An adaptive gait planning method and system based on whole-body motion coordination optimization

Abstract

This invention relates to an adaptive gait planning method and system based on whole-body motion cooperative optimization, belonging to the field of robot control. The method includes obtaining virtual terrain impedance through multimodal perception fusion; performing dynamic CPG modulation based on the virtual terrain impedance to obtain an adaptive gait rhythm and generate a basic gait pattern, the adaptive gait rhythm including leg phase, gait amplitude, and gait frequency; and correcting the basic gait pattern based on whole-body motion cooperative optimization to obtain the ideal foot placement point and the desired centroid acceleration vector. This invention integrates virtual terrain impedance perception, dynamic CPG modulation, and whole-body motion cooperative optimization, achieving unified processing of static terrain properties and dynamic disturbances, significantly improving the robot's adaptability and robustness in unknown environments.

Description

Technical Field

[0001] This invention belongs to the field of robot control technology, specifically relating to an adaptive gait planning method and system based on whole-body motion coordination optimization. Background Technology

[0002] With the widespread application of robotics in rescue, exploration, and service fields, higher demands are placed on the robot's mobility in complex terrains (such as sand, gravel, and slopes). Existing gait planning methods mainly fall into three categories: First, model-based control methods rely on accurate dynamic models and prior environmental knowledge. They perform well in structured environments, but in unknown or dynamic environments, model errors and external disturbances can easily lead to stability problems, such as the robot falling over when subjected to sudden external forces or ground movement. Second, sensor-based reflective control methods, such as using IMUs or force sensors to adjust joint torques in real time. However, these methods are usually slow to react and lack predictability, making it difficult to handle changes in terrain properties (such as switching from hard ground to soft ground), resulting in uncoordinated gait or sinking. Third, bio-inspired methods, such as central pattern generators (CPGs), can generate rhythmic gait, but the parameters are mostly preset manually, making them unable to adapt to terrain changes and lacking overall body movement coordination. The common shortcomings of existing technologies are the lack of online perception of terrain attributes, the inability to balance static terrain adaptation and dynamic disturbance suppression, and the disconnect between gait parameters and whole-body movement, which results in robots moving slowly, consuming a lot of energy and having poor stability in complex environments. Summary of the Invention

[0003] To address the aforementioned problems in the existing technology, this invention provides an adaptive gait planning method and system based on whole-body motion coordination optimization.

[0004] The objective of this invention can be achieved through the following technical solutions:

[0005] An adaptive gait planning method based on whole-body motion coordination optimization is provided. The implementation of the adaptive gait planning method based on whole-body motion coordination optimization includes the following steps:

[0006] Step S1: Obtain virtual terrain impedance through multimodal sensing fusion;

[0007] Step S2: Based on the virtual terrain impedance, perform dynamic CPG modulation to obtain an adaptive gait rhythm and generate a basic gait pattern. The adaptive gait rhythm includes leg phase, gait amplitude, and gait frequency.

[0008] Step S3: Based on the whole-body motion coordination optimization, correct the basic gait pattern to obtain the ideal landing point and the desired center of mass acceleration vector.

[0009] Preferably, step S1 specifically includes:

[0010] The robot's vertical force and sinking velocity are obtained; a moving average filter is applied to the vertical force and sinking velocity to obtain the virtual terrain impedance, mathematically described as follows: ,in, Let be the virtual terrain impedance at time t. Let be the perpendicular force at time t. Let be the sinking velocity at time t. To prevent division by zero of extremely small positive numbers.

[0011] Preferably, the dynamic CPG modulation in step S2 specifically includes:

[0012] Step S201: Preset a CPG model to obtain the robot's leg phase;

[0013] Step S202: Obtain the gait amplitude and the gait frequency based on the virtual terrain impedance.

[0014] Preferably, step S202 specifically includes:

[0015] Obtain the robot's nominal gait amplitude and nominal gait frequency, and use the typical impedance value measured on a flat hard surface as the reference impedance;

[0016] The average value of the virtual terrain impedance over a preset time range is calculated and denoted as the normal virtual terrain impedance. An impedance normalization factor is then obtained based on the normal virtual terrain impedance and the reference impedance, mathematically described as follows: ,in, This is the impedance normalization factor. As the reference impedance, For normal virtual terrain impedance;

[0017] The gait amplitude and gait frequency are obtained based on the nominal gait amplitude, the nominal gait frequency, and the impedance normalization factor, and are mathematically described as follows: ,in, For gait amplitude, The nominal gait amplitude, This is the amplitude adaptive gain coefficient; ,in, For gait frequency, The nominal gait frequency, This is the frequency-adaptive gain coefficient.

[0018] Preferably, the whole-body motion coordination optimization in step S3 specifically includes:

[0019] Step S301: Measure the actual center-of-mass acceleration vector and obtain the predicted center-of-mass acceleration vector. Based on the actual and predicted center-of-mass acceleration vectors, perform perturbation observation to obtain the perturbation force vector, mathematically described as follows: ,in, The disturbance force vector, For robot quality, This is the actual center-of-mass acceleration vector. To predict the acceleration vector of the center of mass;

[0020] Step S302: Perform coordinated foot placement control based on the disturbance force vector, that is, solve for the ideal landing point of the next swing leg and the desired center of mass acceleration vector.

[0021] Preferably, the coordinated gait placement control in step S302 specifically includes:

[0022] The predicted time constant and the desired centroid acceleration vector are obtained through the PD controller, mathematically described as follows: Let the desired center-of-mass acceleration vector be... As the reference centroid position vector, Let be the centroid position vector. As the reference centroid velocity vector, Let the velocity vector be the center of mass. This is the proportional gain coefficient. The differential gain coefficient;

[0023] The ideal landing point coordinates are obtained based on the expected centroid acceleration vector and the predicted time constant, mathematically described as follows: ,in, For the ideal landing point coordinates, Let the velocity vector be the center of mass. Forecast time constant Let the desired center-of-mass acceleration vector be... The nominal center of gravity height, This is the acceleration due to gravity.

[0024] An adaptive gait planning system based on whole-body motion coordination optimization is used to execute the adaptive gait planning method based on whole-body motion coordination optimization described above, including a perception fusion module, a basic gait pattern generation module, and a coordination optimization module;

[0025] The perception fusion module is used to obtain virtual terrain impedance through multimodal perception fusion;

[0026] The basic gait pattern generation module is used to perform dynamic CPG modulation based on the virtual terrain impedance to obtain an adaptive gait rhythm and generate a basic gait pattern. The adaptive gait rhythm includes leg phase, gait amplitude and gait frequency.

[0027] The collaborative optimization module is used to correct the basic gait pattern based on whole-body motion collaborative optimization to obtain the ideal landing point and the desired center of mass acceleration vector.

[0028] The beneficial effects of this invention are as follows:

[0029] (1) By sensing the ground hardness in real time through virtual terrain impedance, the CPG is driven to dynamically adjust the gait amplitude and frequency, increase the leg lift height and reduce the gait frequency on soft ground to prevent sinking, and reduce the impact on hard ground.

[0030] (2) Through whole-body motion coordination optimization, based on disturbance observation and foot placement control, the landing point and center of mass acceleration are corrected in real time, effectively offsetting dynamic disturbances (such as stepping on stones or slipping), avoiding robot falls, significantly improving the robot's motion stability and adaptability in complex terrain (such as sand, gravel, and slopes), while reducing dependence on accurate environmental models and enhancing autonomy. Attached Figure Description

[0031] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0032] Figure 1 This is a flowchart of the adaptive gait planning method based on whole-body motion coordination optimization according to the present invention. Detailed Implementation

[0033] To better understand the invention, various aspects of the invention will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are merely illustrative of exemplary embodiments of the invention and are not intended to limit the scope of the invention in any way. Throughout the specification, the expression "and / or" includes any and all combinations of one or more of the associated listed items. As used herein, the terms "approximately," "about," and similar terms are used as expressions of approximation, not as expressions of degree, and are intended to describe inherent deviations in measured or calculated values ​​that will be recognized by those skilled in the art. Furthermore, the order in which the steps are described in this invention does not necessarily indicate the order in which these steps occur in actual operation, unless otherwise expressly defined or deduced from the context.

[0034] It should also be understood that expressions such as "comprising," "including," "having," "containing," and / or "comprising" are open-ended rather than closed-ended expressions in this specification, indicating the presence of the stated features, elements, and / or components, but not excluding the presence of one or more other features, elements, components, and / or combinations thereof. Furthermore, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features, not just individual elements in the list. Additionally, when describing embodiments of the invention, the word "may" is used to mean "one or more embodiments of the invention." And the term "exemplary" is intended to refer to examples or illustrations.

[0035] Unless otherwise specified, all terms used herein (including engineering and technical terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that, unless expressly stated herein, terms defined in common dictionaries shall be interpreted as having the meaning consistent with their meaning in the context of the relevant art, and not in an idealized or overly formalized sense.

[0036] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other. The invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0037] Example 1:

[0038] Please see Figure 1 An adaptive gait planning method based on whole-body motion coordination optimization includes:

[0039] S1: Virtual terrain impedance is obtained through multimodal perception fusion;

[0040] S2: Based on the virtual terrain impedance, dynamic CPG (central pattern generator) modulation is performed to obtain an adaptive gait rhythm and generate the robot's basic gait pattern (that is, the leg phase, gait amplitude and gait frequency are input together into a foot trajectory generator to generate the desired trajectory of the robot's feet in the air and the periodic landing point sequence). The adaptive gait rhythm includes leg phase, gait amplitude and gait frequency.

[0041] S3: Based on the whole-body motion collaborative optimization, the basic gait pattern is corrected to obtain the ideal footing point and the desired center of mass acceleration vector. The basic gait pattern in step S2 is obtained based on virtual terrain impedance. However, virtual terrain impedance can only reflect the static properties of the terrain and cannot capture dynamic disturbances (such as suddenly stepping on a rolling stone). Therefore, the new footing point obtained through whole-body motion collaborative optimization covers or corrects the footing point in S2. The final instruction received by the robot's bottom controller is: to swing to the position determined in S3 and adjust the center of gravity with the gait amplitude and gait frequency determined in S2.

[0042] In this embodiment, step S1 can be implemented through the following steps:

[0043] S101: The vertical force of the robot is obtained by directly measuring the force along the Z-axis (vertical direction) through a six-dimensional force / torque sensor installed on the robot's ankle or foot; at the same time, the rotation angle of each joint (such as hip, knee, ankle) is measured through the robot's joint encoder, and the position of the robot's foot in three-dimensional space is obtained by using the robot's forward kinematics model. The z-component of the foot position is numerically differentiated to obtain the velocity of the robot's foot in the vertical direction, that is, the robot's sinking speed.

[0044] S102: Perform a moving average filter on the vertical force and the sinking velocity to obtain the virtual terrain impedance, mathematically described as follows: ,in, The virtual terrain impedance at time t (kg / s) 2 The value reflects the degree to which the earth resists subsidence. The higher the value, the harder the earth's surface; the lower the value, the softer the earth's surface and the more easily it subsides. Let t be the vertical force (N) at time t. Let t be the sinking velocity (m / s) at time t. To prevent division by extremely small positive numbers, the formula must remain valid when the robot's foot is stationary or nearly stationary. Example: On a hard surface (concrete or hard rock, etc.), at the instant the robot's foot touches the ground... It dropped rapidly to 0, and A sharp rise led to Extremely high, meaning the control system is told that the contact surface is extremely hard; on soft ground (such as sand), the robot's feet will continue to sink. Larger, and due to the flow and compression of the sand, Slow growth ultimately led to "Very low" means that the control system is telling the contact surface is very soft and the robot is sinking.

[0045] In this embodiment, the dynamic CPG modulation specifically refers to:

[0046] S201: A pre-defined CPG model (e.g., a set of coupled nonlinear oscillators) is used to obtain the robot's leg phase, which is a nominal, periodic rhythmic signal that determines the timing of the robot's leg in the swing phase and the support phase (when the leg phase is in the range [0,π), the robot's leg lifts up and swings forward in the air; when the leg phase is in the range [π,2π), the robot's leg contacts the ground, supports the body, and pushes off the ground backward).

[0047] S202: The gait amplitude and the gait frequency are obtained based on the virtual terrain impedance.

[0048] In this embodiment, step S202 can be implemented through the following steps:

[0049] S202-1: Conduct calibration tests on a flat, hard surface to obtain the robot's nominal gait amplitude and nominal gait frequency, and use the typical impedance value measured on the flat, hard surface as the reference impedance.

[0050] S202-2: Calculate the average value of the virtual terrain impedance within a preset time range (the time span should not be too large, preferably 100-200 milliseconds), denoted as the normal virtual terrain impedance, representing the normal state of the current terrain. Based on the normal virtual terrain impedance and the reference impedance, obtain the impedance normalization factor, mathematically described as follows: ,in, This is the impedance normalization factor. As the reference impedance, For normal virtual terrain impedance;

[0051] S202-3: The gait amplitude and gait frequency are obtained based on the nominal gait amplitude, the nominal gait frequency, and the impedance normalization factor, mathematically described as follows: ,in, This refers to the gait amplitude (which directly determines the height at which the robot lifts its legs). The nominal gait amplitude, This is the amplitude adaptive gain coefficient; ,in, For gait frequency, The nominal gait frequency, For frequency-adaptive gain coefficients; Example: when the ground softens ( When (decreases), (1- ) increases, leading to Enlarging means the robot needs to raise its legs higher to prevent it from getting stuck; at the same time, it is usually set to... If it is a negative value, then <1, the robot reduces its step frequency and walks more slowly and cautiously on soft ground.

[0052] In this embodiment, the whole-body motion coordination optimization specifically refers to:

[0053] S301: The actual center-of-gravity acceleration vector of the robot is obtained through IMU measurement, and the predicted center-of-gravity acceleration vector is obtained through the whole-body dynamics model based on joint torques and ground reaction forces. Based on the actual center-of-gravity acceleration vector and the predicted center-of-gravity acceleration vector, perturbation observation is performed to obtain the perturbation force vector, mathematically described as follows: ,in, It is the perturbation vector (a two-dimensional vector in the x and y directions, with the dimension N). The mass of the robot is expressed in kg. The actual center-of-mass acceleration vector (a two-dimensional vector in the x and y directions, with dimensions m / s²). 2 ), To predict the acceleration vector of the center of mass (a two-dimensional vector in the x and y directions, with dimensions m / s²) 2 );

[0054] S302: Based on the disturbance force vector, perform coordinated foot placement control, that is, solve the ideal landing point of the next swing leg and the desired center of mass acceleration vector, so that the landing point in S2 is covered or finely adjusted by the ideal landing point. At the same time, the robot's torso controller will adjust the body posture according to the desired center of mass acceleration vector, such as actively leaning forward or backward, to generate the required center of mass acceleration.

[0055] In this embodiment, the coordinated foot placement control specifically refers to:

[0056] S302-1: The predicted time constant (representing how long the controller hopes to stabilize the robot) and the expected center-of-mass acceleration vector are obtained through the PD controller. Mathematically, this is described as follows: Let the desired center-of-mass acceleration vector be... As the reference centroid position vector, Let be the centroid position vector. The reference centroid velocity vector (two-dimensional). The velocity vector of the center of mass (two-dimensional). This is the proportional gain coefficient. The differential gain coefficient;

[0057] S302-2: Based on the desired centroid acceleration vector and the predicted time constant, the coordinates of the ideal landing point are obtained, mathematically described as follows: ,in, For the ideal landing point coordinates, Let the velocity vector be the center of mass. Forecast time constant Let the desired center-of-mass acceleration vector be... The nominal center of gravity height, For gravitational acceleration; this step does not determine where the robot's feet should be placed in isolation, but rather based on the current overall motion state ( and ) and perceived disturbances ( The optimal landing point is calculated by combining the results. At the same time, the body's center of gravity also needs to be adjusted accordingly to match this new landing point. This ultimately achieves overall optimization. Example: The robot's right foot steps on a loose pebble (it senses this). The body leans to the left and forward (causing) and (Change), to counteract this tilt, the left foot that is about to step out should not land at point A as originally planned, but should be further to the left and forward, landing at point B (calculated). (The coordinates of point B), and at the same time, the body's center of gravity (center of mass) should be adjusted slightly to the right and rear (according to...). control).

[0058] Example 2:

[0059] An adaptive gait planning system based on whole-body motion coordination optimization includes a perception fusion module, a basic gait pattern generation module, and a coordination optimization module;

[0060] The perception fusion module is used to obtain virtual terrain impedance through multimodal perception fusion;

[0061] The basic gait pattern generation module is used to perform dynamic CPG (central pattern generator) modulation based on the virtual terrain impedance to obtain an adaptive gait rhythm and generate the robot's basic gait pattern (that is, inputting leg phase, gait amplitude and gait frequency together into a foot trajectory generator to generate the desired trajectory of the robot's feet in the air and a periodic landing point sequence). The adaptive gait rhythm includes leg phase, gait amplitude and gait frequency.

[0062] The collaborative optimization module is used to correct the basic gait pattern based on whole-body motion collaborative optimization to obtain the ideal footing point and the desired centroid acceleration vector. The basic gait pattern in step S2 is obtained based on virtual terrain impedance. However, virtual terrain impedance can only reflect the static properties of the terrain and cannot capture dynamic disturbances (such as suddenly stepping on a rolling stone). Therefore, the new footing point obtained through whole-body motion collaborative optimization covers or corrects the footing point in S2. The final instruction received by the robot's underlying controller is: to swing to the position determined in S3 and adjust the center of gravity with the gait amplitude and gait frequency determined in S2.

[0063] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An adaptive gait planning method based on whole-body motion coordination optimization, characterized in that, The implementation of the adaptive gait planning method based on whole-body motion coordination optimization includes the following steps: Step S1: Obtain virtual terrain impedance through multimodal sensing fusion; Step S2: Based on the virtual terrain impedance, perform dynamic CPG modulation to obtain the robot's adaptive gait rhythm and generate a basic gait pattern. The adaptive gait rhythm includes leg phase, gait amplitude, and gait frequency. Step S3: Based on whole-body motion coordination optimization, perform coordinated foot placement control to correct the basic gait pattern and obtain the robot's ideal foot placement point and desired centroid acceleration vector; the coordinated foot placement control specifically includes: The predicted time constant and the desired centroid acceleration vector are obtained through the PD controller, mathematically described as follows: Let the desired center-of-mass acceleration vector be... As the reference centroid position vector, Let be the centroid position vector. As the reference centroid velocity vector, Let the velocity vector be the center of mass. This is the proportional gain coefficient. The differential gain coefficient; The ideal landing point coordinates are obtained based on the expected centroid acceleration vector and the predicted time constant, mathematically described as follows: ,in, For the ideal landing point coordinates, To predict the time constant, The nominal center of gravity height, It is the acceleration due to gravity. The disturbance force vector, For robot quality.

2. The adaptive gait planning method based on whole-body motion coordination optimization according to claim 1, characterized in that, Step S1 specifically includes: The robot's vertical force and sinking velocity are obtained; a moving average filter is applied to the vertical force and sinking velocity to obtain the virtual terrain impedance, mathematically described as follows: ,in, Let be the virtual terrain impedance at time t. Let be the perpendicular force at time t. Let be the sinking velocity at time t. To prevent division by zero of extremely small positive numbers.

3. The adaptive gait planning method based on whole-body motion coordination optimization according to claim 1, characterized in that, The dynamic CPG modulation in step S2 specifically refers to: Step S201: Preset a CPG model to obtain the robot's leg phase; Step S202: Obtain the gait amplitude and the gait frequency based on the virtual terrain impedance.

4. The adaptive gait planning method based on whole-body motion coordination optimization according to claim 3, characterized in that, Step S202 specifically includes: Obtain the robot's nominal gait amplitude and nominal gait frequency, and use the typical impedance value measured on a flat hard surface as the reference impedance; The average value of the virtual terrain impedance over a preset time range is calculated and denoted as the normal virtual terrain impedance. An impedance normalization factor is then obtained based on the normal virtual terrain impedance and the reference impedance, mathematically described as follows: ,in, This is the impedance normalization factor. As the reference impedance, For normal virtual terrain impedance; The gait amplitude and gait frequency are obtained based on the nominal gait amplitude, the nominal gait frequency, and the impedance normalization factor, and are mathematically described as follows: ,in, For gait amplitude, The nominal gait amplitude, This is the amplitude adaptive gain coefficient; ,in, For gait frequency, The nominal gait frequency, This is the frequency-adaptive gain coefficient.

5. The adaptive gait planning method based on whole-body motion coordination optimization according to claim 1, characterized in that, The whole-body motion coordination optimization in step S3 specifically refers to: Step S301: Measure the actual center-of-mass acceleration vector and obtain the predicted center-of-mass acceleration vector. Based on the actual and predicted center-of-mass acceleration vectors, perform perturbation observation to obtain the perturbation force vector, mathematically described as follows: ,in, The disturbance force vector, For robot quality, This is the actual center-of-mass acceleration vector. To predict the acceleration vector of the center of mass; Step S302: Perform coordinated foot placement control based on the disturbance force vector, that is, solve for the ideal landing point of the next swing leg and the desired center of mass acceleration vector.

6. An adaptive gait planning system based on whole-body motion coordination optimization, characterized in that, The system is applied to the adaptive gait planning method based on whole-body motion coordination optimization as described in any one of claims 1-5, and includes a perception fusion module, a basic gait pattern generation module, and a coordination optimization module; The perception fusion module is used to obtain virtual terrain impedance through multimodal perception fusion; The basic gait pattern generation module is used to perform dynamic CPG modulation based on the virtual terrain impedance to obtain the robot's adaptive gait rhythm and generate a basic gait pattern. The adaptive gait rhythm includes leg phase, gait amplitude and gait frequency. The collaborative optimization module is used to correct the basic gait pattern based on whole-body motion collaborative optimization to obtain the robot's ideal footing point and desired center of mass acceleration vector.

Images