Biped Robot Walking Pattern Control via Knee Stretching and Waist Minimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Biped robots exhibit a walking pattern that differs significantly from human walking, leading to low affinity and energy efficiency in human-robot coexistence environments due to improper knee angle control and excessive waist movement.
Innovation Solution
A method and system for controlling a biped robot to generate a walking pattern that maximally stretches the knees and minimizes waist movement, using a zero moment point (ZMP) equation and a human-like walking similarity measure to adjust knee angles during double and single support phases, ensuring the walking pattern similarity to human walking is enhanced.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the knees are stretched to an angle close to 180° to achieve human-like walking, then the walking pattern similarity to human increases, but a singularity pose is generated where the relative distance connecting ankle joint and hip joint equals the length of leg, causing excessive knee joint speed or inability to obtain desired knee joint angle
Solution Approach 1:
The patent applies preliminary action by predicting future knee joint angles and velocities using a state space model before actual movement occurs. The controller calculates anticipated knee joint states and uses this prediction to adjust control inputs in advance, preventing singularity issues before they arise while maintaining human-like walking patterns.
Solution Approach 2:
The patent implements feedback through a state space model that continuously monitors actual knee joint angles and velocities, compares them with predicted values, and adjusts control inputs based on the deviation. This feedback mechanism ensures stable knee joint control while achieving natural walking patterns by correcting errors in real-time.
2Stability of the object's composition
If the waist moves from side to side while maintaining regular height to achieve stable walking, then the walking stability increases, but the walking pattern considerably differs from human walking pattern
Solution Approach 1:
The patent applies dynamics by transitioning from static waist height maintenance to dynamic waist motion control. The controller allows the waist to move naturally in the vertical direction according to the generated trajectory, rather than constraining it to a fixed height. This dynamic approach enables human-like walking patterns while maintaining stability through coordinated leg and torso movements.
3Manufacturing precision
If the knees are stretched and ankles are rotated differently from typical humanoid walking pattern to achieve human-like walking, then the walking pattern similarity increases, but the point of time to stretch knees is improper and the waist shakes
Solution Approach 1:
The patent uses feedback through the state space model to monitor waist position and detect shaking. When waist instability is detected, the controller adjusts knee joint control inputs to correct the timing and magnitude of knee stretching, thereby stabilizing the waist while maintaining human-like walking patterns.
Solution Approach 2:
The patent replaces direct mechanical waist stabilization mechanisms with a computational approach. Instead of using physical dampers or mechanical constraints to prevent waist shaking, the system uses a state space model to predict and control knee joint movements, which indirectly stabilizes the waist through coordinated whole-body dynamics.
Data Source
AI summary
A method of controlling walking a biped robot to generate a walking pattern maximally similar to that of a human includes generating a walking pattern, calculating a walking pattern similarity corresponding to the walking pattern, and comparing the walking pattern similarity with a predetermined reference pattern similarity, and changing the walking pattern based on a result of the comparison. When the robot walks, a knee is maximally stretched and a horizontal movement of a waist is minimized such that the walking pattern of the robot is maximally similar to that of a human, thus enhancing an affinity for a human being and increasing energy efficiency.