A control method for dynamically optimizing contact force distribution of a legged robot

Abstract

The application relates to a control method for dynamically optimizing contact force distribution of a foot-type robot, and belongs to the technical field of foot-type robots. The method solves the problem that the foot end of a foot-type robot is prone to sinking when the foot-type robot travels on soft terrain. The method comprises the following steps: acquiring real-time ground reaction force of each foot of the foot-type robot in a current gait cycle supporting phase and joint control information in a previous gait cycle supporting phase; obtaining an expected ground reaction force time curve of each foot in the current gait cycle supporting phase based on the joint control information of each foot in the previous gait cycle supporting phase and a standard ground reaction force time curve; and obtaining control torque of each foot at each time by using a PID controller based on the expected ground reaction force time curve of each foot in the current gait cycle supporting phase and corresponding real-time ground reaction force, and applying the control torque to each joint motor of each foot to optimize the contact force distribution of each foot. The method realizes walking stability of the foot-type robot on unstructured terrain.

Description

Technical Field

[0001] This invention relates to the field of legged robot technology, and in particular to a control method for dynamically optimizing the contact force distribution of legged robots. Background Technology

[0002] Legged robots are a type of biomimetic robot that possesses flexible joint control capabilities by mimicking the movement mechanisms of the feet in biology. They can move efficiently in unstructured, complex terrain environments and have found wide applications in engineering and automation. For example, in geological exploration and environmental monitoring, legged robots can traverse areas inaccessible to traditional wheeled or tracked robots, enabling data collection and sample analysis of complex terrains. In disaster relief scenarios, legged robots can not only quickly reach damaged buildings or dangerous areas to perform search and rescue missions, but also carry sensors for real-time data collection, providing immediate information support for rescue decisions. In reconnaissance and special operations, legged robots achieve stable movement in unstructured terrain through discrete foot contact, traversing areas that are difficult or impossible for humans to pass safely, greatly enhancing intelligence gathering capabilities.

[0003] Currently, legged robots can walk on flat grass, gravel, and sand, and can even climb stairs and cross thresholds inside buildings. However, the dynamic interaction between the legged locomotion system and the ground determines that its motion stability is highly correlated with the ground's mechanical properties. Traditional control methods for legged robots are mostly based on the assumption of a rigid surface, using fixed friction coefficients and stiffness parameters to calculate contact forces, without considering the rheological properties of soft media such as sand and mud. When moving on muddy or heterogeneous surfaces, traditional control methods often struggle to avoid sinking due to excessive instantaneous impact forces. Furthermore, existing control strategies fail to establish dynamic force distribution optimization mechanisms for this phenomenon, significantly increasing the risk of sinking. This deficiency severely restricts the application effectiveness of legged robots in complex environments. Summary of the Invention

[0004] Based on the above analysis, the present invention aims to provide a control method for dynamically optimizing the contact force distribution of legged robots, in order to solve the problem that the feet of existing legged robots tend to sink when traveling on soft terrain.

[0005] The objective of this invention is mainly achieved through the following technical solutions:

[0006] This invention provides a control method for dynamically optimizing the contact force distribution of a legged robot, comprising the following steps:

[0007] The system acquires the real-time ground reaction force of each foot of the leg in the current gait cycle support phase and the joint control information of the previous gait cycle support phase; wherein, the joint control information is the control torque of each joint at each moment in the previous gait cycle support phase.

[0008] Based on the joint control information and standard ground reaction force time curve of each foot in the previous gait cycle support phase, the expected ground reaction force time curve of each foot in the current gait cycle support phase is obtained.

[0009] Based on the expected ground reaction force time curve of each foot in the current gait cycle support phase and the corresponding real-time ground reaction force, a PID controller is used to obtain the control torque of each foot at each moment and apply it to the joint motors of each foot to optimize the contact force distribution of each foot.

[0010] Furthermore, the expected ground reaction force time curve is a curve showing the relationship between each moment within the gait cycle support phase and the corresponding expected ground reaction force value; wherein, the curve control points, in chronological order, include the impact peak moment and the corresponding impact peak force value, the buffer moment and the corresponding buffer force value, and the support peak moment and the corresponding support peak force value.

[0011] Furthermore, obtaining the expected ground reaction force time curve for each foot in the current gait cycle support phase includes:

[0012] Based on the joint control information of each foot in the previous step-state periodic support phase, the maximum joint control torque τ of each foot in the previous step-state periodic support phase is obtained. max,n and cumulative joint control torque τ sum,n ;

[0013] Based on the maximum joint control torque τ of each foot in the previous step phase of the support cycle. max,n and cumulative joint control torque τ sum,n This yields the curve control points of each foot during the support phase of the current gait cycle;

[0014] Based on the ground contact time and ground lift time of each foot in the current gait cycle support phase, the expected ground reaction time curve of each foot in the current gait cycle support phase is obtained.

[0015] Furthermore, the cumulative joint control torque τ of each foot in the previous step-state periodic support phase is obtained using the following formula. sum,n :

[0016]

[0017] τ total,n (t)=τ hip,n (t)+τ thigh,n (t)+τcalf,n (t)

[0018] Where, τ hip,n (t) represents the control torque of the lateral swing joint of the nth foot at time t; τ thigh,n (t) represents the control torque of the hip joint of the nth foot at time t; τ calf,n (t) represents the control torque of the knee joint of the nth foot at time t; T represents the duration of the support phase of the gait cycle; Δt represents the sampling time interval of the joint sensor.

[0019] Furthermore, obtaining the curve control points of each foot in the support phase of the current gait cycle includes:

[0020] Based on the maximum joint control torque of the previous gait cycle support phase, the peak impact force value of the current gait cycle support phase is obtained using the following formula:

[0021] F impact,k,n =F impact,k-1,n *(0.95+0.05*(τ max,k-1,n / v ref ))

[0022] Among them, F impact,k,n τ represents the peak impact force of the nth foot during the support phase of the kth gait cycle; max,k-1,n τ represents the maximum control torque of the joint of the nth foot in the support phase of the (k-1)th gait cycle; ref This indicates the maximum control torque of the joint under standard gait.

[0023] Based on the accumulated joint control torque of the previous gait cycle support phase, the peak support force value of the current gait cycle support phase is obtained using the following formula:

[0024] F stance,k,n =F stance,k-1,n *(1+k2*τ sum,k-1,n )

[0025] Among them, F stance,k,n τ represents the peak support force of the nth foot in the support phase of the kth gait cycle; sum,k-1,n k1 represents the cumulative joint control torque of the nth foot in the support phase during the (k-1)th gait cycle; k2 represents the support peak force adjustment coefficient.

[0026] Based on the knee joint control torque at each moment of the previous gait cycle support phase, the buffer moment of the current gait cycle is obtained using the following formula:

[0027] t valley,k,n =t valley,k-1,n +k1*var calf,k-1,n

[0028] Among them, t valley,k,nThis represents the buffer moment of the nth foot in the support phase of the kth gait cycle; var calf,k-1,n This represents the variance of the knee joint control torque of the nth foot during the support phase of the (k-1)th gait cycle.

[0029] k1 represents the buffer timing adjustment coefficient;

[0030] The impact peak time, buffer force value corresponding to the buffer time, and support peak time of the standard ground reaction force curve are set to the impact peak time, buffer time, and support peak time of the support phase of the current gait cycle.

[0031] Furthermore, based on the ground contact time and ground lift time of each foot at the control point of the curve and the standard ground reaction force curve in the current gait cycle support phase, the expected ground reaction force time curve of each foot in the current gait cycle support phase is obtained by using cubic spline interpolation in chronological order.

[0032] Furthermore, the real-time ground reaction forces of each foot in the support phase of the current gait cycle of the legged robot are obtained, including:

[0033] Obtain the plantar pressure of each foot of the legged robot at the current moment;

[0034] Based on the acceleration of the legged robot at the current moment, the force distribution of each leg of the legged robot at the current moment is obtained;

[0035] The weighted average of the plantar pressure and the distributed force of each foot of the legged robot at the current moment is used to obtain the ground reaction force of each foot of the legged robot at the current moment.

[0036] Furthermore, obtaining the force distribution of each leg of the legged robot at the current moment includes:

[0037] Based on the acceleration of the legged robot at the current moment, the external force acting on the legged robot at the current moment is obtained;

[0038] The force distributed to each foot of the legged robot at the current moment is obtained by distributing the external force acting on the robot evenly to the ground-contacting foot.

[0039] Furthermore, based on the expected and actual ground reaction forces of each foot at each moment during the current gait cycle support phase, the control torque of each foot at each moment is obtained using the following formula:

[0040]

[0041] Where, τ n (t) represents the control torque on the nth foot at time t; F desire,n(t) represents the expected ground reaction force at time t for the nth foot; F real,n (t) represents the actual ground reaction force at time t for the nth foot; K p K i K d These represent the proportional gain, integral gain, and derivative gain of the PID controller, respectively.

[0042] Furthermore, the legged robot is equipped with an inertial navigation device to obtain the acceleration of the legged robot at the current moment.

[0043] The footed robot has a number of pressure sensors installed on the soles of each foot to obtain the sole pressure of each foot at the current moment.

[0044] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0045] 1. The present invention obtains the ground reaction force and joint control information of each foot of the legged robot in real time, and dynamically generates the desired ground reaction force time curve by combining the standard ground reaction force curve. The PID controller is used to precisely adjust the control torque of each joint motor and optimize the contact force distribution, which effectively avoids the sinking phenomenon of the foot end on soft ground and significantly enhances the walking stability of the legged robot in unstructured terrain.

[0046] 2. The present invention is based on the joint control information of the previous gait cycle support phase, and dynamically adjusts the impact peak force value, support peak force value and buffer time of the current gait cycle support phase, so that the distribution of ground reaction force is more reasonable. This not only reduces the damage of the initial impact force to the legged robot and the ground, but also ensures the smooth transition of forces throughout the gait cycle, thereby improving the motion efficiency and mechanical durability of the legged robot.

[0047] 3. The present invention combines data from an inertial navigation device and a plantar pressure sensor to perform a weighted average calculation of the ground reaction force, which can accurately sense the ground hardness characteristics, significantly improve the accuracy of force sensing, realize dynamic optimization of contact force distribution under different terrain conditions, and greatly expand the application range of legged robots in unstructured environments.

[0048] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained from what is particularly pointed out in the description and drawings. Attached Figure Description

[0049] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0050] Figure 1 This is a flowchart illustrating a control method for dynamically optimizing contact force distribution in a legged robot, as described in an embodiment of the present invention.

[0051] Figure 2 This is a schematic diagram of the legged robot in an embodiment of the present invention;

[0052] Figure 3 This is a schematic diagram of the standard ground reaction force curve for rigid ground in an embodiment of the present invention;

[0053] Figure 4 This is a schematic diagram of the standard ground reaction force curve for soft ground in an embodiment of the present invention;

[0054] Figure 5 This is a schematic diagram of PID control in an embodiment of the present invention. Detailed Implementation

[0055] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0056] A specific embodiment of the present invention discloses a control method for dynamically optimizing contact force distribution in a legged robot, such as... Figure 1 As shown, it includes the following steps S1-S3:

[0057] Step S1: Obtain the real-time ground reaction force of each leg of the legged robot in the current gait cycle support phase and the joint control information of the previous gait cycle support phase; wherein, the joint control information is the control torque of each joint at each moment in the previous gait cycle support phase.

[0058] Specifically, the legged robot is a robot that achieves movement through a biomimetic leg structure, mimicking the multi-jointed leg structures of humans and animals (such as dogs, horses, and insects) to achieve functions such as walking, running, and jumping. For example... Figure 2 The legged robot shown includes a body, which is the main part of the robot and carries all mechanical and electronic components; a leg structure, which is the walking component of the robot, usually including multiple joints, each of which is driven by a motor and can achieve multi-degree-of-freedom movement. In this embodiment, the leg structure includes three joints: a lateral swing joint, a hip joint, and a knee joint; and the feet, which are the parts of each leg that are in direct contact with the ground.

[0059] The gait cycle is the time from when the toe of one foot touches the ground to when the toe of that foot touches the ground again during the walking process of a legged robot, including the support phase and the swing phase; wherein, the support phase is the stage when one of the robot's feet contacts the ground and supports the body weight; the swing phase is the stage when one of the robot's feet leaves the ground and swings forward.

[0060] Furthermore, the real-time ground reaction forces of each foot in the support phase of the current gait cycle of the legged robot are obtained, including steps S11-S13:

[0061] Step S11: Obtain the plantar pressure of each foot of the legged robot at the current moment.

[0062] Specifically, the ground reaction force is the force exerted by the ground on the robot's feet. According to Newton's third law, the ground reaction force is equal in magnitude and opposite in direction to the force exerted by the robot's feet on the ground. Therefore, the ground reaction force directly reflects the interaction force between the legged robot and the ground. During walking, the distribution of the ground reaction force directly affects the robot's stability and motion efficiency.

[0063] Furthermore, each foot of the legged robot is equipped with a number of pressure sensors of the same quantity to obtain the foot pressure at the current moment.

[0064] Specifically, the same number of pressure sensors are installed on the soles of each foot of the legged robot to ensure that the pressure monitoring of each foot has the same accuracy and resolution. For example, the number of pressure sensors can be set to four, with each sensor covering a specific area of ​​the sole, achieving comprehensive monitoring of the pressure distribution across the entire sole. When the legged robot walks, pressure sensor data from each foot is collected in real time, with the data output by each sensor representing a local pressure value. Kalman filter technology is used to process the sensor data, reducing noise interference and improving estimation accuracy. For each foot, the pressure values ​​measured by all sensors at that foot are summed to obtain the total reaction force of that foot.

[0065] Step S12: Based on the acceleration of the legged robot at the current moment, obtain the force distribution of each leg of the legged robot at the current moment.

[0066] Furthermore, the legged robot is equipped with an inertial navigation device to obtain the acceleration of the legged robot at the current moment.

[0067] Specifically, during movement, the overall acceleration of the legged robot reflects its current motion state, and the total external force acting on the robot at that moment can be calculated from its acceleration. More specifically, the robot's acceleration can be measured in real time using an inertial navigation device (IMU) mounted on the robot body.

[0068] Furthermore, obtaining the force distribution of each leg of the legged robot at the current moment includes:

[0069] Based on the acceleration of the legged robot at the current moment, the external force acting on the legged robot at the current moment is obtained.

[0070] Specifically, according to Newton's second law, the total external force acting on the legged robot, neglecting other external forces (such as air resistance and external disturbance forces), can be calculated using the robot's total mass and its center of mass acceleration. Therefore, the external force acting on the legged robot at the current moment can be obtained using the following formula:

[0071] F total (t)=m·a(t)

[0072] Among them, F total (t) represents the external force acting on the legged robot at time t; m represents the mass of the legged robot; a(t) represents the acceleration of the legged robot at time t.

[0073] Furthermore, the external force experienced by the legged robot at the current moment is evenly distributed to the ground-contacting foot to obtain the force distribution of each foot of the legged robot at the current moment.

[0074] Specifically, during the movement of a multi-legged robot, the foot that touches the ground will bear a certain force. Therefore, when multiple feet of the robot are in contact with the ground at the same time, it is necessary to distribute the external force that the legged robot is subjected to at that moment evenly to each foot.

[0075] Step S13: Calculate the weighted average of the plantar pressure and the distributed force of each foot of the leg robot at the current moment to obtain the ground reaction force of each foot of the leg robot at the current moment.

[0076] Specifically, plantar pressure sensors directly measure the pressure distribution at the contact point between the foot and the ground. While this data provides local contact force information, it can be affected by factors such as sensor accuracy, noise interference, and uneven ground. Furthermore, plantar pressure sensors can only provide local information and cannot directly reflect the overall motion state of the robot. The force distribution of each foot is calculated based on the robot's overall acceleration and can reflect the robot's overall motion state, but it may not accurately reflect the actual ground reaction force distribution, especially in complex terrain where uncertainties in ground conditions can lead to discrepancies between theoretical and actual reaction forces. Therefore, by weighted averaging of plantar pressure and force distribution, local plantar pressure information can be combined with overall force distribution information to obtain a more comprehensive and accurate ground reaction force.

[0077] Step S2: Based on the joint control information and standard ground reaction force time curve of each foot in the previous gait cycle support phase, obtain the expected ground reaction force time curve of each foot in the current gait cycle support phase.

[0078] Furthermore, the expected ground reaction force time curve is a curve showing the relationship between each moment within the gait cycle support phase and the corresponding expected ground reaction force value; wherein, the curve control points, in chronological order, include the impact peak moment and the corresponding impact peak force value, the buffer moment and the corresponding buffer force value, and the support peak moment and the corresponding support peak force value.

[0079] Specifically, the ground reaction force-time curve is the curve showing the change of ground reaction force over time during the contact between the foot of the legged robot and the ground, reflecting the dynamic change of ground reaction force during this process. The standard ground reaction force-time curve is an ideal reference curve showing the change of ground reaction force over time when the foot of the legged robot contacts the ground under specific ground conditions. It is a standard curve obtained in advance under different ground conditions through experimental measurement or simulation.

[0080] More specifically, the curve control points typically correspond to specific moments and force values ​​on the ground reaction force curve. They are key feature points for controlling the shape of the ground reaction force time curve, determining the shape and dynamic characteristics of the curve. By adjusting the curve control points, the desired ground reaction force time curve can be dynamically generated, thereby optimizing the robot's gait control strategy.

[0081] When the robot's feet contact the rigid ground, such as Figure 3As shown, due to the high rigidity and low deformability of the ground, when the robot's foot first contacts the rigid ground, a large impact peak force value is generated. It then quickly drops to a trough to become a buffer moment, and then forms a second lower peak value due to the support effect, which is the support peak force value. The support peak force value is usually smaller than the impact peak force value, but it is sufficient to support the weight of the robot.

[0082] When the robot's feet touch the soft ground, such as Figure 4 As shown, due to the lower stiffness and higher deformability of soft ground, its reaction force time curve differs significantly from that of rigid ground. When the robot's foot first contacts the soft ground, the soft ground absorbs the impact, and its peak impact force is significantly lower than that of hard ground. Moreover, because ground compression takes time, the peak impact force appears later. At the same time, due to the continuous deformation of the ground, the reaction force does not decrease sharply. During the support phase, the ground reaction force gradually increases and reaches the peak support force. Because the soft ground needs more time to provide sufficient support force, this peak support force is usually greater than the peak impact force.

[0083] More specifically, when legged robots walk in complex terrain, the mechanical properties of soft ground are time-varying and lag-dependent. The foot interaction of the previous cycle affects the current ground response, and a fixed standard curve cannot adapt to this dynamic change. If completely relying on a fixed standard curve, the control error of the previous step cycle will accumulate, leading to instability in subsequent legged robot cycles. Therefore, by analyzing the joint torque, energy accumulation, and actual contact force data of the previous step cycle, the curve control points of the desired curve are corrected in real time. This ensures gait continuity while preventing the legged robot from sinking and losing energy.

[0084] Furthermore, obtaining the expected ground reaction force time curves for each foot in the current gait cycle support phase includes steps S21-S23:

[0085] Step S21: Based on the joint control information of each foot in the previous step-state periodic support phase, obtain the maximum joint control torque τ of each foot in the previous step-state periodic support phase. max,n and cumulative joint control torque τ sum,n .

[0086] Specifically, such as Figure 2 As shown in this embodiment, the leg structure of the legged robot includes three joints: a lateral swing joint, a hip joint, and a knee joint. Torque sensors mounted on the output shaft of each joint motor are used to obtain the joint control torque of each leg at each moment during the previous step-state periodic support phase, thereby obtaining the maximum control torque τ of each leg's joint. max,n , where n represents the nth leg of the legged robot.

[0087] Furthermore, the cumulative joint control torque τ of each foot in the previous step-state periodic support phase is obtained using the following formula. sum,n :

[0088]

[0089] τ total,n (t)=τ hip,n (t)+τ thigh,n (t)+τ calf,n (t)

[0090] Where, τ hip,n (t) represents the control torque of the lateral swing joint of the nth foot at time t; τ thigh,n (t) represents the control torque of the hip joint of the nth foot at time t; τ calf,n (t) represents the control torque of the knee joint of the nth foot at time t; T represents the duration of the support phase of the gait cycle; Δt represents the sampling time interval of the joint sensor.

[0091] Specifically, the cumulative joint control torque is calculated by time integral of the sum of the torques of the three joints using the trapezoidal rule. It reflects the work done by the joint motors in the support phase. By calculating the cumulative joint control torque, the energy consumption of each joint in the support phase can be quantified, thereby predicting the ground conditions that may be encountered in the current gait cycle. For example, if the cumulative torque of the support phase in the previous gait cycle increases significantly, it may indicate that the ground is relatively soft. The legged robot can adjust its control strategy in the support phase of the current gait cycle.

[0092] Step S22: Based on the maximum joint control torque τ of each foot in the previous step-state periodic support phase. max,n and cumulative joint control torque τ sum,n This yields the curve control points of each foot during the current gait cycle support phase.

[0093] Specifically, based on physical laws and prior knowledge, it can be known that the impact peak force value of each foot in the control point of the curve is positively correlated with the maximum torque of the joint, that is, the greater the joint output torque, the more violent the impact of the foot on the ground; the buffer time is positively correlated with the knee torque fluctuation, that is, the buffer time is the transition stage between the impact peak and the support peak. If the knee torque fluctuation is greater, it indicates that the ground is softer, which leads to a longer time for the reaction force to recover to the support peak; the support peak force value is positively correlated with the cumulative torque energy of the joint within the support phase, that is, the more energy is accumulated, the stronger the elastic recovery force generated after the ground is compressed, and the higher the support peak.

[0094] Furthermore, obtaining the curve control points of each foot in the support phase of the current gait cycle includes:

[0095] Based on the maximum joint control torque of the previous gait cycle support phase, the peak impact force value of the current gait cycle support phase is obtained using the following formula:

[0096] F impact,k,n =F impact,k-1,n *(0.95+0.05*(τ max,k-1,n / τ ref ))

[0097] Among them, F impact,k,n τ represents the peak impact force of the nth foot during the support phase of the kth gait cycle; max,k-1,n τ represents the maximum control torque of the joint of the nth foot in the support phase of the (k-1)th gait cycle; ref This represents the maximum control torque of the joint under standard gait, serving as the reference for the maximum joint torque when performing standard gait on an ideal rigid surface.

[0098] Specifically, the peak impact force value is the maximum value of the ground reaction force at the moment the foot touches the ground, reflecting the initial impact intensity of the foot on the ground. If the peak impact force value is too high, there is a risk of direct sinking on soft ground. Therefore, it is necessary to reduce the peak impact force value on soft ground to avoid instantaneous sinking of the foot.

[0099] Furthermore, based on the accumulated joint control torque of the previous gait cycle support phase, the peak support force value of the current gait cycle support phase is obtained using the following formula:

[0100] F stance,k,n =F stance,k-1,n *(1+k2*τ sum,k-1,n )

[0101] Among them, F stance,k,n τ represents the peak support force of the nth foot in the support phase of the kth gait cycle; sum,k-1,n k1 represents the cumulative joint control torque of the nth foot in the support phase of the (k-1)th gait cycle; k2 represents the support peak force adjustment coefficient, which can be 0.0002, indicating that for every 1000 N·s of joint torque energy accumulated, the support peak force needs to be increased by 2% of the baseline value.

[0102] Specifically, the peak support force value is the maximum value of the ground reaction force during the support phase, reflecting the ground's support capacity under continuous pressure from the foot. The larger the peak support force value, the more robust the support can be provided. Therefore, the support force is dynamically enhanced based on energy accumulation to prevent the foot from sinking.

[0103] Furthermore, based on the knee joint control torque at each moment of the support phase of the previous gait cycle, the buffer moment of the current gait cycle is obtained using the following formula:

[0104] t valley,k,n =t valley,k-1,n +k1*varcalf,k-1,n

[0105] Among them, t valley,k,n This represents the buffer moment of the nth foot in the support phase of the kth gait cycle; var calf,k-1,n represents the variance of the knee joint control torque of the nth foot in the support phase of the (k-1)th gait cycle; k1 represents the buffer moment adjustment coefficient, which, for example, can be 0.003, representing the knee joint torque fluctuation of 1 Nm. 2 The buffer time is delayed by 3 milliseconds.

[0106] Specifically, the buffering time is the time corresponding to the lowest point of the reaction force between the impact peak and the support peak, reflecting the transition time of the ground from impact to recovery. In order to buffer the impact force of the foot contacting the ground, the legged robot needs to absorb energy through the dynamic adjustment of the knee joint, which is manifested as the fluctuation of the knee joint torque. If the ground is soft or the robot needs a longer time to buffer the impact force, the fluctuation of the knee joint torque will be larger, and the buffering time will be delayed accordingly.

[0107] More specifically, the variance of the knee joint control torque is obtained by calculating the average value of the knee joint torque data and the average of the squares of the differences between each torque value and the average value, which reflects the degree of fluctuation of the knee joint torque.

[0108] Furthermore, the impact peak time, buffer force value corresponding to the buffer time, and support peak time of the standard ground reaction force curve are set to the impact peak time, buffer time, and support peak time of the current gait cycle support phase.

[0109] Specifically, experimental tests revealed that the impact peak time, buffer time, and support peak time corresponding to the control points of the ground reaction force time curve do not change significantly with the change of joint torque. Therefore, in this embodiment, the control point of the support phase of the current gait cycle is set to the corresponding value of the standard ground reaction force curve.

[0110] Step S23: Based on the ground contact time and ground lift time of each foot in the current gait cycle support phase, the expected ground reaction force time curve of each foot in the current gait cycle support phase is obtained.

[0111] Furthermore, based on the ground contact time and ground lift time of each foot at the control point of the curve and the standard ground reaction force curve in the current gait cycle support phase, the expected ground reaction force time curve of each foot in the current gait cycle support phase is obtained by using cubic spline interpolation in chronological order.

[0112] Specifically, the starting point (contact point) and reaction force of the desired reaction force time curve are both defined as 0; the ending point (departure point) is defined as the departure point of the standard ground reaction force curve, and the reaction force at the end point (departure point) is 0. The control points of the curve are ordered in the following time sequence to ensure they are ordered on the time axis: contact point < impact peak time < buffer time < support peak time < departure time.

[0113] More specifically, the time axis is divided into four intervals based on the timing of the control points of the curve, all control points are defined as interpolation nodes, and the spline coefficients of each interval are calculated according to the mathematical formula of cubic spline interpolation. Within each interval, a smooth curve is generated using the spline coefficients.

[0114] Step S3: Based on the expected ground reaction force time curve and the corresponding real-time ground reaction force of each foot in the current gait cycle support phase, the PID controller is used to obtain the control torque of each foot at each moment and apply it to the joint motors of each foot to optimize the contact force distribution of each foot.

[0115] Specifically, a PID controller is a feedback controller that dynamically adjusts the output signal by calculating the error between the target value and the actual value in real time, so that the system is stabilized in the desired state. The PID controller consists of three parts: proportional (P) – calculating the control torque directly based on the current error, allowing for rapid response to the current error; integral (I) – calculating the control torque based on the accumulation of error over time, eliminating historical error accumulation; and derivative (D) – calculating the control torque based on the rate of change of the error, predicting future error trends. The PID controller then adds the outputs of the proportional, integral, and derivative components to obtain the total control torque.

[0116] Furthermore, based on the expected and actual ground reaction forces of each foot at each moment during the current gait cycle support phase, the control torque of each foot at each moment is obtained using the following formula:

[0117]

[0118] Where, τ n (t) represents the control torque on the nth foot at time t; F desire,n (t) represents the expected ground reaction force at time t for the nth foot; F real,n (t) represents the actual ground reaction force at time t for the nth foot; K p K i K d These represent the proportional gain, integral gain, and derivative gain of the PID controller, respectively, which were obtained through prior experimental adjustments.

[0119] Specifically, such as Figure 5As shown, the error at each moment is obtained by subtracting the expected ground reaction force time curve from the actual ground reaction force in real time. This error is then input into a PID controller, which adjusts the output of the motor joint torque according to the adjusted parameters. This allows for timely correction of the joint torque output value based on the error, ensuring that the actual ground reaction force time curve follows the expected ground reaction force time curve. It should be noted that in this embodiment, since each leg of the legged robot includes three joints, the control torque τ for the nth leg at time t is... n (t) is a joint torque vector that includes lateral joint torque, hip joint torque, and knee joint torque.

[0120] More specifically, the control torque τ on the nth foot at each time point n (t) is applied to the corresponding joint motors of the corresponding foot to precisely adjust the movement of each joint to achieve stable control of the robot's gait.

[0121] In summary, the control method for dynamically optimizing contact force distribution in a legged robot according to embodiments of the present invention has the following beneficial effects:

[0122] 1. The present invention obtains the ground reaction force and joint control information of each foot of the legged robot in real time, and dynamically generates the desired ground reaction force time curve by combining the standard ground reaction force curve. The PID controller is used to precisely adjust the control torque of each joint motor and optimize the contact force distribution, which effectively avoids the sinking phenomenon of the foot end on soft ground and significantly enhances the walking stability of the legged robot in unstructured terrain.

[0123] 2. The present invention is based on the joint control information of the previous gait cycle support phase, and dynamically adjusts the impact peak force value, support peak force value and buffer time of the current gait cycle support phase, so that the distribution of ground reaction force is more reasonable. This not only reduces the damage of the initial impact force to the legged robot and the ground, but also ensures the smooth transition of forces throughout the gait cycle, thereby improving the motion efficiency and mechanical durability of the legged robot.

[0124] 3. The present invention combines data from an inertial navigation device and a plantar pressure sensor to perform a weighted average calculation of the ground reaction force, which can accurately sense the ground hardness characteristics, significantly improve the accuracy of force sensing, realize dynamic optimization of contact force distribution under different terrain conditions, and greatly expand the application range of legged robots in unstructured environments.

[0125] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A control method for dynamically optimizing contact force distribution in a legged robot, characterized in that, Includes the following steps: The system acquires the real-time ground reaction force of each foot of the leg in the current gait cycle support phase and the joint control information of the previous gait cycle support phase; wherein, the joint control information is the control torque of each joint at each moment in the previous gait cycle support phase. Based on the joint control information and standard ground reaction force time curve of each foot in the previous gait cycle support phase, the expected ground reaction force time curve of each foot in the current gait cycle support phase is obtained. Based on the expected ground reaction force time curve of each foot in the current gait cycle support phase and the corresponding real-time ground reaction force, the PID controller is used to obtain the control torque of each foot at each moment and apply it to the joint motors of each foot to optimize the contact force distribution of each foot. The expected ground reaction force time curve is the relationship curve between each moment in the gait cycle support phase and the corresponding expected ground reaction force value; wherein, the curve control points in time order include the impact peak moment and the corresponding impact peak force value, the buffer moment and the corresponding buffer force value, and the support peak moment and the corresponding support peak force value. The method of obtaining the expected ground reaction force time curves for each foot in the current gait cycle support phase includes: Based on the joint control information of each foot in the previous step-cycle support phase, the maximum joint control torque of each foot in the previous step-cycle support phase is obtained. and cumulative joint control torque ; Based on the maximum joint control torque of each foot in the previous step phase of the support cycle. and cumulative joint control torque This yields the curve control points of each foot during the support phase of the current gait cycle; Based on the ground contact time and ground takeoff time of each foot in the current gait cycle support phase, using the curve control point and the standard ground reaction force time curve, the expected ground reaction force time curve of each foot in the current gait cycle support phase is obtained.

2. The method according to claim 1, characterized in that, The cumulative joint control torque of each foot in the previous step-cycle support phase is obtained using the following formula. : in, This represents the control torque of the lateral swing joint of the nth foot at time t; This represents the control torque of the hip joint of the nth foot at time t; The torque represents the control torque of the knee joint of the nth foot at time t; T represents the duration of the support phase of the gait cycle. This indicates the sampling time interval of the joint sensor.

3. The method according to claim 2, characterized in that, The process of obtaining the curve control points of each foot in the current gait cycle support phase includes: Based on the maximum joint control torque of the previous gait cycle support phase, the peak impact force value of the current gait cycle support phase is obtained using the following formula: in, This represents the peak impact force value of the nth foot in the support phase during the kth gait cycle; This represents the maximum control torque of the joint of the nth foot in the support phase during the (k-1)th gait cycle; This indicates the maximum control torque of the joint under standard gait. Based on the accumulated joint control torque of the previous gait cycle support phase, the peak support force value of the current gait cycle support phase is obtained using the following formula: in, This represents the peak support force value of the nth foot in the support phase of the kth gait cycle; This represents the cumulative joint control torque of the nth foot during the support phase of the (k-1)th gait cycle; This indicates the adjustment coefficient for the peak support force. Based on the knee joint control torque at each moment of the previous gait cycle support phase, the buffer moment of the current gait cycle is obtained using the following formula: in, This represents the buffer moment of the nth foot in the support phase of the kth gait cycle; This represents the variance of the knee joint control torque of the nth foot during the support phase of the (k-1)th gait cycle. Indicates the buffer adjustment factor; The impact peak time, buffer force value corresponding to the buffer time, and support peak time of the standard ground reaction force time curve are set to the impact peak time, buffer time, and support peak time of the current gait cycle support phase.

4. The method according to any one of claims 1-3, characterized in that, Based on the ground contact time and ground lift time of each foot in the current gait cycle support phase, using the curve control point and the standard ground reaction force time curve in chronological order, the expected ground reaction force time curve of each foot in the current gait cycle support phase is obtained by cubic spline interpolation.

5. The method according to claim 1, characterized in that, Obtain the real-time ground reaction forces of each foot in the support phase of the current gait cycle of the legged robot, including: Obtain the plantar pressure of each foot of the legged robot at the current moment; Based on the acceleration of the legged robot at the current moment, the force distribution of each leg of the legged robot at the current moment is obtained; The weighted average of the plantar pressure and the distributed force of each foot of the legged robot at the current moment is used to obtain the ground reaction force of each foot of the legged robot at the current moment.

6. The method according to claim 5, characterized in that, The process of obtaining the force distribution of each leg of the legged robot at the current moment includes: Based on the acceleration of the legged robot at the current moment, the external force acting on the legged robot at the current moment is obtained; The force distributed to each foot of the legged robot at the current moment is obtained by distributing the external force acting on the robot evenly to the ground-contacting foot.

7. The method according to claim 6, characterized in that, Based on the expected and actual ground reaction forces of each foot at each moment during the current gait cycle support phase, the control torque of each foot at each moment is obtained using the following formula: in, This represents the control torque applied to the nth foot at time t; This represents the expected ground reaction force at time t for the nth foot. This represents the actual ground reaction force at time t for the nth foot. These represent the proportional gain, integral gain, and derivative gain of the PID controller, respectively.

8. The method according to any one of claims 5-7, characterized in that, The legged robot is equipped with an inertial navigation device to obtain the acceleration of the legged robot at the current moment. The footed robot has a number of pressure sensors installed on the soles of each foot to obtain the sole pressure of each foot at the current moment.

Images