Biped robot control method and device, biped robot, and readable storage medium
The method controls bipedal robots by determining support periods and using ankle joint moment vectors to manage center of gravity motion, addressing underdriven dynamics and improving adaptability in complex environments.
Patent Information
- Application Number
- JP2022210536
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2022-08-10
- Filing Date
- 2022-12-27
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2042-12-27
AI Technical Summary
Bipedal robots face challenges in controlling their motion due to inherent underdriven dynamics during single-leg support phases, as they cannot effectively manage the center of mass momentum, limiting their adaptability in complex and dynamic environments.
A method and apparatus for controlling bipedal robots by determining the support period and obtaining current forward and lateral velocities of the center of gravity, calculating a moment vector for the ankle joint using quadratic programming, and controlling the robot's motion based on this vector to achieve full driveability.
This approach enables robust control of bipedal robots in both fully driven and underdriven states, enhancing their adaptability and stability in diverse terrains by ensuring complete control over the center of gravity and posture.
Smart Images

Figure 0007577108000010 
Figure 0007577108000011 
Figure 0007577108000012
Abstract
Description
[Technical field]
[0001] The present disclosure relates to the field of control technology, and in particular to a control method and device for a biped robot, a biped robot, and a readable storage medium. [Background technology]
[0002] At present, mobile robots play an important role in our daily lives and provide various conveniences to people. Mobile robots are divided into wheeled, caterpillar and bipedal robots. Wheeled and caterpillar robots are greatly restricted in their use in rough, complex and diverse terrain environments, such as gravel ground and rough ground, which affect their passage. Bipedal robots can work only by touching the ground, and have lower requirements for the ground, so they have better prospects for use.
[0003] Take the biped robot as an example, it is a biped / pseudo-robot, based on its bionic structural characteristics, it can adapt to more complex terrains and working environments, and perform dangerous or repetitive tasks instead of humans. The walking and running of biped / pseudo-robots often have the characteristics of high nonlinearity, under-actuation, and mixed continuous and discrete systems, so that a suitable biped robot walking control method is particularly important.
[0004] In locomotion, whether the external contact environment is structured or unstructured, whether the terrain is ideal or uneven, the dynamics of a biped robot are inherently underactuated. This is because (1) a single support phase (SSP) is inevitably present in the gait of legged locomotion, and (2) the supporting leg can only receive support and friction forces from the contact surface, i.e., the ground cannot exert a pulling force on the biped robot. Therefore, in the case of high-dynamic locomotion, during the single support phase, all the actuable degrees of freedom of the supporting leg of the biped robot cannot realize complete control of the floating base (or center of gravity momentum). Summary of the Invention [Problem to be solved by the invention]
[0005] The present disclosure provides a biped robot control method and device, a biped robot, and a readable storage medium to address the deficiencies of related art.
[0006] According to a first aspect of an embodiment of the present disclosure, a method for controlling a bipedal robot is provided, the method including the steps of acquiring a supporting period of the bipedal robot in a current state, acquiring a current forward velocity and a current lateral velocity of the center of gravity of the bipedal robot at the end of the supporting period, acquiring a moment vector of an ankle joint of a supporting foot of the bipedal robot based on the current forward velocity and the current lateral velocity, acquiring a target moment vector based on the moment vector and a predetermined quadratic programming model, and controlling motion of the bipedal robot based on the target moment vector.
[0007] Optionally, the step of acquiring a support period of the biped robot in a current state includes a step of acquiring a current time and a normal component force of the ground that a swinging foot of the biped robot receives; a step of determining whether the current time satisfies a first condition or a second condition, and a step of determining whether the current time and the normal component force satisfy a third condition, wherein the first condition refers to the current time exceeding the sum of a start time of a single-leg support period and a duration of the single-leg support period, and the second condition refers to the current time exceeding the sum of a start time of a biped support period and a duration of the biped support period. the third condition indicating that the current time exceeds the sum of the start time of a single leg supporting period and half the duration of the single leg supporting period and the normal force is greater than the product of a preset coefficient and the gravity of the biped robot; and if the current time satisfies the first condition or the current time and the normal force satisfy the third condition, determining that the biped robot is in a biped supporting period, and if the current time satisfies the second condition, determining that the biped robot is in a single leg supporting period.
[0008] Optionally, the support period includes a single leg support period, the forward velocity includes a first predicted forward velocity, and the lateral velocity includes a first predicted lateral velocity, and the step of acquiring a current forward velocity and a current lateral velocity of the center of gravity of the biped robot at an end of the support period includes the steps of acquiring a first forward position and a first forward velocity and a first lateral position and a first lateral velocity of the center of gravity of the biped robot relative to the biped robot, estimating a first predicted forward velocity of the biped robot based on a first preset forward velocity model, the first forward position, and the first forward velocity, and estimating a first predicted lateral velocity of the biped robot based on a first preset lateral velocity model, the first lateral position, and the first lateral velocity.
[0009] Optionally, the step of acquiring a moment vector of an ankle joint of a supporting foot of the bipedal robot based on the current forward velocity and the current lateral velocity includes the steps of acquiring the forward velocity of the current step and the lateral velocity of the current step of the center of gravity, and the forward velocity of the next step and the lateral velocity of the next step of the center of gravity in the current state at the end of a single-leg support phase in the current state, and acquiring a moment vector of an ankle joint of the supporting foot of the bipedal robot based on the forward velocity of the current step, the lateral velocity of the current step, the forward velocity of the next step, and the lateral velocity of the next step.
[0010] Optionally, the method further includes the steps of obtaining a forward displacement of a current step and a lateral displacement of a current step of a swinging foot of the supporting foot relative to the center of gravity at the end of the one-leg support period, and obtaining a forward velocity of a next step and a lateral velocity of a next step of the swinging foot based on the forward displacement of the current step of the swinging foot, the forward velocity of the current step of the center of gravity, and the first predicted forward velocity.
[0011] Optionally, the support period includes a bipedal support period, the forward velocity includes a second predicted forward velocity, and the lateral velocity includes a second predicted lateral velocity, and the step of acquiring the current forward velocity and the current lateral velocity of the center of gravity of the bipedal robot at the end of the support period includes the steps of acquiring a second forward position and a second forward velocity and a second lateral position and a second lateral velocity of the center of gravity of the bipedal robot relative to the bipedal robot, estimating a second predicted forward velocity of the bipedal robot based on a second preset forward velocity model, the second forward position, and the second forward velocity, and estimating a second predicted lateral velocity of the bipedal robot based on a second preset lateral velocity model, the second lateral position, and the second lateral velocity.
[0012] Optionally, the step of acquiring a moment vector of an ankle joint of a supporting foot of the bipedal robot based on the current forward velocity and the current lateral velocity includes the steps of acquiring a forward velocity of a current step of the center of gravity at the end of a bipedal support period in a current state and a lateral velocity of the current step, acquiring a moment vector of a first supporting foot based on the forward velocity of the current step and the second predicted forward velocity, and acquiring a moment vector of a second supporting foot based on the lateral velocity of the current step and the second predicted lateral velocity, wherein the first supporting foot refers to a supporting foot in a single supporting period prior to the bipedal support period, and the second supporting foot refers to a swinging foot in a single supporting period prior to the bipedal support period.
[0013] Optionally, the step of obtaining a target moment vector based on the moment vector and a predetermined quadratic programming model includes the steps of obtaining a general acceleration vector of the bipedal robot and an external force vector applied to the two feet of the bipedal robot based on the predetermined quadratic programming model and the moment vector, and obtaining a target moment vector for each ankle joint of the bipedal robot based on the general acceleration vector, the external force vector, and a predetermined inverse dynamics model.
[0014] According to a second aspect of an embodiment of the present disclosure, there is provided a control device for a bipedal robot, the device including: a support period acquisition module for acquiring a support period of the bipedal robot in a current state; a current velocity acquisition module for acquiring a current forward velocity and a current lateral velocity of the center of gravity of the bipedal robot at the end of the support period; a moment vector acquisition module for acquiring a moment vector of an ankle joint of a supporting foot of the bipedal robot based on the current forward velocity and the current lateral velocity; a target moment acquisition module for acquiring a target moment vector based on the moment vector and a predetermined quadratic programming model; and a robot control module for controlling motion of the bipedal robot based on the target moment vector.
[0015] Optionally, the support phase acquisition module includes a time acquisition submodule for acquiring a current time, a normal component force acquisition submodule for acquiring a normal component force of the ground that the swinging foot of the biped robot receives, and a condition determination submodule for determining whether the current time satisfies a first condition or a second condition and determining whether the current time and the normal component force satisfy a third condition, wherein the first condition refers to the current time exceeding the sum of a start time of a single leg support phase and a duration of the single leg support phase, and the second condition refers to the current time exceeding the sum of a start time of a biped support phase and a duration of the biped support phase. a condition judgment submodule for determining that the biped robot is in a bipedal supporting phase if the current time satisfies the first condition or the current time and the normal component force satisfy the third condition, and for determining that the bipedal robot is in a single leg supporting phase if the current time satisfies the first condition or the current time and the normal component force satisfy the third condition, and for determining that the bipedal robot is in a single leg supporting phase if the current time satisfies the second condition.
[0016] Optionally, the support period includes a single-leg support period, the forward velocity includes a first predicted forward velocity, and the lateral velocity includes a first predicted lateral velocity, and the current velocity acquisition module includes a position and velocity acquisition submodule for acquiring a first forward position and a first forward velocity and a first lateral position and a first lateral velocity of a center of gravity of the biped robot relative to the biped robot, a first forward velocity acquisition submodule for estimating a first predicted forward velocity of the biped robot based on a first preset forward velocity model, the first forward position, and the first forward velocity, and a first lateral velocity acquisition submodule for estimating a first predicted lateral velocity of the biped robot based on a first preset lateral velocity model, the first lateral position, and the first lateral velocity.
[0017] Optionally, the moment vector acquisition module includes a current step speed acquisition submodule for acquiring the forward velocity of the current step and the lateral velocity of the current step of the center of gravity at the end of a single-leg support phase in the current state, a next step speed acquisition submodule for acquiring the forward velocity of the next step and the lateral velocity of the next step of the center of gravity in the current state, and a moment vector acquisition submodule for acquiring a moment vector of an ankle joint of a supporting foot of the bipedal robot based on the forward velocity of the current step, the lateral velocity of the current step, the forward velocity of the next step, and the lateral velocity of the next step.
[0018] Optionally, the device further includes a swing foot displacement acquisition module for acquiring a forward displacement of a current step and a lateral displacement of a current step of a swing foot of the supporting foot relative to the center of gravity at the end of the one-leg support period, and a next step speed acquisition module for acquiring a forward speed of a next step and a lateral speed of a next step of the swing foot based on the forward displacement of the current step of the swing foot, the forward speed of the current step of the center of gravity, and the first predicted forward speed.
[0019] Optionally, the support period includes a bipedal support period, the forward velocity includes a second predicted forward velocity, and the lateral velocity includes a second predicted lateral velocity, and the current velocity acquisition module includes a position and velocity acquisition submodule for acquiring a second forward position and a second forward velocity and a second lateral position and a second lateral velocity of a center of gravity of the bipedal robot relative to the bipedal robot, a second forward velocity acquisition submodule for estimating a second predicted forward velocity of the bipedal robot based on a second preset forward velocity model, the second forward position, and the second forward velocity, and a second lateral velocity acquisition submodule for estimating a second predicted lateral velocity of the bipedal robot based on a second preset lateral velocity model, the second lateral position, and the second lateral velocity.
[0020] Optionally, the moment vector acquisition module includes a current step speed acquisition submodule for acquiring a forward velocity of a current step and a lateral velocity of a current step of the center of gravity at the end of a bipedal support period in a current state, a first moment vector acquisition submodule for acquiring a moment vector of a first supporting foot based on the forward velocity of the current step and the second predicted forward velocity, and a second moment vector acquisition submodule for acquiring a moment vector of a second supporting foot based on the lateral velocity of the current step and the second predicted lateral velocity, wherein the first supporting foot refers to the supporting foot in a single supporting period prior to the bipedal support period and the second supporting foot refers to a swinging foot in a single supporting period prior to the bipedal support period.
[0021] Optionally, the desired moment acquisition module includes a vector acquisition sub-module for acquiring a general acceleration vector of the bipedal robot and an external force vector applied to the two feet of the bipedal robot based on the preset quadratic programming model and the moment vector, and a desired moment acquisition sub-module for acquiring a desired moment vector of each ankle joint of the bipedal robot based on the general acceleration vector, the external force vector and a preset inverse dynamics model.
[0022] According to a third aspect of an embodiment of the present disclosure, a bipedal robot is provided, comprising a memory and a processor, the memory being used to store a computer program executable by the processor, and the processor being used to execute the computer program in the memory to realize the above method.
[0023] According to a fourth aspect of an embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, and an executable computer program in the storage medium can be executed by a processor to realize the above method.
[0024] The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects. In the solution provided by the embodiment of the present disclosure, the supporting period of the biped robot in the current state is obtained, and the current forward velocity and the current lateral velocity of the center of gravity of the biped robot at the end of the supporting period are obtained, and then the moment vector of the ankle joint of the supporting foot of the biped robot is obtained based on the current forward velocity and the current lateral velocity, and a target moment vector is obtained based on the moment vector and a preset quadratic programming model, and finally, the motion of the biped robot can be controlled based on the target moment vector. In this way, in this embodiment, the forward motion and lateral motion of the center of gravity of the biped robot are controlled by the moment vector of the ankle joint, thereby achieving the effect of controlling the biped, and in this embodiment, the whole body motion of the biped robot can be controlled by quadratic programming, so that the biped robot (fully driven and under driven) is all in a fully driven state, which is robust.
[0025] It should be noted that the above general description and the following detailed description are merely illustrative and explanatory and do not limit the present disclosure. [Brief description of the drawings]
[0026] The drawings herein are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification serve to explain the principles of the present disclosure. [Figure 1] 1 is a flowchart of a method for controlling a biped robot according to an exemplary embodiment. [Diagram 2] FIG. 2 is a schematic diagram of a single leg support phase and a double leg support phase as shown in an exemplary embodiment. [Diagram 3] 11 is a flowchart for acquiring and determining that a biped robot according to an exemplary embodiment is in a single-leg support phase; [Figure 4] 10 is a flow chart of obtaining a first predicted forward velocity and a first predicted lateral velocity shown in one exemplary embodiment. [Diagram 5] FIG. 13 is a schematic diagram of the height between the center of gravity and the contact surface illustrated by an illustrative embodiment; [Figure 6] 11 is a flowchart of obtaining a second predicted forward velocity and a second predicted lateral velocity shown in an exemplary embodiment. [Figure 7] 11 is a flowchart illustrating a process for acquiring a moment vector of an ankle joint during a single support phase according to an exemplary embodiment. [Figure 8] 11 is a flow chart of obtaining a next step forward velocity and a next step lateral velocity shown by an exemplary embodiment; [Figure 9] 11 is a flowchart of obtaining a moment vector of a supporting foot shown in accordance with an illustrative embodiment; [Figure 10] 11 is a flowchart for obtaining a target moment vector for each joint as shown in an illustrative embodiment; [Figure 11] 11 is a flowchart of another method for controlling a biped robot according to an illustrative embodiment. [Figure 12] FIG. 2 is a block diagram of a control device for a biped robot according to an illustrative embodiment; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0027] Exemplary embodiments will now be described in detail, examples of which are illustrated in the drawings. Where the following description refers to the drawings, the same numerals in different drawings represent the same or similar elements unless otherwise stated. The illustrative embodiments described below do not represent all embodiments consistent with the embodiments of the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of the present invention, as detailed in the appended claims. It should be noted that, where not in conflict, the features in the embodiments and embodiments described below can be combined with each other.
[0028] At present, mobile robots play an important role in our daily lives and provide various conveniences to people. Mobile robots are divided into wheeled, caterpillar and bipedal robots. Wheeled and caterpillar robots are greatly restricted in their use in rough, complex and diverse terrain environments, such as gravel ground and rough ground, which affect their passage. Bipedal robots can work only by touching the ground, and have lower requirements for the ground, so they have better prospects for use.
[0029] Take the biped robot as an example, it is a biped / pseudo-robot, based on its bionic structural characteristics, it can adapt to more complex terrains and working environments, and perform dangerous or repetitive tasks instead of humans. The walking and running of biped / pseudo-robots often have the characteristics of high nonlinearity, under-actuation, and mixed continuous and discrete systems, so that a suitable biped robot walking control method is particularly important.
[0030] In locomotion, whether the external contact environment is structured or unstructured, whether the terrain is ideal or uneven, the dynamics of a biped robot are inherently underactuated. This is because (1) a single support phase (SSP) is inevitably present in the gait of a legged locomotion, and (2) the supporting leg can only receive support force and friction force from the contact surface, i.e., the ground cannot exert a pulling force on the biped robot. Therefore, in the case of high-dynamic locomotion, during the single support phase, all the actuable degrees of freedom of the supporting leg of the biped robot cannot realize the complete control of the floating base (or center of gravity momentum).
[0031] In order to solve the above technical problems, the embodiments of the present disclosure provide a biped robot control method and device, a biped robot, and a readable storage medium, which can be applied to the biped robot. The biped robot can include zero, one, or two ankle joints. When each foot has two ankle joints, the foot can control both pitch and roll motions and belongs to a fully driven state, and when each foot has zero or one ankle joints, the foot cannot control both pitch and roll motions and belongs to an under driven state. For convenience of explanation, the biped robot in the embodiments described below includes robots in a fully driven state and an under driven state. The idea of the present disclosure is to control the forward velocity (movement) and lateral velocity (movement) of the center of gravity of the biped robot by selecting the moment vector of the ankle joint and the foot drop point. Here, the center of gravity refers to the center of mass of the biped robot.
[0032] Alternatively, the velocity of the center of gravity at the end of the single support phase (SSP) and the end of the double support phase (DSP) is tracked by controlling the moment of the ankle joint of the supporting foot, i.e., the position of the center of gravity relative to the supporting foot at the start of the SSP is indirectly controlled by selecting the foot drop point at the end of the SSP, thereby achieving the effect of predictive control.
[0033] And the trajectory tracking for other planned workspaces (such as the height of the center of gravity, the attitude angle of the floating base (such as the robot's body), the position and attitude of the swinging foot, etc.) is realized by the whole body motion control based on the quadratic programming, and the above center of gravity control can be an equality constraint in the quadratic programming model. In the embodiment of the present disclosure, the height and the attitude of the floating base during the single-leg support phase of the biped robot can be ensured in the fully driven state by the control method of first decomposing and integrating, which achieves the effect of complete control, has strong robustness, and can be adapted to the dynamic walking control of the biped robot in the fully driven and under driven states.
[0034] FIG. 1 is a control method for a biped robot according to an exemplary embodiment. Referring to FIG. 1, the control method for a biped robot includes steps 11 to 15.
[0035] In step 11, the support period of the biped robot in the current state is obtained.
[0036] In this embodiment, the processor of the biped robot can obtain the supporting period of the biped robot in the current state. The supporting period includes a single leg supporting period and a biped supporting period. With reference to FIG. 2, the biped robot includes a supporting leg (left leg in the left figure) that contacts the ground during the single leg supporting period and a swinging leg (right leg in the left figure) that leaves the ground, and the biped robot includes two supporting legs (left leg and right leg in the right figure) that contact the ground during the biped supporting period. Considering that the biped supporting period has been changed from the single leg supporting period, the supporting leg during the single leg supporting period, which is one of the supporting legs in the biped supporting period, is called the first supporting leg, and the swinging leg during the single leg supporting period, which is the other supporting leg in the biped supporting period, is called the second supporting leg.
[0037] In this embodiment, the step in which the processor determines the supporting period of the biped robot based on whether or not the swinging foot of the biped robot has fallen to the ground includes steps 31 to 33 with reference to FIG.
[0038] In step 31, the processor can obtain the current time and the normal force of the ground that is being experienced by the swinging foot of the biped robot.
[0039] In step 32, the processor can determine whether the current time satisfies a first condition or a second condition, and determine whether the current time and the normal component force satisfy a third condition. The first condition refers to the current time exceeding the sum of the start time and the duration of a single-leg support period, which is the end time of the single-leg support period, the second condition refers to the current time exceeding the sum of the start time and the duration of a biped support period, which is the end time of the biped support period, and the third condition refers to the current time exceeding the sum of the start time and the duration of a single-leg support period, which is the half of the duration of the single-leg support period, and the normal component force is greater than the product of a preset coefficient and the gravity of the biped robot. The sum of the start time and the half of the duration of the single-leg support period is the end time of the half of the single-leg support period.
[0040] In step 33, the processor may determine that the bipedal robot is in a bipedal support phase if the current time satisfies the first condition or the current time and the normal force satisfy the third condition, or may determine that the bipedal robot is in a single-leg support phase if the current time satisfies the second condition.
[0041] For example, let the current time be t and the start time of the one-leg support period be t. s0 The surface support time length during the one-leg support period is T s Then the first condition that the current time satisfies is t>t s0 +T s Here, t s0 +T s means that the start time t s0 Since it refers to the end time obtained by delaying Ts from the base point, t>t s0 +T s is the time when the current time is the end time (i.e. t s0 +T sIn the embodiment described later, when adding a time and a duration, this also means finding the end time.
[0042] For example, the normal force acting on the estimated swinging foot is f n where α is a coefficient greater than 0 and less than 0.5, m is the mass of the biped robot, and g is the gravitational acceleration, then the third condition that the current time and the normal force satisfy is (t>t s0 +0.5T s )I(f n >αmg), where mg represents the gravity of the biped robot.
[0043] For example, let the current time be t and the start time of the bipedal support period be t. d0 The surface bipedal support period time is T d Then the second condition that the current time satisfies is t>t d0 +T d It is determined that.
[0044] In step 12, the current forward velocity and the current lateral velocity of the center of gravity of the biped robot at the end of the support period are obtained.
[0045] In one embodiment, for example, the support period includes a one-leg support period, in which case the forward velocity includes a first predicted forward velocity and the lateral velocity includes a first predicted lateral velocity. Referring to FIG. 4, the step in which the processor acquires the current forward velocity and the current lateral velocity of the center of gravity of the bipedal robot at the end of the support period includes steps 41 to 42.
[0046] In step 41, the processor may obtain a first forward position and a first forward velocity and a first lateral position and a first lateral velocity of the centre of gravity of the biped robot relative to the biped robot.
[0047] In this step, when predicting the state of the center of gravity at the end of the one-legged support phase, the biped robot is considered as a Linear Inverted Pendulum Model (LIP). It is assumed that the forward position (hereinafter referred to as the first forward position), forward velocity (hereinafter referred to as the first forward velocity), lateral position (hereinafter referred to as the first lateral position), and lateral velocity (hereinafter referred to as the first lateral velocity) of the center of gravity of the biped robot with respect to the supporting leg are obtained through the forward motion capture and state estimation.
[0048] In step 42, the processor may estimate a first predicted forward velocity of the bipedal robot based on a first preset forward velocity model, the first forward position and the first forward velocity, and estimate a first predicted lateral velocity of the bipedal robot based on a first preset lateral velocity model, the first lateral position and the first lateral velocity.
[0049]
number
[0050] In one embodiment, for example, the support period includes a bipedal support period, in which case the forward velocity includes a second predicted forward velocity and the lateral velocity includes a second predicted lateral velocity. Referring to FIG. 6, the step in which the processor obtains the current forward velocity and the current lateral velocity of the center of gravity of the bipedal robot at the end of the support period includes steps 61 to 62.
[0051]
number
[0052] In equation (2), T d represents the surface bipedal support period duration, and t d0 represents the start time of the bipedal support period.
[0053] It should be noted that the second predicted forward velocity model and the second predicted lateral velocity model in step 62 are constructed based on a constant velocity model.
[0054] In step 13, a moment vector of an ankle joint of a supporting leg of the biped robot is obtained based on the current forward velocity and the current lateral velocity.
[0055] In one embodiment, taking a bipedal robot in a single-leg supporting phase as an example, the step of the processor obtaining a moment vector of the ankle joint of the supporting leg of the bipedal robot based on the current forward velocity and the current lateral velocity includes steps 71 to 72, with reference to FIG. 7 .
[0056]
number
[0057] In step 72, the processor can obtain the moment vector of the ankle joint of the supporting foot of the biped robot based on the forward velocity of the current step, the lateral velocity of the current step, the forward velocity of the next step, and the lateral velocity of the next step, as shown in the following equation (3):
[0058]
number
[0059] The moment of the ankle joint of the swinging leg during the single-leg support phase does not affect the control of the center of gravity velocity, but acts on the control of the posture of the swinging leg. When the ankle joint of a biped robot is in an under-actuated state, that is, when the degree of freedom that can be driven at the ankle joint is 0 or 1, J 1 ankle Since the corresponding term in is zero, equation (3) applies to the ankle joint in the fully actuated and underactuated states.
[0060] In one embodiment, the step of the processor correcting the plan for the swinging leg in the single leg support phase includes steps 81 to 82 with reference to FIG.
[0061] In step 81, the processor can obtain the forward displacement and lateral displacement of the current step of the swing foot of the supporting foot with respect to the center of gravity at the end of the one-leg support period. The forward displacement and lateral displacement of the current step with respect to the center of gravity of the swing foot at the end of the one-leg support period (i.e., the time when the swing foot falls to the ground) given by the upper layer plan are respectively foot x - ref and foot y - ref Assume that.
[0062] In step 82, the processor can obtain the forward velocity of the next step and the lateral velocity of the next step of the swinging foot based on the forward displacement of the current step of the swinging foot, the forward velocity of the current step of the center of gravity, and the first predicted forward velocity, as shown in the following equation (4):
[0063]
number
[0064] In another embodiment, taking a bipedal robot in a single-leg support phase as an example, the step of the processor obtaining a moment vector of the ankle joint of the supporting leg of the bipedal robot based on the current forward velocity and the current lateral velocity includes steps 91 to 92, with reference to FIG. 9 .
[0065]
number
[0066] In equation (5), τ 1 qankle and τ 2 qankle represent the moment vectors of the first and second supporting legs, respectively, and J 1 ankle and J. 2 ankle respectively represent the corresponding parts of the forward velocity and lateral velocity of the center of gravity and the ankle joint velocity of the first supporting leg or the second supporting leg in the broad-sense Jacobian matrix of the center of gravity velocity of a biped robot.
[0067] In step 14, a target moment vector is obtained based on the moment vector and a preset quadratic programming model.
[0068] In this embodiment, the step of the processor acquiring a target moment vector based on the moment vector and a preset quadratic programming model includes steps 101 to 102 with reference to FIG.
[0069] In step 101, the processor can obtain a general acceleration vector of the biped robot and an external force vector applied to the two feet of the biped robot based on the preset quadratic programming model and the moment vector. Here, the general acceleration vector includes the acceleration vector of each ankle joint of the biped robot and the acceleration vector of the floating base (e.g., the torso). In this step, the real-time whole-body motion control of the biped robot can be described as a quadratic programming (QP) model. The workspace reference trajectory vector of the biped robot (e.g., the height of the center of gravity, the attitude angle of the floating base, the position and attitude of the swinging foot, etc.) output by the upper-layer planning can be expressed as ξ ref Then, the state-estimated corresponding vector is ξ estIn order for a biped robot to track a reference trajectory while satisfying necessary physical constraints (e.g., moment saturation, contact friction cone constraints, etc.), a quadratic programming model is constructed, as shown in Equation (6) below.
[0070]
number
[0071] Based on the above quadratic programming model, the dynamics model of the biped robot is obtained and shown in Equation (7).
[0072]
number
[0073] In addition to the above constraint conditions, the single-leg support period also includes an equality constraint such as equation (3), and the two-leg support period also includes an equality constraint such as equation (5).
[0074]
number
[0075] In step 15, the motion of the biped robot is controlled based on the desired moment vector.
[0076] In this embodiment, the processor obtains the target moment vector of each ankle joint, and then transmits the target moment vector to each ankle joint so as to move the ankle joint, thereby achieving control of the biped robot.
[0077] Thus, in the solution provided by the embodiment of the present disclosure, the supporting period of the biped robot in the current state is obtained, and the current forward velocity and current lateral velocity of the center of gravity of the biped robot at the end of the supporting period are obtained, and then the moment vector of the ankle joint of the supporting foot of the biped robot is obtained based on the current forward velocity and the current lateral velocity, and a target moment vector is obtained based on the moment vector and a preset quadratic programming model, and finally, the motion of the biped robot can be controlled based on the target moment vector. In this way, the forward motion and lateral motion of the center of gravity of the biped robot are controlled by the moment vector of the midfoot joint in this embodiment, thereby achieving the effect of controlling the biped, and in this embodiment, the whole body motion of the biped robot can be controlled by quadratic programming, so that the biped robot (fully driven and under driven) is all in a fully driven state, which is robust.
[0078] Below, a control method for a bipedal robot provided by an embodiment of the present disclosure will be described in combination with FIG. 11, and with reference to FIG. 11, a processor can determine whether the bipedal robot is in a single-leg support phase.
[0079] When the biped robot is in a single-leg support phase, the forward velocity (i.e., first predicted forward velocity) and lateral velocity (i.e., first predicted lateral velocity) of the center of gravity at the end of the single-leg support phase are predicted, and then the moment vector of the ankle joint of the supporting leg is calculated, and the swing leg plan is corrected. When the biped robot is in a biped support phase, the forward velocity (i.e., second predicted forward velocity) and lateral velocity (i.e., second predicted lateral velocity) of the center of gravity at the end of the biped support phase are predicted, and then the moment vector of the ankle joint of the supporting leg is calculated.
[0080] The processor can obtain the general acceleration variables of the biped robot and the external force vectors acting on the two feet of the biped robot by solving a quadratic programming problem for whole body control based on the predicted forward velocity, the predicted lateral velocity and the quadratic programming model. Then, the processor can calculate the desired moment vectors of each ankle joint based on the inverse dynamics model and control the biped robot based on the desired moment vectors.
[0081] The biped robot control method provided by the embodiment of the present disclosure can be applied to biped robots in a fully driven state and an underdriven state, and controls the forward and lateral movements of the center of gravity of the robot by selecting the joint moment of the ankle joint and the foot drop point, i.e., the center of gravity velocity at the time of the end of SSP and the end of DSP is tracked by controlling the moment vector of the ankle joint of the supporting foot, and the position of the center of gravity relative to the supporting foot at the start time of SSP is indirectly controlled by selecting the foot drop point at the time of the end of SSP, thereby achieving a predictive control effect. In addition, in this embodiment, trajectory tracking for other planned workspaces (e.g., the height of the center of gravity, the attitude angle of the floating base (e.g., the robot's body), the position and attitude of the swinging foot, etc.) is realized by whole-body motion control based on quadratic programming, and this quadratic programming problem is an equality constraint for the center of gravity control. In the embodiment of the present disclosure, the height of the robot and the attitude of the floating base during the single-leg support period can be completely controlled when in a fully driven state by the control method of first decomposing and integrating, and the tracking control effect for the height of the center of gravity and the attitude of the floating base is significantly improved compared to other methods. At the same time, the control method disclosed herein utilizes predictive information and passive dynamics characteristics, and is highly robust and well adapted to the dynamic walking control of fully actuated and under actuated bipedal robots.
[0082] Based on the control method for a bipedal robot provided by an embodiment of the present disclosure, an embodiment of the present disclosure further provides a control device for a bipedal robot, and referring to FIG. 12 , the device includes: a support period acquisition module 121 for acquiring a support period of the bipedal robot in a current state; a current velocity acquisition module 122 for acquiring a current forward velocity and a current lateral velocity of the center of gravity of the bipedal robot at the end of the support period; a moment vector acquisition module 123 for acquiring a moment vector of an ankle joint of a supporting foot of the bipedal robot based on the current forward velocity and the current lateral velocity; a target moment acquisition module 124 for acquiring a target moment vector based on the moment vector and a predetermined quadratic programming model; and a robot control module 125 for controlling the motion of the bipedal robot based on the target moment vector.
[0083] In one embodiment, the supporting phase acquisition module includes a time acquisition submodule for acquiring a current time, a normal component force acquisition submodule for acquiring a normal component force of the ground that the swinging foot of the biped robot receives, and a condition determination submodule for determining whether the current time satisfies a first condition or a second condition and for determining whether the current time and the normal component force satisfy a third condition, wherein the first condition refers to the current time exceeding the sum of a start time of a single leg supporting phase and a duration of the single leg supporting phase, and the second condition refers to the current time exceeding the sum of a start time of a biped supporting phase and a duration of the biped supporting phase. a condition judgment sub-module for determining that the biped robot is in a bipedal support phase if the current time satisfies the first condition or the current time and the normal component force satisfy the third condition, and for determining that the bipedal robot is in a single leg support phase if the current time satisfies the first condition or the current time and the normal component force satisfy the third condition, and for determining that the bipedal robot is in a single leg support phase if the current time satisfies the second condition.
[0084] In one embodiment, the support period includes a single-leg support period, the forward velocity includes a first predicted forward velocity, and the lateral velocity includes a first predicted lateral velocity, and the current velocity acquisition module includes a position and velocity acquisition submodule for acquiring a first forward position and a first forward velocity and a first lateral position and a first lateral velocity of a center of gravity of the biped robot relative to the biped robot, a first forward velocity acquisition submodule for estimating a first predicted forward velocity of the biped robot based on a first preset forward velocity model, the first forward position, and the first forward velocity, and a first lateral velocity acquisition submodule for estimating a first predicted lateral velocity of the biped robot based on a first preset lateral velocity model, the first lateral position, and the first lateral velocity.
[0085] In one embodiment, the moment vector acquisition module includes a current step speed acquisition submodule for acquiring the forward velocity of the current step and the lateral velocity of the current step of the center of gravity at the end of a single-leg support phase in the current state, a next step speed acquisition submodule for acquiring the forward velocity of the next step and the lateral velocity of the next step of the center of gravity in the current state, and a moment vector acquisition submodule for acquiring a moment vector of an ankle joint of the supporting foot of the bipedal robot based on the forward velocity of the current step, the lateral velocity of the current step, the forward velocity of the next step, and the lateral velocity of the next step.
[0086] In one embodiment, the device further includes a swing foot displacement acquisition module for acquiring a forward displacement of a current step and a lateral displacement of a current step of a swing foot of the supporting foot relative to the center of gravity at the end of the one-leg support period, and a next step speed acquisition module for acquiring a forward speed of a next step and a lateral speed of a next step of the swing foot based on the forward displacement of the current step of the swing foot, the forward speed of the current step of the center of gravity, and the first predicted forward speed.
[0087] In one embodiment, the support period includes a bipedal support period, the forward velocity includes a second predicted forward velocity, and the lateral velocity includes a second predicted lateral velocity, and the current velocity acquisition module includes a position and velocity acquisition submodule for acquiring a second forward position and a second forward velocity and a second lateral position and a second lateral velocity of a center of gravity of the bipedal robot relative to the bipedal robot, a second forward velocity acquisition submodule for estimating a second predicted forward velocity of the bipedal robot based on a second preset forward velocity model, the second forward position, and the second forward velocity, and a second lateral velocity acquisition submodule for estimating a second predicted lateral velocity of the bipedal robot based on a second preset lateral velocity model, the second lateral position, and the second lateral velocity.
[0088] In one embodiment, the moment vector acquisition module includes a current step speed acquisition submodule for acquiring the forward velocity of the current step and the lateral velocity of the current step of the center of gravity at the end of a bipedal support period in a current state; a first moment vector acquisition submodule for acquiring a moment vector of a first supporting foot based on the forward velocity of the current step and the second predicted forward velocity; and a second moment vector acquisition submodule for acquiring a moment vector of a second supporting foot based on the lateral velocity of the current step and the second predicted lateral velocity, wherein the first supporting foot refers to the supporting foot in the single supporting period prior to the bipedal support period and the second supporting foot refers to the swinging foot in the single supporting period prior to the bipedal support period.
[0089] In one embodiment, the desired moment acquisition module includes a vector acquisition sub-module for acquiring a general acceleration vector of the bipedal robot and an external force vector applied to the two feet of the bipedal robot based on the preset quadratic programming model and the moment vector, and a desired moment acquisition sub-module for acquiring a desired moment vector of each ankle joint of the bipedal robot based on the general acceleration vector, the external force vector, and a preset inverse dynamics model.
[0090] In addition, the device embodiment shown in this embodiment is consistent with the contents of the method embodiment described above, and the contents of the method embodiment described above can be referenced, so the description will be omitted here.
[0091] In an exemplary embodiment, a bipedal robot is further provided, comprising a memory and a processor, the memory being adapted to store a computer program executable by the processor, and the processor being adapted to execute the computer program in the memory to implement the above method.
[0092] In an exemplary embodiment, a non-transitory computer readable storage medium, such as a memory containing instructions, said executable computer program is executable by the processor, is further provided, where the readable storage medium may be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.
[0093] Those skilled in the art can easily conceive of other embodiments of the present disclosure after considering the specification and practicing the technical solutions disclosed herein. The present disclosure is intended to cover any modifications, applications or adaptations, which modifications, applications or adaptations follow the general principles of the present disclosure and include well-known or commonly used technical means in the art that are not disclosed in the present disclosure. The specification and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the following claims.
[0094] It should be understood that the present disclosure is not limited to the exact construction described above and illustrated in the drawings, and various modifications and variations are possible without departing from the scope of the present disclosure, which is limited only by the appended claims.
Claims
1. A method for controlling a biped robot, comprising the steps of: obtaining a support period of the biped robot in a current state; obtaining a current forward velocity and a current lateral velocity of the center of gravity of the biped robot at the end of the support period; obtaining a moment vector of an ankle joint of a supporting foot of the biped robot based on the current forward velocity and the current lateral velocity; obtaining a target moment vector based on the moment vector and a predetermined quadratic programming model; and controlling motion of the biped robot based on the desired moment vector; The support period includes a one-leg support period and a two-leg support period. A method for controlling a biped robot.
2. The step of obtaining a support period of the biped robot in a current state includes: Obtaining a current time and a ground normal force acting on a swing foot of the biped robot; a step of determining whether the current time satisfies a first condition or a second condition, and determining whether the current time and the normal component force satisfy a third condition, wherein the first condition indicates that the current time exceeds the sum of a start time of a single-leg support period and a duration of the single-leg support period, the second condition indicates that the current time exceeds the sum of a start time of a biped support period and a duration of the biped support period, and the third condition indicates that the current time exceeds the sum of a start time of a single-leg support period and half the duration of the single-leg support period, and the normal component force is greater than a product of a preset coefficient and the gravity of the biped robot; determining that the biped robot is in a biped support phase when the current time satisfies the first condition or the current time and the normal force satisfy the third condition, and determining that the biped robot is in a single leg support phase when the current time satisfies the second condition.
2. The method for controlling a biped robot according to claim 1 .
3. When the support period is the single-leg support period, the forward velocity includes a first predicted forward velocity and the lateral velocity includes a first predicted lateral velocity; The step of obtaining a current forward velocity and a current lateral velocity of the center of gravity of the biped robot at the end of the support period includes: obtaining a first forward position and a first forward velocity and a first lateral position and a first lateral velocity of a center of gravity of the biped robot relative to the biped robot; estimating a first predicted forward velocity of the biped robot based on a first preset forward velocity model, the first forward position, and the first forward velocity, and estimating a first predicted lateral velocity of the biped robot based on a first preset lateral velocity model, the first lateral position, and the first lateral velocity.
2. The method for controlling a biped robot according to claim 1 .
4. if the support period is the bipedal support period, the forward velocity comprises a second predicted forward velocity and the lateral velocity comprises a second predicted lateral velocity; The step of obtaining a current forward velocity and a current lateral velocity of the center of gravity of the biped robot at the end of the support period includes: obtaining a second forward position and a second forward velocity and a second lateral position and a second lateral velocity of a center of gravity of the biped robot relative to the biped robot; estimating a second predicted forward velocity of the biped robot based on a second preset forward velocity model, the second forward position, and the second forward velocity, and estimating a second predicted lateral velocity of the biped robot based on a second preset lateral velocity model, the second lateral position, and the second lateral velocity.
2. The method for controlling a biped robot according to claim 1 .
5. The step of obtaining a moment vector of an ankle joint of a supporting foot of the biped robot based on the current forward velocity and the current lateral velocity includes: obtaining a forward velocity of the current step and a lateral velocity of the current step of the center of gravity at the end of a bipedal support phase in a current state; obtaining a moment vector of a first supporting foot based on a forward velocity of the current step and the second predicted forward velocity, and obtaining a moment vector of a second supporting foot based on a lateral velocity of the current step and the second predicted lateral velocity, wherein the first supporting foot refers to a supporting foot in a single supporting period prior to the bipedal supporting period, and the second supporting foot refers to a swinging foot in a single supporting period prior to the bipedal supporting period.
5. The method for controlling a biped robot according to claim 4.
6. The step of obtaining a target moment vector based on the moment vector and a predetermined quadratic programming model includes: acquiring a general acceleration vector of the biped robot and an external force vector acting on two feet of the biped robot based on the preset quadratic programming model and the moment vector; and obtaining a target moment vector for each ankle joint of the biped robot based on the general acceleration vector, the external force vector, and a preset inverse dynamics model.
2. The method for controlling a biped robot according to claim 1 .
7. A control device for a biped robot, comprising: A support period acquisition module for acquiring a support period of the biped robot in a current state; a current velocity acquisition module for acquiring a current forward velocity and a current lateral velocity of a center of gravity of the biped robot at the end of the support period; a moment vector acquisition module for acquiring a moment vector of an ankle joint of a supporting foot of the biped robot based on the current forward velocity and the current lateral velocity; a target moment obtaining module for obtaining a target moment vector based on the moment vector and a predetermined quadratic programming model; a robot control module for controlling motion of the biped robot based on the desired moment vector; The support period includes a one-leg support period and a two-leg support period. A control device for a biped robot.
8. The support period acquisition module includes: a time acquisition submodule for acquiring the current time; a normal component force acquisition sub-module for acquiring a normal component force of the ground that is applied to the swing foot of the biped robot; a condition determination submodule for determining whether the current time satisfies a first condition or a second condition, and for determining whether the current time and the normal component force satisfy a third condition, wherein the first condition indicates that the current time exceeds the sum of a start time of a single-leg support period and a duration of the single-leg support period, the second condition indicates that the current time exceeds the sum of a start time of a biped support period and a duration of the biped support period, and the third condition indicates that the current time exceeds the sum of a start time of a single-leg support period and half the duration of the single-leg support period and the normal component force is greater than a product of a preset coefficient and the gravity of the biped robot; a support phase determination submodule for determining that the biped robot is in a biped support phase if the current time satisfies the first condition or the current time and the normal force satisfy the third condition, and for determining that the biped robot is in a one-leg support phase if the current time satisfies the second condition. The control device for a biped robot according to claim 7 .
9. When the support period is the single-leg support period, the forward velocity includes a first predicted forward velocity and the lateral velocity includes a first predicted lateral velocity; The current speed acquisition module includes: a position and velocity acquisition sub-module for acquiring a first forward position and a first forward velocity and a first lateral position and a first lateral velocity of a center of gravity of the biped robot with respect to the biped robot; a first forward velocity acquisition sub-module for estimating a first predicted forward velocity of the biped robot based on a first preset forward velocity model, the first forward position, and the first forward velocity; a first lateral velocity acquisition sub-module for estimating a first predicted lateral velocity of the biped robot based on a first preset lateral velocity model, the first lateral position, and the first lateral velocity. The control device for a biped robot according to claim 7 .
10. if the support period is the bipedal support period, the forward velocity comprises a second predicted forward velocity and the lateral velocity comprises a second predicted lateral velocity; The current speed acquisition module includes: a position and velocity acquisition sub-module for acquiring a second forward position and a second forward velocity and a second lateral position and a second lateral velocity of a center of gravity of the biped robot with respect to the biped robot; a second forward velocity acquisition sub-module for estimating a second predicted forward velocity of the biped robot based on a second preset forward velocity model, the second forward position, and the second forward velocity; a second lateral velocity acquisition sub-module for estimating a second predicted lateral velocity of the biped robot based on a second preset lateral velocity model, the second lateral position, and the second lateral velocity. The control device for a biped robot according to claim 7 .
11. The moment vector acquisition module includes: a current step speed acquisition sub-module for acquiring a current step forward speed of the center of gravity and a current step lateral speed at the end of the biped support phase in the current state; a first moment vector acquisition sub-module for acquiring a moment vector of a first supporting foot based on the forward velocity of the current step and the second predicted forward velocity; a second moment vector acquisition sub-module for acquiring a moment vector of a second supporting foot based on the lateral velocity of the current step and the second predicted lateral velocity, the first supporting foot referring to a supporting foot in a single supporting period preceding the bipedal supporting period, and the second supporting foot referring to a swinging foot in a single supporting period preceding the bipedal supporting period; The control device for a biped robot according to claim 10 .
12. The target moment acquisition module includes: a vector acquisition submodule for acquiring a general acceleration vector of the biped robot and an external force vector acting on the two feet of the biped robot based on the preset quadratic programming model and the moment vector; and a target moment acquisition sub-module for acquiring a target moment vector of each ankle joint of the biped robot based on the general acceleration vector, the external force vector, and a preset inverse dynamics model. The control device for a biped robot according to claim 7 .
13. A bipedal robot, A memory and a processor are included, the memory is used to store a computer program executable by the processor; The processor is used to execute a computer program in the memory so as to realize the biped robot control method according to any one of claims 1 to 6. A bipedal robot characterized by
14. A non-transitory computer-readable storage medium, comprising: When the executable computer program in the storage medium is executed by a processor, the biped robot control method according to any one of claims 1 to 6 can be realized. A non-transitory computer-readable storage medium comprising:
Citation Information
Patent Citations
Spring load inverted pendulum model of biped robot and gait planning method
CN113050645A
Walk control device for leg type mobile robot
JP1994079657A
Control device of legged mobile robot
JP2011079092A
