Pipeline robot composite walking self-adaptive control method and system

Abstract

The present application relates to the technical field of pipeline robot motion control, and particularly relates to a pipeline robot composite walking self-adaptive control method and system. The method comprises the following steps: receiving attitude parameters, angular velocity data, foot end contact pressure, motion commands and telemetry time stamps, calculating data age, generating a candidate running state, forming a stable running state through a first confirmation time or a second confirmation time; calling mode parameter sets according to the stable running state and the motion commands to generate a basic foot end trajectory; generating attitude compensation and contact compensation according to attitude deviation and pressure error, converting the two types of compensation into foot end reference position correction and superimposing the foot end reference position correction on the basic foot end trajectory, generating joint logic angles through foot end reachable workspace correction and inverse kinematics solution; mapping and limiting to generate rudder control instructions, and generating a safety attitude instruction in an abnormal state. The present application continuously connects state determination, foot end compensation and joint execution, and improves the stability of composite motion control.

Description

Technical Field

[0001] This invention relates to the field of motion control technology for pipeline robots, and in particular to an adaptive control method and system for composite walking of pipeline robots. Background Technology

[0002] In the field of pipeline robot motion control technology, existing pipeline robots typically employ wheeled, tracked, legged, or hybrid drive mechanisms, and are equipped with inertial measurement units, pressure sensors, and joint actuators. The controller generates foot trajectories or joint control quantities based on posture parameters, foot contact pressure, motion commands, and preset motion patterns, driving the robot to walk, roll, or overcome obstacles in straight pipe sections, curved pipe sections, sloping pipe sections, and local obstacle locations. Existing quadruped robot control schemes also frequently employ periodic gait, posture feedback, foot contact detection, and inverse kinematics solutions to generate joint angles.

[0003] Existing solutions often switch motion modes directly based on attitude parameters, pressure thresholds, or motion commands within a single control cycle, with little uniform constraint on the sampling time and transmission delay of different telemetry data. When the update times of attitude parameters, angular velocity data, and contact pressure at each leg end are inconsistent, or when telemetry data is delayed or interrupted, the controller may continue to use the data from the previous moment to determine the operating status. When the measured values ​​fluctuate around the threshold, the motion mode and its corresponding gait parameters are also prone to frequent switching, resulting in discontinuities in the connection between operating status determination, mode parameter calling, and gait cycle progression.

[0004] Furthermore, existing posture control and foot contact control typically generate joint correction quantities or actuator control quantities separately, and the correction results corresponding to posture deviations and leg contact pressure deviations lack a unified expression for foot position. When the two types of correction quantities are applied to joint commands separately, subsequent corrections may change the results of previous corrections, and the combined foot target is difficult to simultaneously accept the constraints of the foot-reachable workspace and inverse kinematics. Therefore, it is necessary to address the problem of discontinuous connection between operating state, foot target correction, and joint logic angle generation when telemetry data has timeliness differences and posture deviations and leg contact pressure deviations coexist, focusing on the composite motion control of pipeline robots. Summary of the Invention

[0005] To address the above problems, this invention provides a composite walking adaptive control method and system for pipeline robots. This method solves the problem of how to form joint logic angles constrained by the reachable workspace of the feet by synthesizing stable operating state gating and foot-end compensation before inverse kinematics when there are time differences in telemetry data of pipeline robots and posture deviations and contact pressure deviations of each leg.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] An adaptive control method for composite walking of a pipeline robot includes:

[0008] S100: Receives attitude parameters, angular velocity data, contact pressure at each leg and foot end, motion commands, and telemetry timestamps; calculates data age; generates candidate operating states based on attitude disturbance, angular velocity disturbance, pressure unevenness, and data age; determines stable operating state according to the first or second confirmation time of the state transition direction; updates stable operating state to safe state when data age exceeds data timeout threshold.

[0009] S200. Based on the stable operating state and the movement command, determine the target movement mode, call the mode parameter set including gait cycle, swing duty cycle, cycle offset of each leg and foot reference position offset of each leg, and generate the basic foot trajectory.

[0010] S300. Generate attitude compensation amount based on the deviation between attitude parameters and calibrated attitude and angular velocity data. Generate contact compensation amount for each leg based on the pressure error between the contact pressure at the foot end of each leg and the target pressure. Convert the two types of compensation amount into foot end reference position correction amount, superimpose it with the basic foot end trajectory, and generate joint logic angle through foot end reachable workspace correction and inverse kinematics solution.

[0011] S400: Map the joint logic angle to the hardware target angle and limit it according to the joint travel range to generate servo control commands; when the stable operating state is a safe state or the command transmission fails, stop the gait cycle propulsion and generate a safe attitude command.

[0012] Furthermore, the attitude parameters include pitch angle and roll angle;

[0013] The attitude disturbance is generated based on the deviation of the pitch angle and roll angle relative to the calibrated attitude;

[0014] The angular velocity disturbance is generated based on the X-axis angular velocity and the Y-axis angular velocity;

[0015] The pressure unevenness is generated based on the pressure error between the contact pressure of each leg and foot end and the corresponding target pressure.

[0016] The data age is the difference between the current time and the most recent valid telemetry timestamp.

[0017] Furthermore, in addition to the safety status, when the level of the candidate operating status is higher than the current stable operating status, the first confirmation time is used, and the stable operating status is updated after the first confirmation time is continuously met when the entry threshold is reached.

[0018] When the level of the candidate operating state is lower than the current stable operating state, the second confirmation time is used, and the stable operating state is updated after the exit threshold continues to meet the second confirmation time.

[0019] The second confirmation time is longer than the first confirmation time, and the corresponding confirmation time is reset when the candidate running state changes.

[0020] Furthermore, the target movement mode is selected from the safe standing mode, the normal walking mode, the high-pass walking mode, the longitudinal rolling mode, and the lateral rolling mode;

[0021] The high-speed walking mode corresponds to the posture compensation being enabled, and the lateral rolling mode corresponds to the contact compensation being enabled.

[0022] If the current motion mode and the target motion mode do not both generate a mode switching state, an intermediate posture between the current motion mode and the target motion mode is invoked, and the current motion mode is updated after the intermediate posture is completed.

[0023] Furthermore, generating the basic foot trajectory includes:

[0024] Add the time within the gait cycle to the cycle offset of the corresponding leg and normalize it according to the gait cycle to generate the local cycle of each leg;

[0025] By comparing the local period of each leg with the swing duty cycle, it is determined whether the corresponding leg is in the swing phase or the support phase.

[0026] In the swing phase, a foot-lifting trajectory and a forward-moving trajectory are generated; in the support phase, a retreating trajectory is generated.

[0027] The trajectory is corrected based on the swing gain of each leg, and the offset of the reference position of each leg foot is superimposed.

[0028] Furthermore, generating the attitude compensation amount includes:

[0029] Calculate the attitude deviations of the pitch and roll angles relative to the calibrated attitude;

[0030] Based on the attitude deviation and angular velocity data, a proportional-differential calculation is performed and the amplitude is limited according to the maximum compensation amount;

[0031] The attitude compensation amount corresponding to the pitch angle is converted into the foot reference position correction amount for the front and rear legs, respectively, and the attitude compensation amount corresponding to the roll angle is converted into the foot reference position correction amount for the left and right legs, respectively.

[0032] Furthermore, the generation of contact compensation amounts for each leg includes:

[0033] Calculate the pressure error between the contact pressure at the foot of each leg and the corresponding target pressure;

[0034] When the pressure error is within the pressure dead zone, the corresponding contact compensation amount is maintained.

[0035] When the pressure error is outside the pressure dead zone, the corresponding contact compensation amount is discretely accumulated based on the pressure error and the sampling time interval.

[0036] The contact compensation amount is limited by the maximum compensation amount and converted into a correction amount for the foot end reference position along the normal direction of the pipe wall of the corresponding leg.

[0037] Furthermore, the reachable workspace at the foot end is determined based on the length of the leg link and the range of joint travel;

[0038] When the foot target, after the two types of compensation amounts are superimposed on the basic foot trajectory, exceeds the foot reachable workspace, the foot target is corrected to the boundary of the foot reachable workspace;

[0039] The inverse kinematics of the corrected foot target are solved, and the joint logic angles are converted into hardware target angles according to the actuator channel mapping relationship, installation direction, zero position and proportional relationship.

[0040] The target angle of the hardware is limited based on the joint travel range and the maximum angle change in a single control cycle.

[0041] Furthermore, in the safe state, the safe posture command is maintained, the confirmation time of the candidate operating state is reset, and the update of the posture compensation amount and the contact compensation amount of each leg is stopped;

[0042] If the age of the continuously received attitude parameters, angular velocity data, and foot contact pressure data does not exceed the data timeout threshold, the stable operating state is restored from the safe state according to the second confirmation time.

[0043] Furthermore, a composite walking adaptive control system for a pipeline robot includes: a running state determination module, a motion mode and foot trajectory generation module, a foot compensation and inverse kinematics solution module, and an actuator mapping and safety control module; the system is used to implement the method described in any of the above embodiments.

[0044] The key innovations of this invention include:

[0045] (1) The attitude disturbance, angular velocity disturbance, pressure unevenness and data age are used together to generate candidate operating states, and the first confirmation time or the second confirmation time is called according to the state transition direction to form a stable operating state; when the data age exceeds the data timeout threshold, the stable operating state is updated to a safe state.

[0046] (2) Using stable operating state and motion command as common matching conditions for target motion mode, call the mode parameter set including gait cycle, swing duty cycle, cycle offset of each leg and reference position offset of each leg foot; divide the swing phase and support phase according to the local cycle of each leg, and generate the basic foot trajectory corresponding to each leg parameter.

[0047] (3) Generate attitude compensation amount based on the deviation between attitude parameters and calibrated attitude and angular velocity data, and generate contact compensation amount for each leg based on the pressure error between the contact pressure of each foot end and the target pressure; convert the two types of compensation amount into foot end reference position correction amount and superimpose it with the basic foot end trajectory, and perform foot end reachable workspace correction, inverse kinematics solution, hardware target angle mapping and joint stroke limit on the superimposed foot end target in sequence.

[0048] The following are its main beneficial effects:

[0049] (1) To address the issues of inconsistent update times for different telemetry data and discontinuous state transitions when the operating characteristics fluctuate around the threshold, data age verification, continuous confirmation of candidate operating states, and asymmetric confirmation time are used to ensure that telemetry data exceeding the data timeout threshold no longer maintains the normal operating state, and the target motion mode uses the continuously confirmed stable operating state as the basis for invocation.

[0050] (2) To address the issue of asynchronous updates between motion mode switching and gait parameters and foot reference position parameters of each leg, the target motion mode, local cycle of each leg, and foot reference position offset of each leg are updated in the same control process by the association call of stable running state and mode parameter set, forming a basic foot trajectory corresponding to the current running state and motion command.

[0051] (3) To address the problem that the correction results of posture compensation and contact compensation are difficult to unify when they are applied to joint control quantities, the two types of compensation quantities are converted to the foot reference position and superimposed before inverse kinematics, so that the posture deviation and the contact pressure deviation of each leg act together on the same foot target. The superimposed foot target is uniformly constrained by the foot reachable workspace, inverse kinematic relationship and joint stroke range, forming the joint logic angle and hardware target angle corresponding to the corrected foot target. Attached Figure Description

[0052] Figure 1 A flowchart illustrating a composite walking adaptive control method for a pipeline robot provided in an embodiment of this application;

[0053] Figure 2 This is a structural block diagram of a composite walking adaptive control system for a pipeline robot provided in an embodiment of this application. Detailed Implementation

[0054] To enable those skilled in the art to better understand the technical solution, the present invention will be described in detail below with reference to embodiments. The description in this part is only exemplary and explanatory, and should not be used to limit the scope of protection of the present invention in any way.

[0055] Example 1: Refer to Figure 1 This is a flowchart illustrating a composite walking adaptive control method for a pipeline robot provided in an embodiment of the present invention. The flowchart may include at least steps S100-S400:

[0056] S100: Receives attitude parameters, angular velocity data, contact pressure at each leg and foot end, motion commands, and telemetry timestamps; calculates data age; generates candidate operating states based on attitude disturbance, angular velocity disturbance, pressure unevenness, and data age; determines stable operating state according to the first or second confirmation time of the state transition direction; updates stable operating state to safe state when data age exceeds data timeout threshold.

[0057] S200. Based on the stable operating state and the movement command, determine the target movement mode, call the mode parameter set including gait cycle, swing duty cycle, cycle offset of each leg and foot reference position offset of each leg, and generate the basic foot trajectory.

[0058] S300. Generate attitude compensation amount based on the deviation between attitude parameters and calibrated attitude and angular velocity data. Generate contact compensation amount for each leg based on the pressure error between the contact pressure at the foot end of each leg and the target pressure. Convert the two types of compensation amount into foot end reference position correction amount, superimpose it with the basic foot end trajectory, and generate joint logic angle through foot end reachable workspace correction and inverse kinematics solution.

[0059] S400: Map the joint logic angle to the hardware target angle and limit it according to the joint travel range to generate servo control commands; when the stable operating state is a safe state or the command transmission fails, stop the gait cycle propulsion and generate a safe attitude command.

[0060] S100 receives attitude parameters, angular velocity data, contact pressure at each leg foot end, motion commands, and telemetry timestamps, and calculates the data age; generates candidate operating states based on attitude disturbance, angular velocity disturbance, pressure unevenness, and data age; determines stable operating states according to the first or second confirmation time of the state transition direction; and updates stable operating states to safe states when the data age exceeds the data timeout threshold.

[0061] This step is executed by the inertial measurement unit (IMU), pressure sensors located at the ends of the four legs, a telemetry gateway, and the state reception and validity determination module and the operating state and mode decision module in the data processor. The IMU is mounted on the robot body and outputs attitude parameters and angular velocity data. The four pressure sensors correspond to the four legs and output the contact pressure at each leg end. The telemetry gateway receives data frames from the IMU and pressure sensors and forwards them to the data processor. Motion commands are sent by the robot's controller or an external control terminal. These commands include forward, backward, stop, turn, longitudinal roll, and lateral roll commands. The telemetry timestamp is the sampling time of the corresponding data or the reception time of the telemetry gateway.

[0062] The attitude parameters include pitch angle and roll angle, representing the attitude changes of the robot body relative to the calibration attitude in the pitch and roll directions, respectively. The angular velocity data includes X-axis angular velocity and Y-axis angular velocity, where the X-axis is arranged longitudinally along the robot body and the Y-axis is arranged laterally along the robot body. The foot contact pressure represents the contact pressure between the foot of the corresponding leg and the pipe wall or support surface. The calibration attitude is the attitude parameters recorded when the pipeline robot is installed and in a preset support position, and the calibration attitude includes calibration pitch angle and calibration roll angle.

[0063] The data processor reads attitude parameters, angular velocity data, contact pressure at all four feet, motion commands, and telemetry timestamps according to the control cycle. The control cycle is triggered by a timing signal from the data processor; in one embodiment, the control cycle is 10ms to 50ms. When the telemetry gateway receives a new data frame, it writes the data type, leg identifier, measurement value, and telemetry timestamp from the data frame into the corresponding data record; at the beginning of the control cycle, the data processor reads the most recent valid value from each data record and associates the most recent valid value with the corresponding telemetry timestamp.

[0064] The data processor calculates the data age based on the difference between the current time and the most recent valid telemetry timestamp. For attitude parameters, angular velocity data, and contact pressure at each of the four feet, the attitude data age, angular velocity data age, and leg pressure data age can be calculated separately. In this embodiment, the maximum value among the data ages is used as the data age for the current control cycle. The data timeout threshold is stored in the data memory; when a data record does not contain a measurement value, the measurement value exceeds the sensor's range, the telemetry timestamp is earlier than the previous valid timestamp, or the data age exceeds the data timeout threshold, the corresponding data record is marked as invalid.

[0065] In one implementation, the data processor generates attitude disturbances based on the deviations of the pitch and roll angles from the calibrated attitude. Let the current pitch angle be θ, the calibrated pitch angle be θ0, the current roll angle be φ, and the calibrated roll angle be φ0; then the pitch deviation is θ - θ0, and the roll deviation is φ - φ0. The attitude disturbance is the larger of the absolute values ​​of the pitch and roll deviations. In another implementation, the attitude disturbance is a weighted sum of the absolute values ​​of the pitch and roll deviations, with the weights for the pitch and roll directions recorded in the data memory.

[0066] The angular velocity disturbance is generated based on the X-axis angular velocity and the Y-axis angular velocity. In one embodiment, the angular velocity disturbance is the sum of the absolute values ​​of the X-axis angular velocity and the Y-axis angular velocity; in another embodiment, the angular velocity disturbance is the larger of the absolute values ​​of the X-axis angular velocity and the Y-axis angular velocity. The angular velocity disturbance reflects the rotational change of the robot body within the current control cycle. The angular velocity disturbance and the attitude disturbance are stored separately to avoid merging attitude deviations and instantaneous rotations into indistinguishable single data.

[0067] The pressure unevenness is generated based on the pressure error between the foot contact pressure of the four legs and the corresponding target pressure. The target pressure of each leg is associated with and stored in relation to the target motion mode; during control cycles where the target motion mode selection has not yet been completed, the target pressure corresponding to the current motion mode is read. Let the foot contact pressure of the i-th leg be Fi, and the target pressure be Fi0, then the corresponding pressure error is ei = Fi - Fi0. In one embodiment, the pressure unevenness is the maximum value among the four absolute pressure errors; in another embodiment, the pressure unevenness is the difference between the maximum and minimum values ​​among the four foot contact pressures. If the data record of one pressure sensor is invalid, the foot contact state of the corresponding leg is marked as a low-confidence state, and the valid data state is marked as invalid; in another embodiment, when the other three pressure sensors are all valid and contact compensation is not enabled in the current motion mode, the current stable operating state is retained, and the update of the contact compensation amount for the leg corresponding to the invalid pressure sensor is stopped.

[0068] The operation status and mode decision module assembles an operation feature vector from attitude disturbance, angular velocity disturbance, pressure unevenness, data age, foot contact state, movement command, and current movement mode. Each data point in the operation feature vector retains its corresponding physical meaning, and cross-type numerical addition is not performed. The operation status and mode decision module compares the attitude disturbance with an attitude threshold, the angular velocity disturbance with an angular velocity threshold, the pressure unevenness with a pressure threshold, and the data age with a data timeout threshold. Based on the comparison results, it matches the pre-stored operation status rules to generate candidate operation states.

[0069] In one implementation, the operating status levels include a safe state, a normal operating state, a high-disturbance operating state, a longitudinal rolling operating state, and a lateral rolling operating state. A safe state corresponds to invalid data, data timeout, or an emergency stop signal; a normal operating state corresponds to attitude disturbance, angular velocity disturbance, and pressure unevenness all being within their respective entry thresholds; a high-disturbance operating state corresponds to attitude disturbance or angular velocity disturbance reaching its corresponding entry threshold; a longitudinal rolling operating state corresponds to a longitudinal rolling command and the foot contact state matching that command; and a lateral rolling operating state corresponds to a lateral rolling command and the pressure unevenness and foot contact state matching that command. The operating status rules do not directly replace motion commands with a single measurement value; instead, motion commands and measurement results jointly participate in the matching of candidate operating states.

[0070] The operation status and mode decision module stores the current stable operation status, candidate operation statuses, and the confirmation time of each candidate operation status. When a newly generated candidate operation status differs from the candidate operation status of the previous control cycle, the operation status and mode decision module resets the corresponding confirmation time. When a candidate operation status remains unchanged, the operation status and mode decision module accumulates the confirmation time according to the control cycle. When the level of a candidate operation status is higher than the current stable operation status, the first confirmation time is invoked, and after the first confirmation time is continuously met at the corresponding entry threshold, the candidate operation status is written into the stable operation status. When the level of a candidate operation status is lower than the current stable operation status, the second confirmation time is invoked, and after the second confirmation time is continuously met at the corresponding exit threshold, the candidate operation status is written into the stable operation status. The second confirmation time is longer than the first confirmation time.

[0071] Entry and exit thresholds are stored separately, and the entry and exit thresholds for the same operating feature are different. For example, when the attitude disturbance reaches the attitude entry threshold, the candidate operating state can be updated to the high-disturbance operating state; when the stable operating state is already in the high-disturbance operating state, the stable operating state is only updated to the normal operating state after the attitude disturbance falls back to below the attitude exit threshold and continues to meet the second confirmation time. Angular velocity disturbance and pressure unevenness adopt the same confirmation method. The first or second confirmation time is not executed in the safe state; when the data age exceeds the data timeout threshold, the data validity status is invalid, or an emergency stop signal is received, the operating state and mode decision module directly updates the stable operating state to the safe state.

[0072] In an embodiment operating within a pipeline, the robot body is in a straight pipe section. The pitch and roll angles output by the inertial measurement unit are close to the calibrated attitude, the contact pressure at the four feet is near the corresponding target pressure, and the data age is less than the data timeout threshold. The operating state and mode decision module generates a normal operating state as a candidate operating state. After the corresponding confirmation time expires, the stable operating state is updated to the normal operating state. After the robot enters a sloping pipe section, the pitch deviation and X-axis angular velocity continuously increase. The attitude disturbance and angular velocity disturbance reach the entry threshold and continue to meet the first confirmation time, and the stable operating state is updated to the high-disturbance operating state. After the robot leaves the sloping pipe section, the attitude disturbance falls back to within the exit threshold and continues to meet the second confirmation time, and the stable operating state returns to the normal operating state.

[0073] The data generated in this step includes data age, attitude disturbance, angular velocity disturbance, pressure unevenness, candidate operating states, confirmation time of candidate operating states, and stable operating states. The stable operating states and motion commands serve as inputs to step S200; the attitude parameters, angular velocity data, foot contact pressure, and their valid data status are retained, and attitude compensation and contact compensation for each leg are generated in step S300. If a data timeout occurs, the stable operating state is updated to a safe state, and safe attitude processing is performed in step S400.

[0074] S200. Based on the stable operating state and movement command, determine the target movement mode, call the mode parameter set including gait cycle, swing duty cycle, cycle offset of each leg and reference position offset of each leg foot, and generate the basic foot trajectory.

[0075] This step is executed by the running state and mode decision module, the foot trajectory generation unit, and the data storage unit in the data processor. The running state and mode decision module receives the stable running state formed in step S100 and reads the current motion mode and motion command; the data storage unit stores the composite motion mode and its corresponding mode parameter set; the foot trajectory generation unit generates the basic foot trajectory of the four legs according to the target motion mode and the mode parameter set.

[0076] The composite motion modes include a safe standing mode, a normal walking mode, a high-through-traffic walking mode, a longitudinal rolling mode, and a lateral rolling mode. The safe standing mode corresponds to a safe state; the normal walking mode corresponds to a normal operating state and a forward or backward command; the high-through-traffic walking mode corresponds to a high-disturbance operating state; the longitudinal rolling mode corresponds to a longitudinal rolling operating state and a longitudinal rolling command; and the lateral rolling mode corresponds to a lateral rolling operating state and a lateral rolling command. When the motion command is a stop command, the target motion mode is updated to the safe standing mode or the stopping posture corresponding to the current motion mode, the specific selection being recorded by the stop rules in the mode parameter set.

[0077] The mode parameter set is stored in the data memory according to the target motion mode. The mode parameter set includes gait cycle, swing duty cycle, cycle offset for each of the four legs, and foot reference position offset for each of the four legs. Gait cycle represents the time corresponding to the completion of one cycle change of the foot movement; swing duty cycle represents the proportion of time of the swing phase in the gait cycle; cycle offset represents the time offset of the local cycle of the corresponding leg relative to the starting point of the common cycle; foot reference position offset represents the coordinate correction of the foot reference position of the corresponding leg relative to the calibrated foot reference position.

[0078] In one implementation, the mode parameter set also records the swing gain of each leg, the target pressure of each leg, the attitude compensation enabled status, the contact compensation enabled status, the intermediate attitude, and the maximum angle change per control cycle. The normal walking mode can record the attitude compensation disabled or enabled status; the high-pass walking mode records the attitude compensation enabled status; and the lateral rolling mode records the contact compensation enabled status. These enabled statuses control whether the attitude compensation and contact compensation amounts in step S300 are updated, without changing the processing order in which the two types of compensation amounts act on the foot reference position in step S300.

[0079] The operation status and mode decision module matches the stable operation status and movement commands with the mode rules of the composite movement mode to generate a target movement mode. When the current movement mode is the same as the target movement mode, the current movement mode is maintained, and the foot trajectory generation unit continues to advance the current gait cycle; when the current movement mode is different from the target movement mode, a mode switching state is generated, and the intermediate posture between the current movement mode and the target movement mode is read from the data storage. The intermediate posture includes the foot reference positions or corresponding joint logical angles of the four legs, and records the movement sequence of the feet of each leg during the mode switching process.

[0080] After the mode switching state is established, the foot trajectory generation unit first generates a transitional foot trajectory based on the intermediate posture; step S300 compensates for and solves the inverse kinematics of the transitional foot trajectory, and step S400 executes hardware target angle limiting based on the maximum angle change in a single control cycle. When the difference between the current foot reference position of each of the four legs and the corresponding foot reference position in the intermediate posture is within a preset position difference range, or the difference between the current joint logic angle and the corresponding joint logic angle in the intermediate posture is within a preset angle difference range, the running state and mode decision module updates the mode switching state to the completed state and updates the current motion mode to the target motion mode. Before the mode switching is completed, the foot trajectory generation unit does not directly call the complete cycle trajectory of the target motion mode.

[0081] When the target movement mode is a regular walking mode or a high-through-traffic walking mode, the foot trajectory generation unit reads the time within the current gait cycle. Let the gait cycle be T, the current cycle time be t, and the cycle offset of the i-th leg be δi. Then, t and δi are added together, and the remainder is taken with respect to the gait cycle T to obtain the cycle time of the corresponding leg. The cycle time is then divided by T to generate local cycles for each leg, with values ​​ranging from 0 to 1. When the current cycle time exceeds the gait cycle, the foot trajectory generation unit subtracts one gait cycle from the current cycle time while keeping the cycle offset of each leg unchanged.

[0082] The foot trajectory generation unit compares the local period of each leg with the swing duty cycle. When the local period of each leg is less than the swing duty cycle, the corresponding leg is in the swing phase; when the local period of each leg is not less than the swing duty cycle, the corresponding leg is in the support phase. A lifting trajectory and a forward trajectory are generated within the swing phase, and a retreating trajectory is generated within the support phase. The lifting trajectory indicates that the foot moves away from the support surface, the forward trajectory indicates that the foot moves in the direction corresponding to the motion command, and the retreating trajectory indicates that the foot moves in the opposite direction relative to the robot body while maintaining contact.

[0083] In one embodiment, the normalized time of the swing phase is the ratio of the local period of each leg to the swing duty cycle; the foot lift is generated according to the sine curve corresponding to the normalized time, and the foot forward movement is generated according to the continuous curve corresponding to the normalized time. The normalized time of the support phase is the difference between the local period of each leg minus the swing duty cycle and 1, and the swing duty cycle; the foot retraction is continuously generated according to the normalized time of the support phase. The continuous curve maintains positional continuity at the start and end points of the swing phase, the start and end points of the support phase.

[0084] The foot trajectory generation unit corrects the lifting, forward movement, and retraction of each leg based on the swing amplitude gain of each leg. The swing amplitude gain of each leg is used to compensate for the differences in link length, joint zero position, and assembly position of the four legs. After completing the swing amplitude correction, the foot end reference position offset of each leg is superimposed onto the corresponding leg's foot end position. The foot end reference position offset includes X-axis offset, Y-axis offset, and Z-axis offset. The X-axis offset corresponds to the longitudinal position correction of the robot body, the Y-axis offset corresponds to the lateral position correction of the robot body, and the Z-axis offset corresponds to the distance correction of the foot end relative to the support surface.

[0085] When the target motion mode is longitudinal rolling mode, the mode parameter set records the longitudinal rolling support positions of the four legs; the foot trajectory generation unit generates a basic foot trajectory based on the current foot reference position and the longitudinal rolling support position, and maintains the corresponding foot reference position after the four legs reach the longitudinal rolling support position. The driving direction of the longitudinal rolling mechanism is given by the motion command. When the target motion mode is lateral rolling mode, the mode parameter set records the lateral wall-hugging support positions of the four legs and the target pressure; the foot trajectory generation unit generates a basic foot trajectory from the current foot reference position to the lateral wall-hugging support position, and passes the contact compensation enabled state to step S300.

[0086] In an embodiment of pipe joint operation, the stable operating state output in step S100 is updated from the normal operating state to the high-disturbance operating state; the operating state and mode decision module updates the target motion mode from the normal walking mode to the high-through walking mode and generates a mode switching state. The foot trajectory generation unit first calls the intermediate posture between the normal walking mode and the high-through walking mode; after the intermediate posture is completed, it reads the gait cycle, swing duty cycle and period offset of each leg corresponding to the high-through walking mode. The foot lift trajectory height of the high-through walking mode is greater than that of the normal walking mode. The four legs generate the swing phase trajectory and support phase trajectory in sequence according to their respective local cycles, and superimpose the foot reference position offset of each leg.

[0087] This step outputs the target motion mode, mode switching status, mode parameter set, local cycles of each leg, and basic foot trajectory. The basic foot trajectory includes the foot reference positions corresponding to each of the four legs; the posture compensation enabled status, contact compensation enabled status, target pressure of each leg, and maximum compensation amount in the mode parameter set are transmitted to step S300; the target motion mode and mode switching status are transmitted to step S400 and participate in the distribution of hardware target angles and the update of the current motion mode.

[0088] S300. Generate attitude compensation amount based on the deviation between attitude parameters and calibrated attitude and angular velocity data. Generate contact compensation amount for each leg based on the pressure error between the contact pressure at the foot end of each leg and the target pressure. Convert the two types of compensation amount into foot end reference position correction amount, superimpose it with the basic foot end trajectory, and generate joint logic angle through foot end reachable workspace correction and inverse kinematics solution.

[0089] This step is executed by the foot trajectory and compensation module and the inverse kinematics and actuator mapping module in the data processor. The foot trajectory and compensation module receives the basic foot trajectory and mode parameter set formed in step S200, and reads the effective attitude parameters, angular velocity data and foot contact pressure of each leg retained in step S100; the inverse kinematics and actuator mapping module reads the leg link length, joint travel range and installation position of each joint, and performs inverse kinematics solution on the corrected foot target.

[0090] In this embodiment, the X-axis of the robot's body coordinate system is set along the longitudinal direction of the robot body, the Y-axis is set along the transverse direction of the robot body, and the Z-axis points from the center of the robot body to the normal direction of the current main support surface. A separate leg coordinate system is established for each leg, and the reference position of each leg's foot is initially represented in the robot body coordinate system. Before solving the inverse kinematics, the foot reference position in the robot body coordinate system is transformed to the corresponding leg coordinate system based on the hip joint mounting position of each leg. The normal direction of the pipe wall in the lateral rolling mode is recorded by the corresponding mode parameter set and transformed to the robot body coordinate system.

[0091] The attitude compensation amount is generated based on the attitude deviations of the pitch and roll angles relative to the calibrated attitude, as well as angular velocity data. Let the pitch deviation be eθ, the roll deviation be eφ, the X-axis angular velocity be ωx, and the Y-axis angular velocity be ωy. In one embodiment, the pitch compensation amount is calculated using proportional-differential calculations based on eθ and the corresponding angular velocity, and the roll compensation amount is calculated using proportional-differential calculations based on eφ and the corresponding angular velocity. The proportional and differential coefficients are stored according to the target motion mode; the calculated pitch and roll compensation amounts are each limited to the maximum compensation amount.

[0092] During attitude compensation, the foot trajectory and compensation module converts the pitch compensation amount into foot reference position correction amounts for the front and rear legs, respectively. When the pitch angle increases towards the front, the front and rear legs use foot reference position correction amounts in opposite directions along the Z-axis; when the pitch angle increases towards the rear, the correction directions are opposite. The roll compensation amount is converted into foot reference position correction amounts for the left and right legs, respectively; when the robot body tilts to the left, the left and right legs use correction amounts in opposite directions along the Z-axis; when the robot body tilts to the right, the correction directions are opposite.

[0093] In one implementation, the attitude compensation amount includes Y-axis correction in addition to Z-axis correction. In lateral roll mode, the roll compensation amount is converted into the reference position correction amount of each leg end according to the tube wall normal direction recorded in the mode parameter set; when the tube wall normal does not coincide with the Z-axis of the robot body coordinate system, the foot trajectory and compensation module allocates the roll compensation amount according to the components of the tube wall normal on the Y-axis and Z-axis. When the attitude compensation enabled state is disabled, the attitude compensation amount for this control cycle is set to zero; when the attitude data is invalid, the attitude compensation amount is stopped from being updated, and the zero compensation amount or the effective compensation amount after limiting from the previous control cycle is used, the specific method of which is given by the configuration record of the safety control module.

[0094] The contact compensation amount for each leg is generated based on the pressure error between the foot contact pressure and the target pressure of the corresponding leg. The target pressure for each leg is read from the mode parameter set called in step S200. The foot trajectory and compensation module calculates the pressure error in each control cycle; when the pressure error is within the pressure dead zone, the contact compensation amount of the corresponding leg in the previous control cycle is maintained; when the pressure error is outside the pressure dead zone, the contact compensation amount of the corresponding leg is discretely accumulated based on the pressure error and the sampling time interval.

[0095] In one implementation, the contact compensation amount of the i-th leg in the k-th control cycle is calculated based on the contact compensation amount of the previous control cycle, the pressure error, the contact compensation coefficient, and the sampling time interval. If the calculated result is greater than the maximum compensation amount, the maximum compensation amount is used; if the calculated result is less than the maximum negative compensation amount, the maximum negative compensation amount is used. The pressure error indicates that when the actual foot contact pressure is less than the target pressure, the contact compensation amount changes along the direction closer to the pipe wall; the pressure error indicates that when the actual foot contact pressure is greater than the target pressure, the contact compensation amount changes along the direction farther from the pipe wall.

[0096] The contact compensation amount is converted into a foot reference position correction amount for the corresponding leg along the pipe wall normal. In the normal walking mode and high-pass walking mode, the pipe wall normal can be taken as the normal direction of the current support surface; in the lateral rolling mode, the pipe wall normal is determined according to the lateral wall-attaching direction recorded in the mode parameter set. When the contact compensation enable state is disabled, the contact compensation amount of each leg in this control cycle is set to zero; when the data from a pressure sensor is invalid, the update of the contact compensation amount of the corresponding leg stops, and the foot contact state of the corresponding leg is transmitted to the safety control module.

[0097] After calculating the attitude compensation and contact compensation for each leg, the foot trajectory and compensation module converts these two types of compensation into foot reference position corrections in the robot's body coordinate system. For the i-th leg, let the foot reference position in the basic foot trajectory be Pi0, the foot reference position correction corresponding to attitude compensation be ΔPia, and the foot reference position correction corresponding to contact compensation be ΔPic. Then, the superimposed foot target is the vector sum of Pi0, ΔPia, and ΔPic. The two types of compensation are superimposed at the foot reference position layer, and then proceed to the foot reachable workspace correction and inverse kinematics solution. No compensation angle is added after the hardware target angle is generated.

[0098] The reachable workspace at the foot end is determined based on the leg link length and joint range of motion. Each leg of the quadrupedal locomotion mechanism includes a hip roll joint, a hip pitch joint, and a knee joint; the data storage records the rotation direction, joint range of motion, hip joint mounting position, thigh link length, and lower leg link length for each joint. The inverse kinematics and actuator mapping module establishes the reachable workspace at the foot end for each leg based on the above parameters and compares the superimposed foot end target with the reachable workspace at the foot end.

[0099] When the superimposed foot target is within the reachable workspace of the foot, it is used as the corrected foot target; when the superimposed foot target exceeds the reachable workspace, the foot target is corrected to the boundary of the reachable workspace along the direction from the calibrated foot reference position to the foot target. In another embodiment, the radial distance, lateral position, and longitudinal position of the foot target in the leg coordinate system are respectively limited; coordinate components exceeding the corresponding range are corrected to the corresponding boundary values. The foot target before correction, the foot target after correction, and the leg identifier that triggered the correction are recorded in the joint limiting record.

[0100] The inverse kinematics and actuator mapping module corrects the transformation of the foot target from the robot's body coordinate system to the leg coordinate system of each leg. The hip roll joint logic angle is solved based on the positional relationship of the foot target in the lateral and normal directions of the leg. After completing the hip roll joint rotation conversion, the foot target is projected onto the plane containing the hip pitch and knee joints. Based on the thigh link length, the lower leg link length, and the distance between the projected foot target and the hip pitch joint, the knee joint logic angle is solved. Then, the hip pitch joint logic angle is solved based on the foot target direction and the link angle.

[0101] When calculating the knee joint logical angle, the inverse kinematics and actuator mapping module calculates the cosine term; the cosine term is set to 1 if it is greater than 1, and to -1 if it is less than -1. If the corrected foot target still cannot obtain a valid joint logical angle after workspace correction, the valid joint logical angle of the corresponding leg in the previous control cycle is retained, and an inverse kinematics abnormal state is written to the safety control module; when the same leg continuously exhibits an inverse kinematics abnormal state, the safety control module updates the stable operating state to a safe state in subsequent steps.

[0102] In an embodiment of lateral wall-hugging operation, step S200 outputs the basic foot trajectory corresponding to the lateral rolling mode and enables contact compensation; the actual foot contact pressure of the two left legs is less than the target pressure, while the actual foot contact pressure of the two right legs is greater than the target pressure. The foot trajectory and compensation module discretely accumulate the pressure errors of the four legs respectively. The contact compensation amount of the left leg changes along the direction closer to the pipe wall, and the contact compensation amount of the right leg changes along the direction away from the pipe wall; at the same time, attitude compensation amount is generated based on the roll deviation. After the two types of compensation are converted into foot reference position correction amounts, they are superimposed with the basic foot trajectory, corrected by the foot reachable workspace, and then the logical angles of the hip roll joint, hip pitch joint, and knee joint of the four legs are solved respectively.

[0103] This step outputs corrected foot targets and joint logic angles. The corrected foot targets include the reference positions of the four legs in the robot's body coordinate system, and record the attitude compensation components, contact compensation components, and workspace correction components. The joint logic angles are arranged according to leg and joint identifiers and serve as inputs for actuator mapping and joint travel limiting in step S400. The attitude compensation and contact compensation values ​​for each leg are saved to the next control cycle. They are maintained when the pressure error is within the pressure dead zone, and updates cease in step S400 when the system enters a safe state.

[0104] S400: Map the joint logic angle to the hardware target angle and limit it according to the joint travel range to generate servo control commands; when the stable operating state is a safe state or the command transmission fails, stop the gait cycle propulsion and generate a safe attitude command.

[0105] This step is executed by the inverse kinematics and actuator mapping module, safety control module, servo actuator controller, and multiple joint actuators in the data processor. The inverse kinematics and actuator mapping module receives the joint logic angles generated in step S300; the safety control module reads the stable operation status, mode switching status, current motion mode, joint travel range, maximum angle change in a single control cycle, and command transmission status; the servo actuator controller receives servo control commands or safety attitude commands and drives multiple joint actuators of the quadrupedal walking mechanism.

[0106] The joint logic angles are arranged according to leg and joint identifiers. Each leg corresponds to the hip roll joint, hip pitch joint, and knee joint; the joint logic angles of the four legs form 12 joint angle records. The data storage device stores the actuator channel mapping relationship between each joint and the servo actuator controller channel; the actuator channel mapping relationship includes leg identifier, joint identifier, actuator channel, installation direction, zero position, proportional relationship, and joint travel range.

[0107] The inverse kinematics and actuator mapping module first arranges the joint logic angles to the corresponding actuator channels according to the actuator channel mapping relationship; then, it converts the joint logic angles into hardware target angles based on the installation direction, zero position, and scaling relationship. In one implementation, the hardware target angle of the j-th joint is obtained by multiplying the joint logic angle by the scaling factor, then by the installation direction factor, and adding it to the zero position; the installation direction factor is 1 or -1. Joints of the same name in different legs separately store the installation direction and zero position, without using a common zero position to replace the installation parameters of the four legs.

[0108] After generating the hardware target angles, the inverse kinematics and actuator mapping module compares each hardware target angle with the corresponding joint travel range. If the hardware target angle is greater than the upper limit of the joint travel range, the hardware target angle is corrected to the upper limit; if the hardware target angle is less than the lower limit of the joint travel range, the hardware target angle is corrected to the lower limit. When amplitude limiting occurs, the leg identifier, joint identifier, angle before amplitude limiting, angle after amplitude limiting, and the current target motion mode are recorded.

[0109] The safety control module also compares the hardware target angle of the current control cycle with the hardware target angle issued in the previous control cycle. If the absolute value of the difference is greater than the maximum angle change in a single control cycle, the hardware target angle of the current control cycle is corrected according to the sum or difference between the hardware target angle issued in the previous control cycle and the maximum angle change in a single control cycle. When the current motion mode is different from the target motion mode and the mode switching state is not completed, the hardware target angle of each joint gradually approaches the hardware target angle corresponding to the intermediate posture according to the aforementioned change limit.

[0110] The safety control module encapsulates the hardware target angles—which define the channel arrangement, joint travel limits, and angle variation restrictions within a single control cycle—into servo control commands. These commands include a control cycle identifier, the current motion mode, the actuator channel, and the corresponding hardware target angle for each actuator channel. In one implementation, the servo control commands also include a transmission timestamp and a verification field. The servo actuator controller parses the servo control commands and drives the corresponding joint actuators according to the actuator channels.

[0111] The command transmission status is generated by the communication status between the data processor and the servo actuator controller. In one embodiment, the servo actuator controller returns to the receiving state after receiving a servo control command; if the data processor does not receive a receiving status within a preset time, it marks the command transmission status of the current control cycle as failed. In another embodiment, the data processor generates the command transmission status based on the write status of the communication interface. When command transmission fails, the current gait cycle is not continued, and the stable operating state is updated to a safe state.

[0112] When the stable operating state is a safe state, the safety control module stops updating the current gait cycle; the current cycle time in step S200 remains at the value before entering the safe state, or is reset to the safe cycle start point recorded in the mode parameter set. The safety control module stops updating the posture compensation amount and the contact compensation amount of each leg, resets the confirmation time of the candidate operating state, and generates a safe posture command. The safe posture command is generated by the safe posture angle recorded in the data storage, and the safe posture angle corresponds to the hip roll joint, hip pitch joint, and knee joint of the four legs, respectively.

[0113] When transitioning from the current hardware target angle to the safe attitude angle, the amplitude is still limited according to the joint travel range and the maximum angle change in a single control cycle. If the command transmission fails and the servo actuator does not receive the normal servo control command, the data processor will prioritize sending the safe attitude command after communication is restored; the transmission status of the safe attitude command is recorded, and the safe attitude command is maintained after successful transmission until the safe state is released.

[0114] Before the safe state is lifted, the state receiving and validity determination module continues to receive attitude parameters, angular velocity data, and foot contact pressure of each leg, but does not directly use a single valid control cycle as a recovery condition. If the data age corresponding to the continuously received attitude parameters, angular velocity data, and foot contact pressure does not exceed the data timeout threshold, and the second confirmation time is continuously met, the operating state and mode decision module restores the stable operating state from the safe state to the operating state matching the current operating feature vector. If the candidate operating state changes during the recovery period, the second confirmation time is reset.

[0115] After restoring a stable operating state, the operating state and mode decision module does not directly restore the target motion mode before entering the safe state. Instead, it re-executes step S200 based on the restored stable operating state and the current motion command. When the target motion mode differs from the safe standing mode, step S200 generates a mode switching state and calls the corresponding intermediate posture; step S300 recalculates the posture compensation amount and the contact compensation amount of each leg; and step S400 transitions from the safe posture to the intermediate posture according to the maximum angle change in a single control cycle.

[0116] In one embodiment of telemetry interruption, the pipeline robot is in high-through-travel walking mode, and the telemetry gateway has not received data frames from the inertial measurement unit for several consecutive control cycles. The data age calculated in step S100 exceeds the data timeout threshold, and the stable operating state is updated to a safe state. The safety control module stops gait cycle progression, stops updating attitude compensation and leg contact compensation, and generates multiple consecutive safe attitude commands based on the current hardware target angle and safe attitude angle; the angle change within each control cycle does not exceed the maximum angle change in a single control cycle. After telemetry recovery, the attitude parameters, angular velocity data, and foot contact pressure remain valid and meet the second confirmation time, restoring the stable operating state; the operating state and mode decision module reselects the target motion mode and completes the mode transition according to the intermediate posture.

[0117] In another embodiment, the mode parameter set, actuator channel mapping relationship, joint travel range, and safe posture angle are stored in the data memory in association with the robot body identifier and parameter version. When the data processor starts, it reads the robot body identifier and parameter version and verifies whether the read parameters match the current robot body. If the verification result matches, the corresponding parameters are loaded; if the verification result does not match, pre-stored safety parameters are loaded, and the initial stable operating state is set to a safe state. After the parameters are loaded, the state receiving and validity determination module begins receiving telemetry data.

[0118] After the servo actuator drives multiple joint actuators, the robot's posture and the contact pressure at the ends of each leg change. In the next acquisition cycle, the inertial measurement unit and pressure sensors generate new posture parameters, angular velocity data, and foot contact pressure, which are then returned to step S100 via the telemetry gateway. The data processor writes the stable operating state, target motion mode, mode switching state, joint logic angle, hardware target angle, command transmission status, and anomaly flags of this control cycle into the operation record. The current motion mode and the hardware target angle from the previous control cycle in the operation record participate in the mode selection, angle change limitation, and anomaly handling of the next control cycle.

[0119] Example 2: Figure 2 A structural block diagram of a composite walking adaptive control system for a pipeline robot according to an embodiment of the present invention is shown. Figure 2 As shown, the structure may include:

[0120] The operating status determination module 01 receives attitude parameters, angular velocity data, contact pressure at each leg end, motion commands, and telemetry timestamps. It calculates the data age, generates candidate operating states based on attitude disturbance, angular velocity disturbance, pressure unevenness, and the data age, determines a stable operating state according to the first or second confirmation time corresponding to the state transition direction, and updates the stable operating state to a safe state when the data age exceeds a data timeout threshold. Specifically, the attitude parameters, angular velocity data, contact pressure at each leg end, and telemetry timestamps are input according to the same control cycle, and the motion commands are associated with the control cycle. The operating status determination module reads the current time and the most recent valid telemetry timestamp, uses the time difference between the two to form the data age, and performs validity checks on the attitude parameters, angular velocity data, and contact pressure at each leg end. The validity check includes checking whether the input object is missing, whether it exceeds the corresponding acquisition range, and whether the telemetry timestamp is earlier than the previous valid telemetry timestamp. If the check fails, the corresponding input object is recorded as invalid, and the stable operating state is updated to the safe state. The attitude disturbance is generated based on the deviation of the attitude parameters from the calibrated attitude; the angular velocity disturbance is generated based on the angular velocity data; and the pressure unevenness is generated based on the pressure error between the contact pressure at each leg and foot end and the corresponding target pressure. The operating state determination module compares the attitude disturbance, angular velocity disturbance, and pressure unevenness with the entry threshold or exit threshold, respectively, and combines the motion command and the data age to match the operating state, generating the candidate operating state. When the candidate operating state differs from the current stable operating state, the first confirmation time or the second confirmation time is called according to the state transition direction between the two; when the candidate operating state remains unchanged within the corresponding confirmation time, it is written into the stable operating state; when the candidate operating state changes before the confirmation time expires, the corresponding confirmation time is reset. When the data age exceeds the data timeout threshold, the accumulation of the first confirmation time or the second confirmation time is not performed, and the stable operating state is directly updated to the safe state. The running state determination module provides the stable running state and movement command to the motion mode and foot trajectory generation module, provides the foot compensation and inverse kinematics solution module with validated posture parameters, angular velocity data and contact pressure of each leg foot end, and provides the stable running state to the actuator mapping and safety control module.

[0121] The movement mode and foot trajectory generation module 02, connected to the running state determination module, is used to determine the target movement mode based on the stable running state and the movement command. It calls a set of mode parameters, including gait cycle, swing duty cycle, leg cycle offset, and leg foot reference position offset, to generate a basic foot trajectory. Specifically, the movement mode and foot trajectory generation module receives the stable running state and movement command output by the running state determination module, matches them with the correspondence of composite movement modes, and forms the target movement mode. The composite movement modes include a safe standing mode, a normal walking mode, a high-pass walking mode, a longitudinal rolling mode, and a lateral rolling mode. When the stable running state and movement command match a composite movement mode, the corresponding composite movement mode is recorded as the target movement mode. The movement mode and foot trajectory generation module calls the corresponding set of mode parameters according to the target movement mode and reads the gait cycle, swing duty cycle, leg cycle offset, and leg foot reference position offset. The current time within the gait cycle is added to the cycle offset of the corresponding leg, and then normalized according to the gait cycle to form a local cycle for each leg. The local cycle of each leg is compared with the swing duty cycle. When the local cycle of each leg is in the swing phase, a lifting trajectory and a forward trajectory for the corresponding leg are generated; when the local cycle of each leg is in the support phase, a retreat trajectory for the corresponding leg is generated. The lifting trajectory, forward trajectory, and retreat trajectory are superimposed with the foot reference position offset of the corresponding leg to form the basic foot trajectory for each of the four legs. When the current movement mode differs from the target movement mode, the movement mode and foot trajectory generation module calls an intermediate posture based on the current movement mode and the target movement mode, first generating a basic foot trajectory from the current foot reference position to the intermediate posture. After the intermediate posture is completed, the gait cycle is advanced according to the mode parameter set corresponding to the target movement mode. When the mode parameter set is missing or one of the cycle offsets is not associated with the corresponding leg, the mode parameter set corresponding to the current movement mode is maintained, the target movement mode is not updated, and the calling status of the mode parameter set is passed to the actuator mapping and safety control module. The motion pattern and foot trajectory generation module transmits the basic foot trajectory and the pattern parameter set to the foot compensation and inverse kinematics solving module; wherein, the basic foot trajectory is used as the superposition of posture compensation and contact compensation of each leg, and the target pressure in the pattern parameter set is used as the input for generating contact compensation of each leg.

[0122] The foot-end compensation and inverse kinematics solution module 03 connects the motion mode and foot trajectory generation module and the running state determination module. It generates a posture compensation amount based on the deviation between the posture parameters and the calibrated posture, and the angular velocity data. It also generates a contact compensation amount for each leg based on the pressure error between the foot-end contact pressure and the target pressure. The posture compensation amount and the contact compensation amount for each leg are converted into a foot-end reference position correction amount and superimposed on the basic foot-end trajectory. After correction of the reachable workspace and inverse kinematics solution, a joint logic angle is generated. Specifically, the foot-end compensation and inverse kinematics solution module receives the basic foot-end trajectory and mode parameter set output by the motion mode and foot trajectory generation module, and receives the posture parameters, angular velocity data, and foot-end contact pressure for each leg provided by the running state determination module. The foot-end compensation and inverse kinematics solution module calculates the posture deviation of the posture parameters relative to the calibrated posture, performs proportional-differential calculations on the posture deviation and the angular velocity data to generate the posture compensation amount, and limits the amplitude according to the maximum compensation amount. The posture compensation amount is converted according to the positional relationship between the front leg, rear leg, left leg, and right leg to form the foot reference position correction amount corresponding to each leg. The foot compensation and inverse kinematics solution module calculates the pressure error between the foot contact pressure and the corresponding target pressure of each leg. When the pressure error is within the pressure dead zone, the contact compensation amount of the corresponding leg in the previous control cycle is maintained. When the pressure error is outside the pressure dead zone, the contact compensation amount of the corresponding leg is discretely accumulated according to the pressure error and the sampling time interval, and the amplitude is limited according to the maximum compensation amount. The contact compensation amount of each leg is converted into the foot reference position correction amount according to the tube wall normal of the corresponding leg. The foot reference position correction amount corresponding to the posture compensation amount and the foot reference position correction amount corresponding to the contact compensation amount of each leg are superimposed with the basic foot trajectory to form the foot target of each leg. The foot-end compensation and inverse kinematics solution module compares the foot-end target with the foot-end reachable workspace. When the foot-end target is within the reachable workspace, it is used as the corrected foot-end target; when the foot-end target exceeds the reachable workspace, it is corrected to the boundary of the reachable workspace. The module performs inverse kinematics solution based on the corrected foot-end target, leg link length, and joint travel range, generating joint logic angles corresponding to the hip roll, hip pitch, and knee joints of each leg. If the inverse kinematics solution does not generate valid joint logic angles, it retains the valid joint logic angles from the previous control cycle for the corresponding leg and passes the state of the inverse kinematics solution to the actuator mapping and safety control module. The module arranges the joint logic angles according to the correspondence between the leg and joints and passes the arranged joint logic angles to the actuator mapping and safety control module.

[0123] The actuator mapping and safety control module 04, connected to the foot compensation and inverse kinematics solving module and the running state determination module, is used to map the joint logic angles to hardware target angles and limit them according to the joint travel range, generating servo control commands; when the stable running state is the safe state or the command transmission fails, the gait cycle propulsion is stopped and a safe posture command is generated. Specifically, the actuator mapping and safety control module receives the joint logic angles output by the foot compensation and inverse kinematics solving module and receives the stable running state output by the running state determination module; the joint logic angles are matched with the actuator channel mapping relationship according to the correspondence between each leg and each joint. The actuator mapping and safety control module converts the joint logic angles into corresponding hardware target angles according to the installation direction, zero position, and proportional relationship of each joint; when the hardware target angle is greater than the upper limit of the corresponding joint travel range, it is corrected to the upper limit; when the hardware target angle is less than the lower limit of the corresponding joint travel range, it is corrected to the lower limit. The actuator mapping and safety control module also compares the hardware target angle of the current control cycle with that of the previous control cycle; when the angle change between the two exceeds the maximum angle change in a single control cycle, the hardware target angle of the current control cycle is corrected according to the maximum angle change in a single control cycle. The hardware target angles after actuator channel mapping and limiting are arranged according to the actuator channels to form the servo control command, which is then output to the servo actuator controller. The transmission status of the servo control command is recorded within the current control cycle; if no corresponding transmission status is received or the transmission status is recorded as a failure, the control cycle is marked as a command transmission failure. When the stable operating state is the safe state or a command transmission failure occurs, the actuator mapping and safety control module stops the gait cycle advancement of the motion mode and foot trajectory generation module, stops updating the attitude compensation amount and the contact compensation amount of each leg, and calls the hardware target angle corresponding to the safe attitude to generate the safe attitude command. The safe attitude command is limited according to the joint travel range and the maximum angle change in a single control cycle before being output to the servo actuator controller. After the servo actuator controller executes the servo control command or safety attitude command, the new attitude parameters, angular velocity data, contact pressure of each leg foot end, and telemetry timestamp are transmitted to the running state determination module in the next control cycle. The running state determination module regenerates the candidate running state and stable running state accordingly. After the stable running state releases the safety state, the motion mode and foot trajectory generation module calls the mode parameter set again and generates the basic foot trajectory.

[0124] It should be noted that, in this document, the terms "comprising," "including," and any other variations are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Specific examples have been used in this document to illustrate the principles and implementation methods of the present invention. These examples are merely for the purpose of helping to understand the method and core ideas of the present invention. The above descriptions are only preferred embodiments of the present invention. It should be pointed out that, due to the limitations of written expression and the objective existence of infinite specific structures, those skilled in the art can make several improvements, modifications, or variations without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, variations, or combinations, or the direct application of the concept and technical solution of the present invention to other situations without modification, should all be considered within the scope of protection of the present invention.

Claims

1. An adaptive control method for composite walking of a pipeline robot, characterized in that, include: S100 receives attitude parameters, angular velocity data, contact pressure at the end of each leg, motion commands, and telemetry timestamps, and calculates the age of the data. Candidate operating states are generated based on attitude disturbance, angular velocity disturbance, pressure unevenness and data age. Stable operating states are determined according to the first confirmation time or the second confirmation time of the state transition direction. When the data age exceeds the data timeout threshold, the stable operating state is updated to a safe state. S200. Based on the stable operating state and the movement command, determine the target movement mode, call the mode parameter set including gait cycle, swing duty cycle, cycle offset of each leg and foot reference position offset of each leg, and generate the basic foot trajectory. S300. Generate attitude compensation amount based on the deviation between attitude parameters and calibrated attitude and angular velocity data. Generate contact compensation amount for each leg based on the pressure error between the contact pressure at the foot end of each leg and the target pressure. Convert the two types of compensation amount into foot end reference position correction amount, superimpose it with the basic foot end trajectory, and generate joint logic angle through foot end reachable workspace correction and inverse kinematics solution. S400: Map the joint logic angle to the hardware target angle and limit it according to the joint travel range to generate servo control commands; when the stable operating state is a safe state or the command transmission fails, stop the gait cycle propulsion and generate a safe attitude command.

2. The adaptive control method for composite walking of a pipeline robot according to claim 1, characterized in that, The attitude parameters include pitch angle and roll angle; The attitude disturbance is generated based on the deviation of the pitch angle and roll angle relative to the calibrated attitude; The angular velocity disturbance is generated based on the X-axis angular velocity and the Y-axis angular velocity; The pressure unevenness is generated based on the pressure error between the contact pressure of each leg and foot end and the corresponding target pressure. The data age is the difference between the current time and the most recent valid telemetry timestamp.

3. The adaptive control method for composite walking of a pipeline robot according to claim 1, characterized in that, In addition to the aforementioned safety status, when the level of the candidate operating status is higher than the current stable operating status, the first confirmation time is used, and the stable operating status is updated after the first confirmation time is continuously met upon entering the threshold. When the level of the candidate operating state is lower than the current stable operating state, the second confirmation time is used, and the stable operating state is updated after the exit threshold continues to meet the second confirmation time. The second confirmation time is longer than the first confirmation time, and the corresponding confirmation time is reset when the candidate running state changes.

4. The adaptive control method for composite walking of a pipeline robot according to claim 1, characterized in that, The target movement mode is selected from safe standing mode, normal walking mode, high-pass walking mode, longitudinal rolling mode, and lateral rolling mode; The high-speed walking mode corresponds to the posture compensation being enabled, and the lateral rolling mode corresponds to the contact compensation being enabled. If the current motion mode and the target motion mode do not both generate a mode switching state, an intermediate posture between the current motion mode and the target motion mode is invoked, and the current motion mode is updated after the intermediate posture is completed.

5. The adaptive control method for composite walking of a pipeline robot according to claim 1, characterized in that, Generating the basic foot trajectory includes: Add the time within the gait cycle to the cycle offset of the corresponding leg and normalize it according to the gait cycle to generate the local cycle of each leg; By comparing the local period of each leg with the swing duty cycle, it is determined whether the corresponding leg is in the swing phase or the support phase. In the swing phase, a foot-lifting trajectory and a forward-moving trajectory are generated; in the support phase, a retreating trajectory is generated. The trajectory is corrected based on the swing gain of each leg, and the offset of the reference position of each leg foot is superimposed.

6. The adaptive control method for composite walking of a pipeline robot according to claim 1, characterized in that, Generating the attitude compensation amount includes: Calculate the attitude deviations of the pitch and roll angles relative to the calibrated attitude; Based on the attitude deviation and angular velocity data, a proportional-differential calculation is performed and the amplitude is limited according to the maximum compensation amount; The attitude compensation amount corresponding to the pitch angle is converted into the foot reference position correction amount for the front and rear legs, respectively, and the attitude compensation amount corresponding to the roll angle is converted into the foot reference position correction amount for the left and right legs, respectively.

7. The adaptive control method for composite walking of a pipeline robot according to claim 1, characterized in that, The generation of contact compensation values ​​for each leg includes: Calculate the pressure error between the contact pressure at the foot of each leg and the corresponding target pressure; When the pressure error is within the pressure dead zone, the corresponding contact compensation amount is maintained. When the pressure error is outside the pressure dead zone, the corresponding contact compensation amount is discretely accumulated based on the pressure error and the sampling time interval. The contact compensation amount is limited by the maximum compensation amount and converted into a correction amount for the foot end reference position along the normal direction of the pipe wall of the corresponding leg.

8. The adaptive control method for composite walking of a pipeline robot according to claim 1, characterized in that, The reachable working space at the foot end is determined based on the length of the leg linkage and the range of joint travel; When the foot target, after the two types of compensation amounts are superimposed on the basic foot trajectory, exceeds the foot reachable workspace, the foot target is corrected to the boundary of the foot reachable workspace; The inverse kinematics of the corrected foot target are solved, and the joint logic angles are converted into hardware target angles according to the actuator channel mapping relationship, installation direction, zero position and proportional relationship. The target angle of the hardware is limited based on the joint travel range and the maximum angle change in a single control cycle.

9. The adaptive control method for composite walking of a pipeline robot according to claim 1, characterized in that, In the safe state, maintain the safe posture command, reset the confirmation time of the candidate running state, and stop updating the posture compensation amount and the contact compensation amount of each leg; If the age of the continuously received attitude parameters, angular velocity data, and foot contact pressure data does not exceed the data timeout threshold, the stable operating state is restored from the safe state according to the second confirmation time.

10. A composite walking adaptive control system for a pipeline robot, applied to the method according to any one of claims 1 to 9, characterized in that, include: The operating status determination module is used to receive attitude parameters, angular velocity data, contact pressure at each leg and foot end, motion commands and telemetry timestamps, calculate data age, generate candidate operating states based on attitude disturbance, angular velocity disturbance, pressure unevenness and the data age, determine stable operating state according to the first confirmation time or the second confirmation time corresponding to the state transition direction, and update the stable operating state to a safe state when the data age exceeds the data timeout threshold. The movement mode and foot trajectory generation module is connected to the running state determination module. It is used to determine the target movement mode based on the stable running state and the movement command, and call the mode parameter set including gait cycle, swing duty cycle, cycle offset of each leg and foot reference position offset of each leg to generate the basic foot trajectory. The foot compensation and inverse kinematics solution module connects the motion mode and foot trajectory generation module and the running state determination module. It is used to generate posture compensation amount based on the deviation between the posture parameters and the calibrated posture and the angular velocity data, generate contact compensation amount for each leg based on the pressure error between the foot contact pressure of each leg and the target pressure, convert the posture compensation amount and the contact compensation amount of each leg into foot reference position correction amount and superimpose it with the basic foot trajectory, and generate joint logic angle through foot reachable workspace correction and inverse kinematics solution. The actuator mapping and safety control module, connected to the foot compensation and inverse kinematics solution module and the running state determination module, is used to map the joint logic angle to a hardware target angle and limit it according to the joint travel range, and generate servo control commands; when the stable running state is a safe state or the command transmission fails, the gait cycle propulsion is stopped and a safe posture command is generated.

Images