Robot Swing Leg Contact Detection for Stable Touchdown Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Robots face challenges in accurately detecting footstep contact with the ground surface, leading to potential destabilization and increased risk of tripping, as existing systems often fail to timely and accurately identify touchdowns, which can disrupt balance and cause flips or instability.
Innovation Solution
A method and system for footstep contact detection in robots, involving data processing hardware that receives joint dynamics and odometry data to determine unexpected torques on swing legs, classifying impacts as touchdowns or other causes, and generating responses such as changing leg classification or elevating the swing leg based on classified causes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing contact detection systems are used, then the robot can detect footstep contact, but the detection is not timely or accurate enough, leading to destabilization and tripping risks
Solution Approach 1:
The patent segments the contact detection process into multiple independent analysis components: torque analysis from joint dynamics, position analysis from odometry, velocity analysis, and classification modules. Each component processes specific aspects of leg movement separately, then integrates results to achieve comprehensive and accurate touchdown detection, resolving the contradiction between detection accuracy and system reliability.
Solution Approach 2:
The system implements continuous feedback loops where joint dynamics data and odometry data are constantly monitored, compared against expected values, and used to generate real-time corrections. The classification of impact causes feeds back into gait adjustment, allowing the robot to adapt its movement pattern based on detected contact events, thereby maintaining stability while improving detection accuracy.
2Measurement precision
If the robot responds to every detected impact, then contact detection sensitivity increases, but false positives from non-touchdown impacts (scuffing, self-contact) cause unnecessary gait disruptions
Solution Approach 1:
The patent applies different quality thresholds and analysis methods to different types of impacts. Touchdown impacts are detected with high sensitivity using torque and position analysis, while the system specifically filters out scuffing impacts (early swing phase contacts) and self-contact impacts (leg-leg or leg-body collisions) through classification rules. This localized differentiation allows high detection sensitivity without unnecessary gait disruptions from false positives.
Solution Approach 2:
The system changes detection parameters dynamically based on gait phase and impact characteristics. During swing phase, the system monitors for scuffing with specific torque thresholds; during stance phase, it monitors for touchdowns with different thresholds. The classification module adjusts response behavior based on impact parameters, allowing sensitive detection while maintaining gait smoothness by suppressing responses to non-critical impacts.
3Measurement precision
If the robot uses complex classification to distinguish touchdown from other impacts, then detection accuracy improves, but computational complexity and processing time increase
Solution Approach 1:
The classification process is segmented into distinct decision modules: torque magnitude analysis, swing phase percentage calculation, hip joint limit checking, and contralateral leg position verification. Each module handles a specific aspect of classification independently, making the complex overall process more manageable and computationally efficient while maintaining high classification accuracy.
Solution Approach 2:
The system performs preliminary calculations and checks before final classification. It pre-calculates swing phase percentage, pre-identifies hip joint limits, and pre-maps expected touchdown positions based on odometry. These preliminary actions prepare the data structure and thresholds in advance, reducing real-time computational burden while maintaining accurate classification of impact causes.
Data Source
AI summary
A method of footstep contact detection includes receiving joint dynamics for a swing leg of the robot where the swing leg performs a swing phase of a gait of the robot. The method also includes receiving odometry defining an estimation of a pose of the robot and determining whether an unexpected torque on the swing leg corresponds to an impact on the swing leg. When the unexpected torque corresponds to the impact, the method further includes determining whether the impact is indicative of a touchdown of the swing leg on a ground surface based on the odometry and the joint dynamics. When the impact is not indicative of the touchdown of the swing leg, the method includes classifying a cause of the impact based on the odometry of the robot and the joint dynamics of the swing leg.