A control method and control system for a biped heavy-duty robot

Abstract

This invention relates to the field of bipedal robot technology, and provides a control method and control system for a bipedal heavy-duty robot. The method includes: constructing a bipedal heavy-duty robot model; obtaining a dynamic gait frequency adjustment term based on terrain adaptation, energy optimization, and gait frequency coordination compensation terms, and obtaining a gait phase generation function based on the dynamic gait frequency adjustment term; conducting simulation experiments on the bipedal heavy-duty robot model based on the gait phase generation function; and controlling the bipedal heavy-duty robot based on the gait phase generation function if the simulation results are verified. This invention can obtain the gait phase generation function through the dynamic gait frequency adjustment term. Based on this gait phase generation function, it can better adapt to terrain changes or energy efficiency requirements to cope with complex scenarios such as slopes and soft ground, thus improving the dynamic adaptability of the gait frequency.

Description

Technical Field

[0001] This invention relates to the field of bipedal robot technology, and more specifically, to a control method and control system for a bipedal heavy-duty robot. Background Technology

[0002] In the design of modern bipedal heavy-duty robots, the application of computer simulation technology has become an indispensable key technical means. In the early stages of bipedal heavy-duty robot design, robot simulation technology allows for precise calculation and analysis of key parameters such as angles and joint torques. This simulation process provides important reference for selecting appropriate mechanical structure materials, motor models, and other key components. Furthermore, computer simulation technology possesses high repeatability and flexibility, allowing designers to conduct multiple simulation tests under different conditions, thereby comprehensively evaluating the performance of the bipedal heavy-duty robot. This approach effectively prevents interference problems that may occur during actual operation and also helps reduce wear and tear on motors and other drive components, extending the robot's lifespan. Therefore, computer simulation plays a crucial role in the design of bipedal heavy-duty robots, not only improving design efficiency and accuracy but also providing strong technical support for the optimization and improvement of bipedal heavy-duty robots.

[0003] To achieve natural and stable straight-line walking in a bipedal heavy-duty robot, a series of gait parameters can be designed by comprehensively considering factors such as robot size, gait smoothness, and energy efficiency. Based on forward kinematics calculation, the robot's walking posture is dynamically generated by calculating the angles and positions of each joint. Existing gait phase generation models generally use linear phase models with a fixed step frequency, which cannot adapt to changes in terrain or energy efficiency requirements. Summary of the Invention

[0004] To address the problem that existing bipedal heavy-duty robot gait phase generation models, which employ linear phase models with fixed step frequencies, cannot adapt to terrain changes or energy efficiency requirements, this invention provides a bipedal heavy-duty robot control method and control system. The specific technical solution is as follows:

[0005] A control method for a bipedal heavy-duty robot includes the following steps:

[0006] Construct a bipedal heavy-duty robot model;

[0007] Obtain terrain adaptation parameters to adjust cadence based on plantar pressure and ground slope to enhance gait stability;

[0008] Obtain an energy optimization term for adjusting the stride frequency based on the amplitude of torso height fluctuations to minimize torso kinetic energy fluctuations;

[0009] Obtain a gait coordination compensation term used to adjust the stride frequency based on the phase deviation of the left and right legs so that the phase deviation returns to the target value;

[0010] The dynamic gait frequency adjustment term is obtained based on the terrain adaptation term, energy optimization term, and gait frequency coordination compensation term, and the gait phase generation function is obtained based on the dynamic gait frequency adjustment term.

[0011] A simulation experiment of a bipedal heavy-duty robot model is conducted based on the gait phase generation function. If the simulation results are verified, the bipedal heavy-duty robot is controlled according to the gait phase generation function.

[0012] The bipedal heavy-duty robot control method constructs a bipedal heavy-duty robot model and obtains a dynamic gait frequency adjustment term by constructing a terrain adaptation term, an energy optimization term, and a gait frequency coordination compensation term. The gait phase generation function can be obtained through the dynamic gait frequency adjustment term. Based on this gait phase generation function, it can better adapt to terrain changes or energy efficiency requirements to cope with complex scenarios such as slopes and soft ground, thereby improving the dynamic adaptability of gait frequency.

[0013] Preferably, the method for obtaining terrain adaptation items includes:

[0014] Acquire heel pressure, toe pressure, maximum foot pressure, and ground slope;

[0015] Obtain the pressure difference between the toe pressure and the heel pressure, and obtain the pressure ratio between the pressure difference and the maximum foot pressure;

[0016] Obtain the cosine value of the ground slope, and obtain the terrain adaptation term based on the pressure ratio and the weighted sum of the cosine values.

[0017] Preferably, the method for obtaining the energy optimization term includes:

[0018] Obtain the vertical acceleration of the torso's center of mass, the power consumption of the joint motors, and the reference fluctuation threshold;

[0019] The standard deviation of the vertical acceleration is obtained based on a sliding window, and the fluctuation ratio between the standard deviation of the acceleration and the reference fluctuation threshold is obtained.

[0020] Obtain the sign function value of the power consumption of the joint motor, and obtain the energy optimization term based on the product of the fluctuation ratio value and the sign function value.

[0021] Preferably, the method for obtaining the step frequency coordination compensation term includes:

[0022] Obtain the proportional gain and integral gain, and obtain the proportional gain term based on the proportional gain and phase deviation;

[0023] The integral gain term is obtained based on the integral gain and phase deviation.

[0024] The step frequency coordination compensation term is obtained based on the proportional gain term and the integral gain term.

[0025] Preferably, the control method further includes the following steps:

[0026] Obtain the maximum duration of the transition time, and obtain the transition phase and stopping phase of the bipedal heavy-duty robot model based on the maximum duration of the transition time;

[0027] During the transition phase, the phase and amplitude scaling functions are obtained in an exponential decay form to achieve a smooth transition of the bipedal heavy-duty robot's gait.

[0028] During the stopping phase, the phase and amplitude scaling functions are obtained in a linear decay form to achieve stable stopping of the bipedal heavy-duty robot;

[0029] Specifically, the simulation experiment of the bipedal heavy-duty robot model based on the gait phase generation function is as follows: the simulation experiment of the bipedal heavy-duty robot model is carried out based on the gait phase generation function and the phase and amplitude scaling functions.

[0030] Preferably, during the transition phase, the method for obtaining the phase and amplitude scaling functions includes:

[0031] Obtain the timing duration from the start of the stop command, and obtain the oscillation suppression term for generating a half-cycle cosine wave based on the maximum transition time and the timing duration.

[0032] Obtain the oscillation attenuation coefficient, and based on the oscillation attenuation coefficient and the timing duration, obtain the exponential attenuation term used to suppress the oscillation amplitude by envelope.

[0033] The phase and amplitude scaling functions are obtained by multiplying the oscillation suppression term and the exponential decay term.

[0034] Preferably, during the stopping phase, the method for obtaining the phase and amplitude scaling functions includes:

[0035] Obtain the duration difference between the timing duration and the maximum transition time;

[0036] Obtain the linear convergence slope, and obtain the phase and amplitude scaling functions based on the time difference and the linear convergence slope.

[0037] A bipedal heavy-duty robot control system, used to implement the aforementioned bipedal heavy-duty robot control method, includes:

[0038] The simulation model building module is used to build models of bipedal heavy-duty robots.

[0039] The terrain adaptation module acquires terrain adaptation parameters that adjust gait frequency based on plantar pressure and ground slope to enhance gait stability.

[0040] The energy optimization term acquisition module is used to acquire energy optimization terms for adjusting the stride frequency based on the amplitude of torso height fluctuations in order to minimize torso kinetic energy fluctuations.

[0041] The gait coordination compensation term acquisition module is used to acquire gait coordination compensation terms for adjusting the stride frequency based on the phase deviation of the left and right legs so that the phase deviation returns to the target value.

[0042] The gait phase function acquisition module is used to acquire a dynamic gait frequency adjustment term based on the terrain adaptation term, energy optimization term, and gait frequency coordination compensation term, and to acquire a gait phase generation function based on the dynamic gait frequency adjustment term;

[0043] The simulation experiment and result verification module performs simulation experiments on the bipedal heavy-duty robot model based on the gait phase generation function. If the simulation results pass the verification, the bipedal heavy-duty robot is controlled according to the gait phase generation function.

[0044] Preferably, the control system further includes:

[0045] The phase acquisition module is used to acquire the maximum duration of the transition time, and to acquire the transition phase and the stopping phase of the bipedal heavy-duty robot model based on the maximum duration of the transition time.

[0046] The scaling function acquisition module is used to acquire the phase and amplitude scaling functions in an exponential decay form during the transition phase to achieve a smooth transition of the bipedal heavy-duty robot's gait, and to acquire the phase and amplitude scaling functions in a linear decay form during the stopping phase to achieve a stable stop of the bipedal heavy-duty robot.

[0047] Preferably, the scaling function acquisition module includes:

[0048] The oscillation suppression term acquisition unit is used to acquire the timing duration from the start of the trigger stop command, and to acquire the oscillation suppression term for generating a half-cycle cosine wave based on the maximum transition time and the timing duration.

[0049] An exponential decay term acquisition unit is used to acquire an oscillation decay coefficient and, based on the oscillation decay coefficient and timing duration, acquire an exponential decay term used to envelop and suppress the oscillation amplitude.

[0050] The scaling function acquisition unit is used to obtain the phase and amplitude scaling functions based on the product of the oscillation suppression term and the exponential decay term. Attached Figure Description

[0051] The invention will be further understood from the following description taken in conjunction with the accompanying drawings. The components in the drawings are not necessarily drawn to scale, but rather the emphasis is on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0052] Figure 1 This is a schematic diagram of the overall process of a bipedal heavy-duty robot control method according to an embodiment of the present invention;

[0053] Figure 2 This is a three-dimensional schematic diagram of a bipedal heavy-duty robot model according to an embodiment of the present invention. Figure 1 ;

[0054] Figure 3 This is a three-dimensional schematic diagram of a bipedal heavy-duty robot model according to an embodiment of the present invention. Figure 2 ;

[0055] Figure 4 This is a front view of the mechanical mechanism of a bipedal heavy-duty robot in one embodiment of the present invention;

[0056] Figure 5 This is a schematic diagram of the rear view of the mechanical mechanism of a bipedal heavy-duty robot in one embodiment of the present invention;

[0057] Figure 6 This is a flowchart illustrating a method for obtaining terrain adaptation items in one embodiment of the present invention;

[0058] Figure 7 This is a flowchart illustrating a method for obtaining energy optimization terms in one embodiment of the present invention;

[0059] Figure 8 This is a flowchart illustrating a method for obtaining step frequency coordination compensation terms in one embodiment of the present invention;

[0060] Figure 9 This is a diagram showing the hip angle of a humanoid straight line in a bipedal heavy-duty robot according to an embodiment of the present invention.

[0061] Figure 10 This is a diagram showing the joint angles of the humanoid straight lower limbs of a bipedal heavy-duty robot according to an embodiment of the present invention.

[0062] Figure 11 This is a diagram of the humanoid linear thigh joint angle of a bipedal heavy-duty robot according to an embodiment of the present invention;

[0063] Figure 12 This is a diagram showing the angle of the linear lower leg joint of a bipedal heavy-duty robot according to an embodiment of the present invention.

[0064] Figure 13 This is a diagram showing the angle of the linear foot joints of a bipedal heavy-duty robot according to an embodiment of the present invention.

[0065] Figure 14This is a linear gait diagram of a bipedal heavy-duty robot according to an embodiment of the present invention;

[0066] Figure 15 This is a diagram showing the hip angle of a humanoid circular trajectory of a bipedal heavy-duty robot according to an embodiment of the present invention.

[0067] Figure 16 This is a diagram of the lower limb joint angles of a bipedal heavy-duty robot with a humanoid circular trajectory, according to an embodiment of the present invention.

[0068] Figure 17 This is a diagram of the thigh joint angle of a humanoid circular trajectory bipedal heavy-duty robot according to an embodiment of the present invention.

[0069] Figure 18 This is a diagram showing the foot angles of a humanoid circular trajectory of a bipedal heavy-duty robot according to an embodiment of the present invention.

[0070] Figure 19 This is a gait diagram of a bipedal heavy-duty robot in a circular trajectory according to an embodiment of the present invention.

[0071] Explanation of reference numerals in the attached diagram: 1. Control center; 2. Right upper hip joint motor; 3. Right middle hip joint motor; 4. Right lower hip joint motor; 5. Right knee joint motor; 6. Right upper ankle joint motor; 7. Right lower ankle joint motor; 8. Left upper hip joint motor; 9. Left middle hip joint motor; 10. Left lower hip joint motor; 11. Left knee joint motor; 12. Left upper ankle joint motor; 13. Left lower ankle joint motor. Detailed Implementation

[0072] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to its embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of the invention.

[0073] The simulation algorithm is based on forward kinematics calculation, dynamically generating the robot's walking posture by calculating the angles and positions of each joint. Gait phase is the core of controlling the walking rhythm; alternating walking with the left and right legs is achieved through phase difference. Based on time and step frequency calculations, the formula is: Where step_freq represents the step frequency, typically set to 0.6 Hz, time(t) is the current time, and the left leg phase is directly expressed as phase. That is, π. The phase of the right leg is increased by π (i.e., phase + pi) to ensure that the left and right legs are always in opposite gait phases (e.g., the right leg supports the body while the left leg swings). To enhance readability, the phase calculation results are stored as phase_left and phase_right. This gait phase generation model uses a linear phase model with a fixed stride frequency, which has the problem of not being able to adapt to terrain changes or energy efficiency requirements.

[0074] Therefore, such as Figure 1 As shown, an embodiment of the present invention provides a control method for a bipedal heavy-duty robot, comprising the following steps:

[0075] S1, Construct a bipedal heavy-duty robot model.

[0076] Specifically, this bipedal heavy-duty robot model adopts a humanoid structure, comprising six main components: torso, hips, lower limb joints, thighs, calves, and feet. The dimensions of each component are designed with reference to typical bipedal heavy-duty robot parameters, in millimeters, to ensure the model's proportions closely resemble human gait characteristics while meeting practical engineering requirements. For example, the specific dimensions of the bipedal heavy-duty robot model are as follows: torso length: 120 mm, connecting the left and right legs and supporting the upper body; hip length: 131 mm, connecting the torso and lower limbs; lower limb joint length: 139 mm, allowing for lateral fine-tuning; thigh length: 259 mm, providing primary support; calf length: 360 mm, extending stride length; foot length: 190 mm, simulating heel-toe contact. In Matlab, the model is constructed based on a three-dimensional coordinate system, and each component is drawn using three-dimensional plotting functions. The torso is represented by a thick black line, while the hips, lower limb joints, thighs, calves, and feet of the left and right legs are distinguished by different colors (such as red, green, blue, purple, and cyan) to visually demonstrate the dynamic changes in the kinematic chain. To simulate realistic ground constraints, the simulation environment includes a planar mesh ground, covering an area of ​​±500 mm along the x-axis and ±200 mm along the y-axis, using a light gray semi-transparent material (0.3 transparency) to clearly show the ground position while avoiding obscuring the robot's movement trajectory.

[0077] Figure 2 , Figure 3 , Figure 4 as well as Figure 5The diagram shows the structural layout of the bipedal heavy-duty robot model. The detailed correspondences of each number in the diagram are as follows: 1 is the control center of the bipedal robot; 2 is the upper hip joint motor (right leg); 3 is the middle hip joint motor (right leg); 4 is the lower hip joint motor (right leg); 5 is the knee joint motor (right leg); 6 is the upper ankle joint motor (right leg); 7 is the lower ankle joint motor (right leg); 8 is the upper hip joint motor (left leg); 9 is the middle hip joint motor (left leg); 10 is the lower hip joint motor (left leg); 11 is the knee joint motor (left leg); 12 is the upper ankle joint motor (left leg); and 13 is the lower ankle joint motor (left leg). The location corresponding to number 1 is the control center, housing the core controller, core switching sensors, battery, power distribution board, etc., and is the core of the entire bipedal heavy-duty robot. Motors #2 and #8 are upper hip joint motors, model DM-J10010L-2EC; motors #3, #4, #9, and #10 are upper-middle hip joint motors; motors #5 and #11 are knee joint motors, all model DM-J8006-2EC; motors #6, #7, #12, and #13 are ankle joint motors, all model DM-J6006-2EC. There are a total of 12 joint motors. There is a 10-axis sensor on each thigh to monitor the movement status of the bipedal heavy-duty robot in real time. There is also an IMU-BMI088 6-axis sensor on the torso to monitor the actual movement of the torso. The other sensors are interactive sensors.

[0078] This bipedal heavy-duty robot's main body is the torso, with the hip joint, lower limb joint, thigh, calf, and foot connected to each side respectively. The components are connected via flanges and several linkages. Motors drive the connections between the hip and torso, the lower limb joint and hip, the thigh and lower limb joint, the calf and thigh, and the foot and calf. One motor drives the hip and torso to swing back and forth; a motor drives the lower limb and torso to swing left and right; a motor drives the thigh to rotate around the lower limb joint; a motor drives the calf to swing back and forth relative to the thigh (the motor and linkages form a planar four-bar linkage); and two motors drive the foot to swing back and forth and inward and outward relative to the calf (two motors simultaneously adjust the foot angle). Each leg of the bipedal heavy-duty robot has six motors, giving the robot's legs a wide range of degrees of freedom, allowing for flexible posture adjustments and a good mimicking of human gait.

[0079] S2, acquire terrain adaptation parameters for adjusting gait frequency based on plantar pressure and ground slope to enhance gait stability. As a preferred technical solution, such as... Figure 6 As shown, the methods for obtaining terrain adaptation parameters include:

[0080] S21, Get Heel Pressure Toe pressure Maximum foot pressure and ground slope Maximum foot pressure = robot weight × gravitational acceleration. Ground slope can be simulated using LiDAR or a depth camera. Heel pressure can be acquired using a piezoelectric thin-film sensor mounted on the heel, and toe pressure can be acquired using a force-sensitive resistor array mounted on the toe.

[0081] S22, obtain the pressure difference between the toe pressure and the heel pressure, and obtain the pressure ratio between the pressure difference and the maximum foot pressure. When the toe pressure is greater than the heel pressure, it can be understood as being on an uphill slope or soft ground. In this case, the cadence should be reduced to enhance stability. When the ground slope is greater than 5 degrees, the cadence can be reduced proportionally to the slope to prevent slipping.

[0082] S23, obtain the cosine value of the ground slope, and obtain the terrain adaptation term based on the pressure ratio and the weighted sum of the cosine values. Ground slope The physical meaning of the sine value is equivalent to the slope gradient. For example, sin(15°)≈0.26, which is a ground slope of 26%.

[0083] Specifically, when humans go uphill, they increase stride length and decrease stride frequency (to reduce the impact of slope); when going downhill, they control deceleration through heel pressure (to prevent slippage). This terrain adaptation feature integrates plantar pressure difference and slope angle to achieve a similar adaptive adjustment, which is used to simulate human walking strategies on slopes. The terrain adaptation feature can be expressed as follows: .in, These represent the pressure difference weight and the slope weight, respectively, ranging from 0.3 to 0.6 and from 0.4 to 0.8. These represent the pressure difference term and the slope term, respectively. In general, this terrain adaptation term has the following functions: 1. Dynamically adjust the robot's step frequency to cope with complex surfaces (such as slopes, sand, and gravel roads); 2. Predict the risk of slipping by the foot pressure difference and reduce the step frequency in advance to stabilize posture; 3. Intelligently balance the step frequency and torque output when walking on steep slopes to avoid motor overload.

[0084] S3, obtain the energy optimization term used to adjust the stride frequency according to the amplitude of torso height fluctuations in order to minimize torso kinetic energy fluctuations.

[0085] As a preferred technical solution, such as Figure 7 As shown, the methods for obtaining the energy optimization term include:

[0086] S31, obtain the vertical acceleration of the torso's center of mass and the power consumption of the joint motors. and reference fluctuation threshold Vertical acceleration can be simulated using an IMU simulator, reflecting the intensity of the vertical fluctuations in the center of gravity. The power consumption of the joint motors is equal to the sum of the products of the torques and angular velocities of multiple joints, directly reflecting real-time energy consumption. A larger reference fluctuation threshold, to some extent, indicates more severe torso swaying and higher energy consumption.

[0087] S32, Obtain the standard deviation of the vertical acceleration based on a sliding window. And obtain the fluctuation ratio between the acceleration standard deviation and the reference fluctuation threshold. The window length equals the one-step period, typically approximately 1.67 seconds. The reference fluctuation threshold is 0.5. .

[0088] S33, Obtain the sign function value of the power consumption of the joint motor. The energy optimization term is obtained by multiplying the fluctuation ratio value and the sign function value.

[0089] Specifically, the energy optimization term is mainly constructed by simulating the energy-optimal strategy of human walking. When walking, humans spontaneously adjust their stride frequency to reduce vertical center of gravity fluctuations and muscle power change rates. To achieve equilibrium and minimize metabolic energy consumption, the energy optimization term can be expressed as follows: .in, This represents the energy sensitivity coefficient, typically taken as 0.4. Experiments show that: Values ​​greater than 0.5 are prone to instability. A step frequency less than 0.3 indicates insufficient energy consumption optimization. If the standard deviation of acceleration is greater than the reference fluctuation threshold, it can be interpreted as excessive torso swaying, and the step frequency should be reduced to decrease energy consumption; if... A value greater than zero indicates that power consumption is rising too quickly, and the step frequency should be reduced accordingly.

[0090] Overall, this energy optimization feature has the following functions: 1. Suppressing drastic fluctuations in torso height to avoid additional energy consumption caused by instability in the center of gravity; 2. Preventing sudden changes in joint motor power to extend hardware lifespan; 3. Minimizing energy consumption during robot walking.

[0091] S4, obtain a gait coordination compensation term used to adjust the stride frequency based on the phase deviation of the left and right legs so that the phase deviation returns to the target value. As a preferred technical solution, such as... Figure 8 As shown, the method for obtaining the step frequency coordination compensation term includes:

[0092] S41, Obtain the proportional gain and integral gain The proportional gain term is obtained based on the proportional gain and phase deviation. Generally, the proportional gain determines the instantaneous compensation strength of the phase difference, while the integral gain is used to eliminate steady-state errors (such as long-standing small phase deviations). The proportional gain and integral gain can be set to 0.25 and 0.05, respectively. Phase deviation This represents the phase difference between the left and right legs. Ideally: left leg phase = π / 2, right leg phase = 3π / 2, phase deviation. =0. After interference, the left leg phase = π / 2, the right leg phase = π, and the phase deviation Δ is 0. =-π / 2, therefore a correction is needed.

[0093] S42, Obtain the integral gain term based on the integral gain and phase deviation. This integral gain term is mainly used to eliminate steady-state phase error.

[0094] S43, obtain the step frequency coordination compensation term based on the proportional gain term and the integral gain term.

[0095] Specifically, when humans walk, even if one leg is disturbed (such as stepping on a pebble), the nervous system quickly adjusts the gait cycle through the spinal cord's central pattern generator, resynchronizing the movements of the left and right legs. The gait frequency coordination compensation term achieves a similar function using a proportional-integral (PI) controller, primarily used to simulate the phase-locking mechanism of the human cerebellum. The gait frequency coordination compensation term can be expressed as... Among them, when When the value is greater than 0.2 rad, it can be compensated by step frequency coordination term. Adjust the step frequency to bring the phase difference between the left and right legs back to π.

[0096] In summary, this gait frequency coordination compensation item has the following functions: 1. Adjusting the gait phase difference between the left and right legs through closed-loop feedback to ensure precise synchronization of the alternating movements of the left and right legs when the robot walks; 2. Automatically restoring the ideal phase relationship when external disturbances (such as uneven ground or crosswinds) cause gait disorder; 3. Reducing the additional joint torque compensation caused by gait asymmetry and lowering overall energy consumption.

[0097] S5. Obtain a dynamic gait frequency adjustment term based on the terrain adaptation term, energy optimization term, and gait frequency coordination compensation term, and obtain a gait phase generation function based on the dynamic gait frequency adjustment term.

[0098] For example, dynamic cadence adjustment item It can be expressed as .in, The weight coefficients for the terrain adaptation, energy optimization, and gait frequency coordination compensation terms are represented, respectively. These coefficients can be dynamically adjusted through a learning algorithm, and are generally set to 0.5, 0.3, and 0.2, respectively. The dynamic gait frequency adjustment term can be limited to the range of -0.15 to 0.15, corresponding to a gait frequency adjustment range of ±15% (i.e., 0.51-0.69 Hz). This dynamic gait frequency adjustment term integrates pressure-slope joint feedback, dynamic adjustment of acceleration standard deviation, and closed-loop PI compensation, enabling real-time dynamic adjustment of gait frequency and improving the robot's gait adaptability to terrain, energy consumption, and stability.

[0099] Gait phase generation function It can be represented as .in, The base cadence is typically set at 0.6 Hz to maintain steady-state walking.

[0100] S6. Perform a simulation experiment on the bipedal heavy-duty robot model according to the gait phase generation function. If the simulation results pass the verification, control the bipedal heavy-duty robot according to the gait phase generation function.

[0101] In reality, the robot requires a gradual process to transition from a stationary start to stable walking. To avoid initial abrupt changes, a smooth transition is introduced in the simulation within the first second (100 frames, assuming transition_frames=100). The transition weights are calculated using a cosine function, with the amplitude gradually increasing from 0 to 1, as shown in the formula: During the transition phase, the phase and swing amplitude are scaled by `phase_scale`, for example, the left leg phase is adjusted to `phase_left * phase_scale` to ensure a smooth start. After the transition, `phase_scale` is fixed at 1, and the robot enters a stable walking state. The gait phase generation function is input into Matlab software to conduct linear and circular trajectory simulation experiments on the bipedal heavy-duty robot model, and the joint angles of the bipedal heavy-duty robot during humanoid linear walking are recorded.

[0102] like Figures 9-14 As shown, Figure 9 It shows the left and right hip angle curves with a clear alternation pattern; foot trajectory: during the swing phase, the average height of the foot off the ground is 29.5 mm, and the trajectory is a smooth parabola; during the support phase, the foot is completely in contact with the ground, and the transition between the front and rear ends is natural, simulating the biomechanical characteristics of the heel-toe. Figure 11 , Figure 12 as well as Figure 13 The joint angles of the bipedal heavy-duty robot during humanoid linear walking were recorded. Figure 14The simulation results demonstrate the robot's gait in a straight line. The simulation fully validates the effectiveness of the gait design, successfully achieving natural and stable straight-line walking, and verifying the rationality of the gait parameters and algorithm design. Specifically, the results are as follows: Stride length stability: The robot's average stride length is 198.7 mm, with a standard deviation of 1.2 mm and an error of less than 1.5%, indicating precise stride length control, highly consistent with the design value of 200 mm; Trunk dynamics: Trunk height fluctuations are controlled within ±15 mm, with a maximum deviation of 0.8 mm, and the peak tilt angle is π / 50 rad (approximately 3.6°), lower than the design value of π / 48 rad, reflecting excellent posture stability; Joint coordination: The hip angle varies sinusoidally over time with a period of 1.67 seconds (consistent with a step frequency of 0.6 Hz), and the maximum angular acceleration is 0.09 rad / s², indicating smooth movement without significant abrupt changes.

[0103] like Figure 15-19 As shown, Figure 16 It shows the change in knee joint angle over time, with a cycle of 1.67 seconds (step frequency of 0.6 Hz), and the alternation between the left and right legs is clear and regular. Figure 19 The circular trajectory gait diagram of the robot is shown. Simulation results verify the effectiveness of the circular trajectory gait design, specifically in trajectory accuracy. The average deviation between the torso center trajectory and the ideal circular path is 0.5 mm, and the maximum deviation is 1.7 mm, indicating highly accurate path tracking.

[0104] In terms of gait stability: the average stride length was 197.8 mm, with a standard deviation of 1.4 mm, close to the 198.7 mm in linear simulation, with an error of less than 1.5%. The hip angle amplitude was stable at π / 10 radians, with a maximum angular acceleration of 0.08 radians / second², reflecting smooth turning movements. Trunk dynamics: height fluctuations were controlled within ±15 mm, with a standard deviation of 0.6 mm, and the peak tilt angle was π / 52 radians (approximately 3.5 degrees), lower than the design value of π / 48 radians, demonstrating excellent posture control. Knee joint coordination: the average knee joint angle during the swing phase was 42 degrees, with a maximum of 132 degrees; the average knee angle during the stance phase was 2.8 degrees, with a maximum of 9 degrees, conforming to biomechanical constraints.

[0105] Regarding foot trajectory: during the swing phase, the average height of the foot off the ground is 28.0 mm, and the trajectory is arc-shaped, adapting to the curvature of the path; during the support phase, the foot is in contact with the ground, and the transition between the front and rear ends is smooth.

[0106] Compared to linear simulation, circular trajectory simulation significantly improves path adaptability by reducing hip amplitude (π / 10 vs π / 8 radians), introducing knee joint constraints (0-10 degree support phase), and dynamic tangent adjustment, thus verifying the robustness of the algorithm.

[0107] In summary, the bipedal heavy-duty robot control method constructs a bipedal heavy-duty robot model and obtains a dynamic gait frequency adjustment term by constructing a terrain adaptation term, an energy optimization term, and a gait frequency coordination compensation term. The gait phase generation function can be obtained through the dynamic gait frequency adjustment term. Based on this gait phase generation function, the robot can better adapt to terrain changes or energy efficiency requirements to cope with complex scenarios such as slopes and soft ground, thereby improving the dynamic adaptability of the gait frequency.

[0108] The aforementioned transition weight cosine function formula While initial abrupt changes can be avoided, a longer transition time may be present, as mechanical power consumption constraints are not considered. Therefore, to optimize the transition weight cosine function formula... As a preferred technical solution, the control method further includes the following steps:

[0109] S7, obtain the maximum duration of the transition time, and obtain the transition phase and stopping phase of the bipedal heavy-duty robot model based on the maximum duration of the transition time.

[0110] During the transition phase, the robot gradually decelerates, its gait period gradually lengthens (step frequency decreases), and its phase change rate gradually slows down. Therefore, a decaying oscillation function can be used to achieve a smooth transition in this phase. In the stopping phase, the robot has completed its deceleration and is in a stable stopped state. At this point, the phase scaling factor should decrease linearly to zero to ensure the robot comes to a complete stop.

[0111] S8, during the transition phase, phase and amplitude scaling functions are obtained in an exponential decay form to achieve a smooth transition of the bipedal heavy-duty robot's gait.

[0112] Specifically, during the transition phase, the methods for obtaining the phase and amplitude scaling functions include:

[0113] S81, obtain the timeout duration since the stop command was triggered. According to the maximum duration of the transition time And the timing duration is used to obtain the oscillation suppression term for generating a half-cycle cosine wave.

[0114] The timing duration is the independent variable, representing the time elapsed after the stop action begins. The oscillation suppression term can be expressed as: Its function is to generate an oscillating waveform that rises from 0 and then falls, with an oscillation period of 2× (Half a cosine period) has the physical significance of simulating the process of a robot gradually reducing its stride and step frequency, ensuring a smooth transition.

[0115] S82, obtain the oscillation damping coefficient Based on the oscillation attenuation coefficient and the timing duration, an exponential decay term for envelope suppression of the oscillation amplitude is obtained.

[0116] The oscillation decay coefficient is dimensionless and is used to control the rate of oscillation decay during the transition phase. A larger value indicates faster decay. This exponential decay term can be expressed as: Its function is to gradually decay the amplitude of the function to 0 over time, ensuring that the oscillations during the transition phase do not continue indefinitely. Physically, this simulates the process of a robot's kinetic energy decaying over time. The larger the value, the faster the decay and the more quickly it stops. The oscillation decay coefficient is generally proportional to the ground friction coefficient; it can increase with a high friction coefficient. This accelerates the decay.

[0117] S83, obtain the phase and amplitude scaling functions based on the product of the oscillation suppression term and the exponential decay term.

[0118] For example, the phase and amplitude scaling functions can be expressed as follows: .in, This represents the phase and amplitude scaling factor, which achieves gradual adjustment of gait (such as decreasing stride length) through an oscillatory term and ensures that the adjustment amplitude decreases over time through exponential decay to avoid abrupt changes in motion. When the value is 1, the phase changes normally, and when it is 0, the phase stops changing.

[0119] S9, in the stopping phase, the phase and amplitude scaling functions are obtained using a linear decay method to achieve stable stopping of the bipedal heavy-duty robot. Specifically, in the stopping phase, the method for obtaining the phase and amplitude scaling functions includes:

[0120] S91, obtain the duration difference between the timing duration and the maximum transition time;

[0121] S92, obtain the linear convergence slope The phase and amplitude scaling functions are obtained based on the time difference and the linear convergence slope.

[0122] The linear convergence slope controls the rate at which the phase scaling factor decreases linearly during the stopping phase; a larger value results in faster stopping. For example, the phase and amplitude scaling functions... When the timing duration equals the maximum transition time, It is set to 1, aligning with the end value of the transition phase. When... hour, When the value is 0, the bipedal heavy-duty robot comes to a complete stop. During the stopping phase, the robot has already decelerated significantly, and only minor phase adjustments are needed to bring it to a complete stop. Linear descent ensures the stability and controllability of the stopping process, avoiding control complexity caused by nonlinear changes.

[0123] The recommended range for the maximum transition time is 0.5-1.5 seconds. A maximum duration that is too short can easily lead to sudden stop jitter, while a duration that is too long will result in a slow stopping response. Here, the maximum transition time can be adjusted according to the robot's mass; the greater the mass, the greater the inertia, and the longer the maximum transition time. The recommended value range for the oscillation attenuation coefficient is 2-5. The recommended value range for the linear convergence slope is 0.5-2.0, which can be adjusted according to the joint motor response speed; the faster the response speed, the larger the oscillation attenuation coefficient can be selected.

[0124] That is to say, the phase and amplitude scaling functions are expressed as Specifically, when humans stop walking, they gradually reduce their stride length and stride frequency through buffer steps (1-2 steps) to smoothly transition their center of gravity to a stationary state. This phase and amplitude scaling function achieves a similar effect through a two-stage design of exponentially decaying oscillation (transition phase) → (stopping phase), and is mainly used to simulate the gait characteristics at the end of human walking.

[0125] Specifically, the simulation experiment of the bipedal heavy-duty robot model based on the gait phase generation function is as follows: the simulation experiment of the bipedal heavy-duty robot model is carried out based on the gait phase generation function and the phase and amplitude scaling functions.

[0126] Preferably, the linear convergence slope and oscillation damping coefficient can be calibrated by minimizing the weighted sum of stopping energy consumption and impact calibration parameters. That is, through a function .in, Total power of the joints The magnitude of the acceleration of the center of mass. The impact penalty coefficient is typically taken as 0.4. This function measures the time required for a bipedal heavy-duty robot to stop. Here, by mathematically optimizing the oscillation damping coefficient and linear convergence slope in the phase scaling function, the robot's stopping process can simultaneously meet the engineering requirements of high efficiency and high safety. In summary, this function has the following functions: 1. Optimizing energy consumption during the robot's stopping process, improving endurance; 2. Limiting the maximum acceleration impact during stopping, protecting the mechanical structure and electronic components; 3. Coordinating the trade-off between energy consumption and impact through an impact penalty coefficient.

[0127] In summary, the phase and amplitude scaling function has the following functions: 1. It enables a natural and gradual change in step frequency during the robot's start-stop phase, avoiding mechanical shock; 2. It minimizes energy loss during the stopping process through a composite control of exponential decay and oscillation; 3. It prevents instability caused by sudden stops, such as forward or backward tilting of the center of gravity, and is especially suitable for sloping terrain.

[0128] An embodiment of the present invention also provides a bipedal heavy-duty robot control system for implementing the aforementioned bipedal heavy-duty robot control method. The system includes a simulation model construction module, a terrain adaptation term acquisition module, an energy optimization term acquisition module, a gait coordination compensation term acquisition module, a gait phase function acquisition module, and a simulation experiment and result verification module.

[0129] The simulation model construction module is used to construct a bipedal heavy-duty robot model; the terrain adaptation module acquires terrain adaptation terms that adjust the gait frequency based on plantar pressure and ground slope to enhance gait stability; the energy optimization module acquires energy optimization terms that adjust the gait frequency based on the amplitude of torso height fluctuations to minimize torso kinetic energy fluctuations; the gait coordination compensation module acquires gait coordination compensation terms that adjust the gait frequency based on the phase deviation of the left and right legs to make the phase deviation return to the target value; the gait phase function acquisition module acquires dynamic gait frequency adjustment terms based on the terrain adaptation terms, energy optimization terms, and gait frequency coordination compensation terms, and acquires a gait phase generation function based on the dynamic gait frequency adjustment terms; the simulation experiment and result verification module conducts simulation experiments on the bipedal heavy-duty robot model based on the gait phase generation function. If the simulation results pass verification, the bipedal heavy-duty robot is controlled according to the gait phase generation function.

[0130] Specifically, the motion simulation of the bipedal heavy-duty robot in straight lines and circular trajectories can be performed using Matlab simulation software. In the straight-line motion simulation, the bipedal heavy-duty robot model adopts a humanoid structure, comprising six main components: torso, hip, lower limb joints, thigh, calf, and foot. The dimensions of each component are designed with reference to the parameters of typical bipedal heavy-duty robots, in millimeters, to ensure that the model is proportionally close to human gait characteristics while meeting practical engineering requirements. The specific dimensions are as follows: torso length: 120 mm, used to connect the left and right legs and support the upper body; hip length: 131 mm, connecting the torso and lower limbs; lower limb joint length: 139 mm, allowing for lateral fine-tuning; thigh length: 259 mm, providing primary support; calf length: 360 mm, extending stride length; foot length: 190 mm, simulating heel-toe contact. In Matlab, the model is constructed based on a three-dimensional coordinate system, and the components are drawn using three-dimensional plotting functions. The torso is represented by a thick black line, while the hips, lower limb joints, thighs, calves, and feet of the left and right legs are distinguished by different colors (such as red, green, blue, purple, and cyan) to visually demonstrate the dynamic changes of the kinematic chain. To simulate realistic ground constraints, the simulation environment includes a planar mesh ground, covering an area of ​​±500 mm along the x-axis and ±200 mm along the y-axis, using a light gray semi-transparent material (0.3 transparency) to clearly show the ground position while avoiding obscuring the robot's movement trajectory.

[0131] To achieve natural and stable straight-line walking, a series of gait parameters were designed, taking into account robot size, gait smoothness, and energy efficiency. These parameters include: Time vector: simulation time from 0 to 20 seconds, time step 0.01 seconds, ensuring high-precision trajectory calculation; Step frequency: 0.6 Hz, close to the slow walking frequency of humans (approximately 90 steps / minute), suitable for low-speed stable gait; Hip swing amplitude: π / 8 rad (approximately 22.5°), limiting forward and backward swing to maintain center of gravity stability; Lower limb joint swing amplitude: π / 36 rad (approximately 5°), used for lateral fine-tuning to enhance gait naturalness; Stride length: 200 mm, in proportion to leg length (approximately 758 mm, hip to foot), close to 0.25 times the human stride; Trunk height fluctuation amplitude: 15 mm, simulating the natural fluctuation of the center of gravity with gait; Trunk tilt amplitude: π / 48 rad (approximately 3.75°), maintaining posture stability and preventing excessive tilting. These parameters were determined through repeated adjustments to ensure that the gait conforms to both the structural constraints of the robot and the biomechanical characteristics of human walking, thus providing a reliable foundation for subsequent simulations.

[0132] In the circular trajectory motion simulation, the model is constructed based on a three-dimensional coordinate system, and the plot3 function is used to draw each component. The torso is represented by a thick black line, and the hips, lower limb joints, thighs, calves, and feet of the left and right legs are distinguished by different colors (such as red, green, blue, purple, and cyan) to visually demonstrate the dynamic changes of the kinematic chain. To simulate realistic ground constraints, the simulation environment includes a planar mesh ground with a mesh range of ±500 mm on the x-axis and ±200 mm on the y-axis, using a light gray semi-transparent material (0.3 transparency) to clearly present the ground position while avoiding occlusion of the robot's trajectory. These settings are consistent with the linear simulation to ensure the comparability of the model and the environment.

[0133] Regarding gait parameter design: To achieve natural and stable walking along a circular path, the following gait parameters were designed, taking into account robot size, path curvature, and biomechanical characteristics; Time vector: Simulation time ranges from 0 seconds to 20 seconds, with a step length of 0.01 seconds to ensure high accuracy in trajectory calculation; Step frequency: 0.6 Hz, close to the frequency of slow human walking (approximately 90 steps / minute), suitable for low-speed stable gait; Hip swing amplitude: π / 10 radians (approximately 18 degrees), slightly smaller than the π / 8 radians used in linear simulations, to accommodate continuous turning along the circular trajectory. Requirements: Lower limb joint swing amplitude: π / 36 radians (approximately 5 degrees), used for lateral fine-tuning to enhance gait coordination; Trunk height fluctuation amplitude: 15 mm, simulating the natural fluctuation of the center of gravity with gait; Trunk tilt amplitude: π / 48 radians (approximately 3.75 degrees), to maintain posture stability and prevent excessive tilting; Knee joint angle constraints: minimum angle 0 radians (fully extended), maximum angle 135 degrees (human knee flexion limit), maximum angle during the support phase 10 degrees (close to extension), ensuring that knee joint movement conforms to biomechanical principles.

[0134] For circular trajectory localization: The robot's torso center moves along a circular path, defined by the radius and angular velocity. The angle theta on the ring increases linearly with time, calculated as angular_speed * time(t). The global coordinates of the torso center are: x-coordinate: radius * cos(theta); y-coordinate: radius * sin(theta); z-coordinate: based on leg length and amplitude of undulation, approximately hip_length + lower_joint_length + thigh_length + calf_length + trunk_height_amplitude (trunk amplitude * cos(phase)). The tangent direction is [sin(theta), -cos(theta), 0], representing the local forward direction of clockwise movement. Unlike linear simulation, circular trajectory adds a y-axis coordinate and dynamic tangent adjustment to ensure the robot always faces forward on the path.

[0135] like Figures 9-19 As shown, this embodiment realizes motion simulation of a bipedal heavy-duty robot in humanoid linear walking and humanoid circular trajectory walking, and accurately measures the angle changes of the robot's joints in the two movement modes. The control system constructs a bipedal heavy-duty robot model and obtains a dynamic gait frequency adjustment term by constructing a terrain adaptation term, an energy optimization term, and a gait frequency coordination compensation term. The gait phase generation function can be obtained through the dynamic gait frequency adjustment term. Based on this gait phase generation function, it can better adapt to terrain changes or energy efficiency requirements to cope with complex scenarios such as slopes and soft ground, thereby improving the dynamic adaptability of gait frequency.

[0136] As a preferred technical solution, the control system further includes a stage acquisition module and a scaling function acquisition module.

[0137] The phase acquisition module is used to acquire the maximum duration of the transition time, and to acquire the transition phase and the stopping phase of the bipedal heavy-duty robot model based on the maximum duration of the transition time; the scaling function acquisition module is used to acquire the phase and amplitude scaling functions in the form of exponential decay in the transition phase to achieve a smooth transition of the bipedal heavy-duty robot's gait, and to acquire the phase and amplitude scaling functions in the form of linear decay in the stopping phase to achieve a stable stop of the bipedal heavy-duty robot.

[0138] Specifically, the scaling function acquisition module includes an oscillation suppression term acquisition unit, an exponential decay term acquisition unit, and a scaling function acquisition unit.

[0139] The oscillation suppression term acquisition unit is used to acquire the timing duration from the start of the trigger stop command, and acquire the oscillation suppression term for generating a half-cycle cosine wave based on the maximum transition time and the timing duration; the exponential decay term acquisition unit is used to acquire the oscillation decay coefficient, and acquire the exponential decay term for envelope suppression of the oscillation amplitude based on the oscillation decay coefficient and the timing duration; the scaling function acquisition unit is used to acquire the phase and amplitude scaling functions based on the product of the oscillation suppression term and the exponential decay term.

[0140] Transition weight cosine function formula While initial abrupt changes can be avoided, there may be issues with long transition times, and mechanical power consumption constraints are not considered. This phase and amplitude scaling function has the following functions: 1. It enables a natural and gradual change in step frequency during the robot's start-stop phase, avoiding mechanical shock; 2. It minimizes energy loss during the stopping process through a composite control of exponential decay and oscillation; 3. It prevents instability caused by sudden stops, such as forward or backward tilting of the center of gravity, and is particularly suitable for sloping terrain.

[0141] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A control method for a bipedal heavy-duty robot, characterized in that, The control method includes the following steps: Construct a bipedal heavy-duty robot model; Obtain terrain adaptation parameters to adjust cadence based on plantar pressure and ground slope to enhance gait stability; Obtain an energy optimization term for adjusting the stride frequency based on the amplitude of torso height fluctuations to minimize torso kinetic energy fluctuations; Obtain a gait coordination compensation term used to adjust the stride frequency based on the phase deviation of the left and right legs so that the phase deviation returns to the target value; The dynamic gait frequency adjustment term is obtained based on the terrain adaptation term, energy optimization term, and gait frequency coordination compensation term, and the gait phase generation function is obtained based on the dynamic gait frequency adjustment term. A simulation experiment of a bipedal heavy-duty robot model is conducted based on the gait phase generation function. If the simulation results pass the verification, the bipedal heavy-duty robot is controlled according to the gait phase generation function. The terrain adaptation term is expressed as follows: , These represent the pressure difference weight and the slope weight, respectively. These represent the pressure difference term and the slope term, respectively. Indicates heel pressure, Indicates toe pressure, Indicates the maximum pressure on the sole of the foot. Indicates the slope of the ground; The energy optimization term is expressed as , Indicates the energy sensitivity coefficient. Indicates the standard deviation of acceleration. Indicates the reference fluctuation threshold. Indicates the symbolic function value. Indicates the power consumption of the joint motor; The step frequency coordination compensation term is represented as , Indicates proportional gain. Indicates integral gain. Indicates phase deviation; Dynamic cadence adjustment item Expressed as , These represent the weighting coefficients of the terrain adaptation term, energy optimization term, and stride frequency coordination compensation term, respectively. Gait phase generation function Represented as , Based on the base step frequency.

2. The control method for a bipedal heavy-duty robot as described in claim 1, characterized in that, The control method further includes the following steps: Obtain the maximum duration of the transition time, and obtain the transition phase and stopping phase of the bipedal heavy-duty robot model based on the maximum duration of the transition time; During the transition phase, the phase and amplitude scaling functions are obtained in an exponential decay form to achieve a smooth transition of the bipedal heavy-duty robot's gait. During the stopping phase, the phase and amplitude scaling functions are obtained in a linear decay form to achieve stable stopping of the bipedal heavy-duty robot; Specifically, the simulation experiment of the bipedal heavy-duty robot model based on the gait phase generation function is as follows: the simulation experiment of the bipedal heavy-duty robot model is carried out based on the gait phase generation function and the phase and amplitude scaling functions.

3. The control method for a bipedal heavy-duty robot as described in claim 2, characterized in that, During the transition phase, the methods for obtaining the phase and amplitude scaling functions include: Obtain the timing duration from the start of the stop command, and obtain the oscillation suppression term for generating a half-cycle cosine wave based on the maximum transition time and the timing duration. Obtain the oscillation attenuation coefficient, and based on the oscillation attenuation coefficient and the timing duration, obtain the exponential attenuation term used to suppress the oscillation amplitude by envelope. The phase and amplitude scaling functions are obtained by multiplying the oscillation suppression term and the exponential decay term.

4. The control method for a bipedal heavy-duty robot as described in claim 3, characterized in that, During the stopping phase, the methods for obtaining the phase and amplitude scaling functions include: Obtain the duration difference between the timing duration and the maximum transition time; Obtain the linear convergence slope, and obtain the phase and amplitude scaling functions based on the time difference and the linear convergence slope.

5. A control system for a bipedal heavy-duty robot, used to implement the bipedal heavy-duty robot control method as described in any one of claims 1-4, characterized in that, The control system includes: The simulation model building module is used to build models of bipedal heavy-duty robots. The terrain adaptation module acquires terrain adaptation parameters that adjust gait frequency based on plantar pressure and ground slope to enhance gait stability. The energy optimization term acquisition module is used to acquire energy optimization terms for adjusting the stride frequency based on the amplitude of torso height fluctuations in order to minimize torso kinetic energy fluctuations. The gait coordination compensation term acquisition module is used to acquire gait coordination compensation terms for adjusting the stride frequency based on the phase deviation of the left and right legs so that the phase deviation returns to the target value. The gait phase function acquisition module is used to acquire a dynamic gait frequency adjustment term based on the terrain adaptation term, energy optimization term, and gait frequency coordination compensation term, and to acquire a gait phase generation function based on the dynamic gait frequency adjustment term; The simulation experiment and result verification module performs simulation experiments on the bipedal heavy-duty robot model based on the gait phase generation function. If the simulation results pass the verification, the bipedal heavy-duty robot is controlled according to the gait phase generation function.

6. The bipedal heavy-duty robot control system as described in claim 5, characterized in that, The control system further includes: The phase acquisition module is used to acquire the maximum duration of the transition time, and to acquire the transition phase and the stopping phase of the bipedal heavy-duty robot model based on the maximum duration of the transition time. The scaling function acquisition module is used to acquire the phase and amplitude scaling functions in an exponential decay form during the transition phase to achieve a smooth transition of the bipedal heavy-duty robot's gait, and to acquire the phase and amplitude scaling functions in a linear decay form during the stopping phase to achieve a stable stop of the bipedal heavy-duty robot.

7. The bipedal heavy-duty robot control system as described in claim 6, characterized in that, The scaling function acquisition module includes: The oscillation suppression term acquisition unit is used to acquire the timing duration from the start of the trigger stop command, and to acquire the oscillation suppression term for generating a half-cycle cosine wave based on the maximum transition time and the timing duration. An exponential decay term acquisition unit is used to acquire an oscillation decay coefficient and, based on the oscillation decay coefficient and timing duration, acquire an exponential decay term used to envelop and suppress the oscillation amplitude. The scaling function acquisition unit is used to obtain the phase and amplitude scaling functions based on the product of the oscillation suppression term and the exponential decay term.

Images