Legged Robot Posture Control via Direct Acceleration Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional legged locomotion robots face challenges in achieving stable posture control due to high computational load and inaccuracies in calculating acceleration data, which hinders quick and accurate real-time adjustments, especially on uneven surfaces with varying friction coefficients.
Innovation Solution
The implementation of a sensor system with acceleration sensors, angular acceleration sensors, and angular velocity sensors at multiple points on the robot allows for direct measurement of posture and acceleration, enabling the formulation of a ZMP equation for precise posture control, independent of rigid body assumptions and surface conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If acceleration data is calculated through two-step differentiation of position data, then the control system can determine ZMP and external forces, but the computational load increases and measurement accuracy decreases
Solution Approach 1:
The patent replaces the mechanical calculation method (two-step differentiation of position data) with a direct sensor measurement approach. Acceleration sensors directly measure acceleration data, eliminating the need for complex computational differentiation processes. This substitution reduces computational load while improving measurement accuracy by obtaining acceleration data directly from sensors rather than deriving it from position data through mathematical operations.
2Adaptability or versatility
If the robot uses bipedal upright movement to navigate human environments, then it gains flexibility over irregular surfaces, but stability and control difficulty increase
Solution Approach 1:
The patent implements a feedback control system that continuously monitors acceleration data from sensors distributed at multiple points on the robot body. The control unit processes this real-time acceleration data to determine ZMP (zero moment point) and external forces, then adjusts actuator commands to maintain stable posture. This closed-loop feedback mechanism enables the robot to adapt to irregular surfaces and maintain balance despite the inherent instability of bipedal upright movement.
Solution Approach 2:
The patent divides the robot's measurement system into multiple segmented sensor units distributed at different locations (head, torso, hips, knees, ankles, feet). Each sensor measures local acceleration data, providing distributed feedback about the robot's posture and external forces. This segmentation allows the control system to comprehensively monitor and control different body segments independently, improving overall posture stability while maintaining movement flexibility.
3Measurement precision
If sensors are distributed at multiple points on the robot, then measurement of movement parameters improves, but device complexity increases
Solution Approach 1:
The patent employs identical sensor units distributed at multiple locations on the robot body. Each sensor unit serves multiple functions: measuring local acceleration data, determining contributions to overall robot acceleration, and providing information for both ZMP calculation and external force determination. This universal, multi-functional sensor design improves measurement precision through distributed sensing while reducing device complexity by using standardized components rather than specialized sensors for each location.
Data Source
AI summary
The lumbar part of a robot as a controlled-object point where the mass amount becomes maximum is set as the origin of a local coordinate, an acceleration sensor is disposed at the controlled-object point to directly measure the attitude and acceleration at that position to control the robot to take a stable posture on the basis of a ZMP. Further, at each foot which touches the walking surface, there are provided a floor reaction force sensor and acceleration sensor to directly measure a ZMP and force, and a ZMP equation is formulated directly at the foot nearest to a ZMP position. Thus there can be implemented a stricter and quick control of the robot for a stable posture.