Humanoid Walking Control with ZMP Error Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current humanoid robot walking technologies, such as ZMP-based and FSM-based methods, face challenges in achieving stable walking while maintaining energy efficiency and avoiding joint rigidity and impact issues, particularly in compensating for zero moment point (ZMP) errors during FSM-based walking.
Innovation Solution
A humanoid robot system that includes a walking pattern creation unit, ZMP calculation and measurement units, and a compensation unit to calculate and apply compensation values to joint position and torque commands, ensuring stable walking by adjusting the ZMP based on calculated and measured values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If ZMP-based walking control method is used to accurately control joint position and control ZMP, then walking stability is improved, but position servo control gain becomes high causing high motor current and low energy efficiency
Solution Approach 1:
The patent implements dynamic adjustment of servo control gain based on the robot's walking state. The control system switches between high gain (for stability during critical phases) and low gain (for energy efficiency during stable phases), making the control system adaptable rather than static. This resolves the contradiction by applying high control authority only when necessary for stability while maintaining energy efficiency during other phases of the gait cycle.
Solution Approach 2:
The patent changes the servo control gain parameter dynamically based on the walking phase and ZMP error magnitude. By adjusting this critical control parameter according to real-time conditions, the system achieves both stability (when high gain is needed) and energy efficiency (when low gain suffices), directly resolving the trade-off between these two opposing requirements.
2Measurement precision
If ZMP-based walking control method is used with high position servo control gain, then ZMP control precision is improved, but joint rigidity increases causing great impact on obstacles when colliding
Solution Approach 1:
The patent makes the joint rigidity dynamic by adjusting servo control gain based on walking phase and environmental conditions. During phases requiring precise ZMP control, higher gain provides necessary precision. During phases where compliance is beneficial (such as potential contact with obstacles), lower gain reduces rigidity and impact. This dynamic adaptation resolves the contradiction between control precision and impact reduction.
Solution Approach 2:
The patent changes the servo control gain parameter to balance ZMP control precision and joint compliance. By lowering gain when precise ZMP control is less critical and when compliance with obstacles is beneficial, the system reduces impact forces while maintaining adequate control precision when needed, thus resolving the contradiction.
3Adaptability or versatility
If FSM-based walking control method is used to calculate joint torque by referring to operation states, then the robot can take various poses, but a separate balancing operation is required causing time delay and energy waste
Solution Approach 1:
The patent merges the walking control and balancing control into a unified system. Instead of executing separate balancing operations (such as stamping feet) when balance is disturbed, the system integrates balance maintenance into the continuous walking control loop. This eliminates the need for discrete, time-consuming balancing maneuvers while preserving the ability to take various poses through FSM state transitions.
Solution Approach 2:
The patent maintains continuous walking control action rather than interrupting for separate balancing operations. The unified control system continuously adjusts joint torques to maintain both walking pattern and balance simultaneously, eliminating the discontinuities and time delays associated with separate balancing maneuvers while preserving pose versatility through state machine transitions.
4Reliability
If FSM-based walking control method is used with separate balancing operation, then balance can be maintained, but energy is wasted due to unnecessary operations
Solution Approach 1:
The patent merges walking control and balancing control into a single unified system that operates continuously. This eliminates the need for separate, energy-consuming balancing operations (such as stamping feet) while maintaining balance through integrated torque control. The unified system achieves both walking and balance objectives simultaneously, reducing energy waste.
Solution Approach 2:
The patent implements continuous control action that simultaneously addresses walking and balancing objectives. Instead of interrupting the walking control for separate balancing operations, the system continuously adjusts torques to maintain both gait pattern and balance, eliminating the energy waste associated with discrete balancing maneuvers while preserving reliable balance maintenance.
Data Source
AI summary
Disclosed herein are a humanoid robot that compensates for a zero moment point (ZMP) error during finite state machine (FSM)-based walking to achieve stable walking and a walking control method thereof. The humanoid robot compensates for a joint position trajectory command or a joint torque command using compensation values calculated based on situations divided according to the position of a calculated ZMP and the position of a measured ZMP in a stable region of the robot.