Modularized bionic joint and dynamic balance control humanoid robot system

Abstract

The invention discloses a modular bionic joint and dynamic balance control humanoid robot system, which relates to the technical field of robots and comprises a sensing module and a decision module, the sensing module comprises a multi-modal sensor, and the decision module comprises a bionic strategy module and an intelligent algorithm module. The bionic strategy module comprises an ankle joint and hip joint cooperation module and a gait planning module, the intelligent algorithm module comprises a real-time dynamic prediction control module and a strategy optimization module, and the execution module comprises a modular bionic joint module and a control system module. The intelligent algorithm module comprises a real-time dynamic prediction control module and a strategy optimization module, the real-time dynamic prediction control module is used for providing real-time dynamic balance control and optimizing current control input by predicting a future state, and the strategy optimization module is used for providing long-term strategy optimization and learning optimal gait parameters through a trial and error mechanism; therefore, efficient and robust gait optimization can be realized.

Description

Technical Field

[0001] The invention relates to the technical field of robots, and in particular to a humanoid robot system with modular bionic joints and dynamic balance control. Background Art

[0002] A humanoid robot, also known as a bionic man, is a robot designed to mimic human appearance and behavior, especially one with a human-like body. Until recently, the concept of humanoid robots was mainly in the realm of science fiction, often seen in movies, television, comics, novels, etc. Advances in robotics have made it possible to design functional, realistic humanoid robots.

[0003] In the existing humanoid robot technology, in the application of bionic joints, according to CN110027002B-A bionic joint control system and method based on multi-motor drive, it is recorded that "and real-time control of the stiffness of the joints, so that multiple motors can work together effectively like multiple muscles of a biological organism, improving the flexibility and variable stiffness performance of the bionic joints, and solving the problems of poor variable stiffness performance and poor control flexibility. Since the joints are controlled by multiple motors, when a motor fails, the system can still operate normally, thereby improving the robustness and reliability of the robot; in addition, through the bionic joints driven by multiple motors, the cost of manufacturing bionic robots in high-torque scenarios can also be reduced." It can be seen from the content that the bionic joints in the existing technology are currently at the stage of imitating the rigidity and flexibility of movements, but there is a lack of effective development for the subsequent combination of dynamic balance.

[0004] Therefore, a humanoid robot system with modular bionic joints and dynamic balance control is designed. Summary of the invention

[0005] In order to overcome the above-mentioned deficiencies, the present invention provides a humanoid robot system with modular bionic joints and dynamic balance control.

[0006] The present invention achieves the above-mentioned purpose through the following technical solutions:

[0007] A humanoid robot system with modular bionic joints and dynamic balance control, comprising

[0008] Perception module: The perception module includes multimodal sensors, which can identify the surrounding environment, the working status of the robot's bionic joints, and the robot's motion posture;

[0009] Decision-making module: including bionic strategy module and intelligent algorithm module. According to the data information collected by the perception module, the bionic strategy module and intelligent algorithm module are combined to make gait decisions for the robot.

[0010] The bionic strategy module includes the ankle and hip joint coordination module and the gait planning module. The ankle and hip joint coordination module synergizes to imitate the natural mechanism of human walking and achieves anti-interference balance by dynamically adjusting the center of gravity. The ankle joint is mainly responsible for rapid response, and the hip joint provides stable support. The two work together to ensure that the robot remains stable in complex terrains. The gait planning module generates a natural and efficient gait based on the dynamic model and bionic strategy to adapt to different terrains and task requirements. The gait planning module achieves stable walking by optimizing parameters such as step length, step frequency, support phase and swing phase.

[0011] The intelligent algorithm module includes a real-time dynamic prediction control module and a strategy optimization module. The real-time dynamic prediction control module is used to provide real-time dynamic balance control and optimize the current control input by predicting the future state. The strategy optimization module is used to provide long-term strategy optimization and learn the optimal gait parameters through a trial-and-error mechanism.

[0012] Execution module: includes modular bionic joint module and control system module. The control system module drives the modular bionic joint module to execute actions by receiving action instructions from the decision module.

[0013] Preferably, the multimodal sensor includes a visual sensor for identifying environmental terrain and obstacles, a force sensor for real-time monitoring of joint forces and ground reaction forces, an inertial measurement unit for detecting robot posture and acceleration, and a tactile sensor for enhancing operational accuracy and environmental interaction capabilities.

[0014] Preferably, the implementation of the ankle joint and hip joint coordination module comprises the following steps:

[0015] S11, perception data collection, obtain robot status and environment information through multimodal sensors to provide data support for collaborative control;

[0016] S12, center of gravity offset detection, used to monitor the center of gravity state in real time and provide a basis for coordinated adjustment. The center of gravity offset detection includes calculating the center of gravity position and detecting the offset. The center of gravity position is calculated based on an inverted pendulum model or a multi-body dynamics model to calculate the relative position of the center of gravity and the support point. The offset detection is based on the inertial measurement unit data to determine whether the center of gravity deviates from the stable area;

[0017] S13, ankle joint fast response, used to quickly adjust the ankle joint angle through servo motor or hydraulic drive, compensate for the center of gravity offset, and ensure instant adjustment of the robot's posture;

[0018] S14, hip joint stabilization support, used to predict and adjust the hip joint torque through the intelligent algorithm module, provide stable support, and ensure the overall balance of the robot;

[0019] S15, closed-loop feedback and optimization, evaluates the adjustment effect based on the data from the force sensor and inertial measurement unit, and dynamically optimizes the control parameters of the ankle and hip joints to achieve closed-loop control. Through the feedback mechanism, the accuracy and robustness of the coordinated control are ensured.

[0020] Preferably, the implementation of the gait planning module comprises the following steps:

[0021] S21, Task and environment analysis: Determine gait goals according to task types (such as walking, climbing, carrying), and use visual sensors to identify terrain features (such as slopes, obstacles), thereby providing task and environment information for gait planning;

[0022] S22, gait parameter initialization, according to the task requirements and terrain characteristics, initialize the step length and step frequency, and determine the time ratio of the support phase (legs touching the ground) and the swing phase (legs moving in the air), so as to provide initial parameters for gait generation;

[0023] S23, gait generation and optimization, is used to predict, generate gait and optimize synchronously through intelligent algorithm modules, so as to ensure the efficiency and adaptability of gait;

[0024] S24, real-time adjustment and feedback, is used to evaluate the gait execution effect according to the data of force sensor and inertial measurement unit, and adjust the gait parameters in real time according to the feedback data to adapt to environmental changes, thereby ensuring the robustness and adaptability of the gait through closed-loop control.

[0025] Preferably, the implementation of the real-time dynamic prediction control module requires the following steps:

[0026] S31, state prediction, predicting the future state, providing a basis for optimal control, the formula used is as follows:

[0027] x t+1 =f(x t ,u t )

[0028] Among them, x t is the system state at time t, u t is the control input (e.g., joint torque) at time t, and f is the system dynamics module;

[0029] S32, objective function optimization, is used to optimize the control input, minimize the state error and control cost, and the optimization formula is as follows:

[0030]

[0031] Among them, x refis the reference state (such as the expected center of gravity position), Q and R are weight matrices, representing the cost of state error and control input respectively, and N is the prediction time domain, indicating the number of future prediction steps.

[0032] Preferably, the implementation of the strategy optimization module requires the following steps:

[0033] S41, state-action value function calculation, used to evaluate the long-term value of the action and guide strategy optimization. The calculation formula is as follows:

[0034]

[0035] is the state-action value function, which means that in state s t Next, perform action a t Expected cumulative reward of

[0036] r t is the immediate reward at time t (such as gait stability score), γ is the discount factor (0≤γ≤1), indicating the importance of future rewards;

[0037] S42, deterministic strategy update, is used to select the optimal gait strategy through the maximum Q function, and the determination formula is as follows:

[0038]

[0039] Among them, μ(s t ) is a deterministic strategy, indicating that in state s t The best action to choose.

[0040] Preferably, the modular bionic joint module includes a drive module and a feedback module. The drive module includes a servo motor and a hydraulic drive mechanical structure to control various links of the robot to realize gait execution. The feedback module includes a force sensor and an inertial measurement unit for real-time feedback of joint status.

[0041] Preferably, the control system module includes a CPU, and the CPU is used to receive data from the perception module and the decision-making module to control the driving module to work.

[0042] Preferably, the specific implementation of the system includes the following specific steps:

[0043] Step 1: Perception data collection: multimodal sensors obtain robot status and environment information;

[0044] Step 2: Gait planning: the gait planning module is combined with the intelligent algorithm module to generate a natural and efficient gait that adapts to the task and terrain requirements;

[0045] Step 3: Adjust the center of gravity. The ankle and hip joint coordination module is combined with the intelligent algorithm module to adjust the center of gravity to ensure dynamic balance.

[0046] Step 4: Gait execution: the modular bionic joint module realizes precise movement according to the planning instructions of steps 2 and 3;

[0047] Step 5: Feedback and optimization: dynamically optimize gait and collaborative control parameters based on sensor feedback data in the modular bionic joint module.

[0048] The beneficial effects of the present invention are as follows: in the humanoid robot system with modular bionic joints and dynamic balance control:

[0049] 1. The intelligent algorithm module includes a real-time dynamic predictive control module and a strategy optimization module. The real-time dynamic predictive control module is used to provide real-time dynamic balance control and optimize the current control input by predicting the future state. The strategy optimization module is used to provide long-term strategy optimization and learn the optimal gait parameters through a trial and error mechanism, thereby achieving efficient and robust gait optimization.

[0050] 2. Based on the sensor feedback data in the modular bionic joint module, the gait and collaborative control parameters are dynamically optimized, and dynamic balance optimization can be performed. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention will now be described by way of example with reference to the accompanying drawings, in which:

[0052] Figure 1 It is a step diagram of the present invention. DETAILED DESCRIPTION

[0053] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0054] like Figure 1 As shown, a humanoid robot system with modular bionic joints and dynamic balance control includes

[0055] Perception module: The perception module includes multimodal sensors, which can identify the surrounding environment, the working status of the robot's bionic joints, and the robot's motion posture;

[0056] Decision-making module: including bionic strategy module and intelligent algorithm module. According to the data information collected by the perception module, the bionic strategy module and intelligent algorithm module are combined to make gait decisions for the robot.

[0057] The bionic strategy module includes the ankle and hip joint coordination module and the gait planning module. The ankle and hip joint coordination module synergizes to imitate the natural mechanism of human walking and achieves anti-interference balance by dynamically adjusting the center of gravity. The ankle joint is mainly responsible for rapid response, and the hip joint provides stable support. The two work together to ensure that the robot remains stable in complex terrains. The gait planning module generates a natural and efficient gait based on the dynamic model and bionic strategy to adapt to different terrains and task requirements. The gait planning module achieves stable walking by optimizing parameters such as step length, step frequency, support phase and swing phase.

[0058] The intelligent algorithm module includes a real-time dynamic prediction control module and a strategy optimization module. The real-time dynamic prediction control module is used to provide real-time dynamic balance control and optimize the current control input by predicting the future state. The strategy optimization module is used to provide long-term strategy optimization and learn the optimal gait parameters through a trial-and-error mechanism.

[0059] Execution module: includes modular bionic joint module and control system module. The control system module drives the modular bionic joint module to execute actions by receiving action instructions from the decision module.

[0060] Specifically, the multimodal sensor includes a visual sensor for identifying environmental terrain and obstacles, a force sensor for real-time monitoring of joint forces and ground reaction forces, an inertial measurement unit for detecting robot posture and acceleration, and a tactile sensor for enhancing operational accuracy and environmental interaction capabilities.

[0061] Specifically, the implementation of the ankle joint and hip joint coordination module includes the following steps:

[0062] S11, perception data collection, obtain robot status and environment information through multimodal sensors to provide data support for collaborative control;

[0063] S12, center of gravity offset detection, used to monitor the center of gravity state in real time and provide a basis for coordinated adjustment. The center of gravity offset detection includes calculating the center of gravity position and detecting the offset. The center of gravity position is calculated based on an inverted pendulum model or a multi-body dynamics model to calculate the relative position of the center of gravity and the support point. The offset detection is based on the inertial measurement unit data to determine whether the center of gravity deviates from the stable area;

[0064] S13, ankle joint fast response, used to quickly adjust the ankle joint angle through servo motor or hydraulic drive, compensate for the center of gravity offset, and ensure instant adjustment of the robot's posture;

[0065] S14, hip joint stabilization support, used to predict and adjust the hip joint torque through the intelligent algorithm module, provide stable support, and ensure the overall balance of the robot;

[0066] S15. Closed-loop feedback and optimization: Based on the data from the force sensor and the inertial measurement unit, evaluate the adjustment effect, and at the same time dynamically optimize the control parameters of the ankle joint and the hip joint to achieve closed-loop control. Through the feedback mechanism, ensure the accuracy and robustness of the coordinated control.

[0067] Specifically, the implementation of the gait planning module includes the following steps:

[0068] S21. Task and environment analysis: According to the task type (such as walking, climbing, carrying), determine the gait goal, and at the same time, through the vision sensor, identify the terrain features (such as slope, obstacles), so as to provide task and environment information for gait planning;

[0069] S22. Gait parameter initialization: According to the task requirements and terrain features, initialize the step length and step frequency, and at the same time determine the time ratio of the support phase (the leg touches the ground) and the swing phase (the leg moves in the air), so as to provide initial parameters for gait generation;

[0070] S23. Gait generation and optimization: Used to predict and generate the gait and synchronously optimize through the intelligent algorithm module, so as to ensure the efficiency and adaptability of the gait;

[0071] S24. Real-time adjustment and feedback: Used to evaluate the gait execution effect according to the data from the force sensor and the inertial measurement unit, and at the same time, according to the feedback data, real-time adjust the gait parameters to adapt to environmental changes, so as to ensure the robustness and adaptability of the gait through closed-loop control.

[0072] Specifically, the implementation of the real-time dynamic prediction control module requires the following steps:

[0073] S31. State prediction: Predict the future state to provide a basis for optimal control. The formula used is as follows:

[0074] x t+1 =f(x t ,u t

[0075] where x t is the system state at time t, u t is the control input at time t (such as joint torque), and f is the system dynamics module;

[0076] S32. Objective function optimization: Used to optimize the control input, minimize the state error and the control cost. The optimization formula is as follows:

[0077]

[0078] where x refis the reference state (such as the expected center of gravity position), Q and R are weight matrices, representing the cost of state error and control input respectively, and N is the prediction time domain, indicating the number of future prediction steps.

[0079] Specifically, the implementation of the strategy optimization module requires the following steps:

[0080] S41, state-action value function calculation, used to evaluate the long-term value of the action and guide strategy optimization. The calculation formula is as follows:

[0081]

[0082] is the state-action value function, which means that in state s t Next, perform action a t Expected cumulative reward of

[0083] r t is the immediate reward at time t (such as gait stability score), γ is the discount factor (0≤γ≤1), indicating the importance of future rewards;

[0084] S42, deterministic strategy update, is used to select the optimal gait strategy through the maximum Q function, and the determination formula is as follows:

[0085]

[0086] Among them, μ(s t ) is a deterministic strategy, indicating that in state s t The best action to choose.

[0087] Specifically, the modular bionic joint module includes a drive module and a feedback module. The drive module includes a servo motor and a hydraulic drive mechanical structure to control various links of the robot to realize gait execution. The feedback module includes a force sensor and an inertial measurement unit for real-time feedback of joint status.

[0088] Specifically, the control system module includes a CPU, and the CPU is used to receive data from the perception module and the decision-making module to control the driving module to work.

[0089] Specifically, the specific implementation of the system includes the following specific steps:

[0090] Step 1: Perception data collection: multimodal sensors obtain robot status and environment information;

[0091] Step 2: Gait planning: the gait planning module is combined with the intelligent algorithm module to generate a natural and efficient gait that adapts to the task and terrain requirements;

[0092] Step 3: Adjust the center of gravity. The ankle and hip joint coordination module is combined with the intelligent algorithm module to adjust the center of gravity to ensure dynamic balance.

[0093] Step 4: Gait execution: the modular bionic joint module realizes precise movement according to the planning instructions of steps 2 and 3;

[0094] Step 5: Feedback and optimization: dynamically optimize gait and collaborative control parameters based on sensor feedback data in the modular bionic joint module.

[0095] Embodiment 1:

[0096] Application of Strategy Optimization Module in Slope Terrain:

[0097] Status definition: s t The robot posture (such as tilt angle, center of gravity position) and terrain slope;

[0098] Action definition: a t are step length, step frequency, ankle and hip joint angles;

[0099] Reward function: r t Scores based on gait stability (e.g., center of gravity shift) and energy efficiency (e.g., joint torque);

[0100] Application Process:

[0101] 1. Initial strategy, the robot generates an initial gait on flat ground;

[0102] 2. Trial and error learning, performing gaits on the slope and adjusting parameters such as step length and step frequency according to the reward function;

[0103] 3. Strategy optimization: Through the algorithm of the strategy optimization module, the optimal gait strategy that adapts to the slope is gradually learned.

[0104] Embodiment 2:

[0105] Application of strategy optimization module on uneven ground:

[0106] Status definition: s t Changes in robot posture and ground height (detected by visual sensors or force sensors);

[0107] Action definition: a t are step length, step frequency, ankle and hip joint angles;

[0108] Reward function: r t Scores based on gait stability (e.g., center of gravity shift) and terrain adaptability (e.g., foot contact force);

[0109] Application Process:

[0110] Initial strategy, the robot generates an initial gait on flat ground;

[0111] Trial and error learning, performing gaits on uneven ground, adjusting parameters such as step length and step frequency according to the reward function;

[0112] Strategy optimization, through the algorithm of the strategy optimization module, gradually learn the optimal gait strategy to adapt to uneven terrain.

[0113] The above is based on the present invention as an inspiration. Through the above description, relevant staff can make various changes and modifications without departing from the technical idea of ​​this invention. The technical scope of this invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A humanoid robot system with modular bionic joints and dynamic balance control, characterized by: include Perception module: The perception module includes multimodal sensors, which can identify the surrounding environment, the working status of the robot's bionic joints, and the robot's motion posture; Decision-making module: including bionic strategy module and intelligent algorithm module. According to the data information collected by the perception module, the bionic strategy module and intelligent algorithm module are combined to make gait decisions for the robot. The bionic strategy module includes the ankle and hip joint coordination module and the gait planning module. The ankle and hip joint coordination module synergizes with the natural mechanism of human walking and achieves anti-interference balance by dynamically adjusting the center of gravity. The gait planning module generates a natural and efficient gait based on the dynamic model and bionic strategy to adapt to different terrains and task requirements. The intelligent algorithm module includes a real-time dynamic prediction control module and a strategy optimization module. The real-time dynamic prediction control module is used to provide real-time dynamic balance control and optimize the current control input by predicting the future state. The strategy optimization module is used to provide long-term strategy optimization and learn the optimal gait parameters through a trial-and-error mechanism. Execution module: includes modular bionic joint module and control system module. The control system module drives the modular bionic joint module to execute actions by receiving action instructions from the decision module.

2. The humanoid robot system with modular bionic joints and dynamic balance control according to claim 1, characterized in that: The multimodal sensor includes a visual sensor, a force sensor, an inertial measurement unit and a tactile sensor.

3. The humanoid robot system with modular bionic joints and dynamic balance control according to claim 1, characterized in that: The implementation of the ankle joint and hip joint coordination module includes the following steps: S11, perception data collection, obtain robot status and environment information through multimodal sensors to provide data support for collaborative control; S12, center of gravity offset detection, used to monitor the center of gravity state in real time and provide a basis for coordinated adjustment. The center of gravity offset detection includes calculating the center of gravity position and detecting the offset. The center of gravity position is calculated based on an inverted pendulum model or a multi-body dynamics model to calculate the relative position of the center of gravity and the support point. The offset detection is based on the inertial measurement unit data to determine whether the center of gravity deviates from the stable area; S13, ankle joint fast response, used to quickly adjust the ankle joint angle through servo motor or hydraulic drive, compensate for the center of gravity offset, and ensure instant adjustment of the robot's posture; S14, hip joint stabilization support, used to predict and adjust the hip joint torque through the intelligent algorithm module, provide stable support, and ensure the overall balance of the robot; S15, closed-loop feedback and optimization, evaluates the adjustment effect based on the data from the force sensor and inertial measurement unit, and dynamically optimizes the control parameters of the ankle and hip joints to achieve closed-loop control. Through the feedback mechanism, the accuracy and robustness of the coordinated control are ensured.

4. The humanoid robot system with modular bionic joints and dynamic balance control according to claim 1, characterized in that: The implementation of the gait planning module includes the following steps: S21, Task and environment analysis, determines the gait goal according to the task type, and identifies the terrain features through visual sensors, thereby providing task and environment information for gait planning; S22, gait parameter initialization, according to the task requirements and terrain characteristics, initialize the step length and step frequency, and determine the time ratio of the support phase and the swing phase, so as to provide initial parameters for gait generation; S23, gait generation and optimization, is used to predict, generate gait and optimize synchronously through intelligent algorithm modules, so as to ensure the efficiency and adaptability of gait; S24, real-time adjustment and feedback, is used to evaluate the gait execution effect according to the data of force sensor and inertial measurement unit, and adjust the gait parameters in real time according to the feedback data to adapt to environmental changes, thereby ensuring the robustness and adaptability of the gait through closed-loop control.

5. The humanoid robot system with modular bionic joints and dynamic balance control according to claim 1, characterized in that: The implementation of the real-time dynamic prediction control module requires the following steps: S31, state prediction, predicting the future state, providing a basis for optimal control, the formula used is as follows: x t+1 =f(x t ,u t ) Among them, x t is the system state at time t, u t is the control input (e.g., joint torque) at time t, and f is the system dynamics module; S32, objective function optimization, is used to optimize the control input, minimize the state error and control cost, and the optimization formula is as follows: Among them, x ref is the reference state (such as the expected center of gravity position), Q and R are weight matrices, representing the cost of state error and control input respectively, and N is the prediction time domain, indicating the number of future prediction steps.

6. The humanoid robot system with modular bionic joints and dynamic balance control according to claim 1, characterized in that: The implementation of the strategy optimization module requires the following steps: S41, state-action value function calculation, used to evaluate the long-term value of the action and guide strategy optimization. The calculation formula is as follows: is the state-action value function, which means that in state s t Next, perform action a t Expected cumulative reward of r t is the immediate reward at time t (such as gait stability score), γ is the discount factor (0≤γ≤1), indicating the importance of future rewards; S42, deterministic strategy update, is used to select the optimal gait strategy through the maximum Q function, and the determination formula is as follows: Among them, μ(s t ) is a deterministic strategy, indicating that in state s t The best action to choose.

7. The humanoid robot system with modular bionic joints and dynamic balance control according to claim 1, characterized in that: The modular bionic joint module includes a drive module and a feedback module. The drive module includes a servo motor and a hydraulic drive mechanical structure to control various links of the robot to realize gait execution. The feedback module includes a force sensor and an inertial measurement unit for real-time feedback of joint status.

8. The humanoid robot system with modular bionic joints and dynamic balance control according to claim 1, characterized in that: The control system module includes a CPU, and the CPU is used to receive data from the perception module and the decision-making module to control the driving module to work.

9. The humanoid robot system with modular bionic joints and dynamic balance control according to claim 1, characterized in that: The specific implementation of this system includes the following specific steps: Step 1: Perception data collection: multimodal sensors obtain robot status and environment information; Step 2: Gait planning: the gait planning module is combined with the intelligent algorithm module to generate a natural and efficient gait that adapts to the task and terrain requirements; Step 3: Adjust the center of gravity. The ankle and hip joint coordination module is combined with the intelligent algorithm module to adjust the center of gravity to ensure dynamic balance. Step 4: Gait execution: the modular bionic joint module realizes precise movement according to the planning instructions of steps 2 and 3; Step 5: Feedback and optimization: dynamically optimize gait and collaborative control parameters based on sensor feedback data in the modular bionic joint module.

Images