, and obstacle type as the terrain map.

4. A method for adaptive obstacle crossing control of an electric drive columnar wheel-legged robot based on travel path recognition according to claim 1, characterized in that, In S2: According to the comparison results of the ratio of the obstacle height in the terrain map to the maximum lifting height of the robot's wheel-feet, the ratio of the groove width to the maximum span of the wheel-feet, and the slope to the wheel-foot tipping threshold, dynamically trigger the climbing platform, crossing the ditch, or slope mode.

5. The adaptive obstacle-crossing control method for an electric drive columnar wheel-legged robot based on traveling path recognition according to claim 4, wherein In S2: If it is identified as a "slope type" obstacle, start the normal passing process; If it is identified as a "step type" obstacle, start the climbing platform action process; If it is identified as a "groove type" obstacle, start the crossing the ditch action process.

6. The adaptive obstacle-crossing control method for an electric drive columnar wheel-legged robot based on traveling path recognition according to claim 1, characterized in that The parameter setting in the slope mode includes: According to the slope angle and surface roughness level in the topographic map, based on a preset friction coefficient mapping table, calculate the real-time friction coefficient between each foot end and the slope ground ; Based on the friction coefficient and the current load weight of the robot, calculate the minimum anti-slip pressure threshold at the end of each wheel foot through the static equilibrium equation , denote the minimum anti-slip pressure threshold at the end of the i-th wheel foot, and set the pressure safety threshold at the end of the wheel foot.

7. The adaptive obstacle-crossing control method for an electric drive columnar wheel-legged robot based on traveling path recognition according to claim 1, wherein The parameter setting in the climbing platform mode includes: Generate a wheel-foot lifting and propulsion command according to the step height H in the terrain map, and control the hydraulic cylinder to press , as the lifting margin, lift the front-wheel support height, and at the same time, based on the projection position of the robot's center of mass, plan the lifting timing of the rear wheel-foot so that the wheel-feet climb over the step in the action sequence of "front-wheel lifting → front-wheel extending → rear-wheel following" in turn.

8. The adaptive obstacle-crossing control method for an electric drive column-wheel-legged robot based on traveling path recognition according to claim 1, wherein The parameter setting in the crossing the ditch mode includes: According to the groove width W in the terrain map, generate a wheel-foot rotation and propulsion command, control the expansion of the front-wheel angle, and calculate the rotation angle and propulsion speed of each wheel-foot based on the inverse kinematics module to make the wheel-feet perform the crossing action of "front-wheel rotation into the groove edge → middle and rear wheels synchronously propel → front-wheel reset" in sequence.

9. The adaptive obstacle-crossing control method for an electric drive columnar wheel-legged robot based on travel path recognition according to claim 1, characterized in that, The above-mentioned S4 includes constructing a projection boundary of the support polygon based on the position coordinates and contact states of each round foot at present, and obtaining the position coordinates of the center of gravity in real time through the center-of-gravity sensor of the robot; if the position coordinates of the center of gravity exceed the projection boundary of the support polygon, it is determined that there is a risk of tipping over.

10. The adaptive obstacle-crossing control method for an electric drive columnar wheel-legged robot based on travel path recognition according to claim 9, characterized in that, In the above-mentioned S4, after it is determined that there is a risk of tipping over, a buffer support section is inserted into the preset action sequence, and at least a pair of diagonal round feet are controlled to touch the ground synchronously and the pressure at the foot end is increased until the center of gravity returns within the projection boundary of the support polygon.

Images