Humanoid Balance Control via Non-Pathological Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional balance control methods for humanoids, such as biped robots, rely on complex computations involving Center of Mass (CoM) and Zero Moment Point (ZMP) that lack controllability, leading to instability and performance issues due to improper discretization of dynamics equations and lack of theoretical basis for initial position and velocity.
Innovation Solution
A system and method that utilize discrete-time dynamics equations with a non-pathological sampling frequency to compute ZMP and CoM position and velocity, ensuring controllability and stability by forming a closed loop system that guarantees the ZMP remains within a supporting region, thereby stabilizing balance control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional linear inverted pendulum model is used for balance control, then computation is simplified, but controllability of system is not guaranteed
Solution Approach 1:
The patent changes the sampling frequency parameter from arbitrary to specifically non-pathological sampling frequencies, which transforms the discretized dynamics equation to guarantee controllability while maintaining computational efficiency. This parameter change resolves the contradiction by selecting specific sampling frequencies that preserve system controllability properties.
2Productivity
If discretization is performed without considering controllability, then computation is faster, but system stability deteriorates
Solution Approach 1:
The patent specifies non-pathological sampling frequencies as a critical parameter change that ensures both computational efficiency and system stability. By selecting sampling frequencies that avoid pathological cases, the system maintains fast computation while guaranteeing stability through the preserved controllability of the discretized dynamics equation.
3Ease of operation
If constant values are input for CoM initial position and velocity, then control is simpler, but balance control performance deteriorates over time
Solution Approach 1:
The patent applies preliminary action by computing optimal initial CoM position and velocity values before executing the balance control task. Using the discretized dynamics equation with non-pathological sampling frequency, the system pre-calculates initial conditions that guarantee future balance stability, rather than using arbitrary constant values. This preliminary computation ensures both simplicity and long-term performance.
4Adaptability or versatility
If ZMP is not constrained to supporting region, then motion flexibility is higher, but balance stability decreases
Solution Approach 1:
The patent implements feedback by continuously monitoring ZMP position relative to the supporting region and using the discretized dynamics equation to adjust CoM trajectory. The closed-loop control system uses the guaranteed controllability of the non-pathological sampling frequency to dynamically constrain ZMP within the supporting region while maintaining motion flexibility through adaptive trajectory planning.
Data Source
AI summary
A system for balance control of a humanoid includes a control input generation unit which determines Zero Momentum Position (ZMP) of the humanoid with respect to ground, and a control output generation unit which determines Center of Mass (CoM) position and velocity of the humanoid, wherein the control input generation unit and the control output generation unit determine the ZMP and the CoM position and velocity, respectively, by applying a discrete-time dynamics equation sampled by applying a non-pathological sampling frequency.