Robotic Arm Insertion Control Using Force-Position Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current robotic assembly systems face challenges in accurately detecting anomalies during insertion processes due to uncertainties and complexities, leading to inefficiencies in error recovery and increased computational complexity in data-driven models.
Innovation Solution
A machine learning-based anomaly detection system using probabilistic functional models that reduce dimensionality from 12 or 18 dimensions to 2 dimensions by establishing a stable relationship between force and position, employing Gaussian process regression and local Gaussian process models for real-time anomaly detection and error recovery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data-driven models with high dimensionality (12 or 18 dimensions) are used for anomaly detection, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent extracts only the essential variables (force and position along the insertion line) from the full 12 or 18-dimensional measurement space. By identifying and removing redundant dimensions, the system maintains anomaly detection capability while significantly reducing computational complexity. This is achieved by focusing on the one-dimensional insertion line where force and position measurements are taken.
Solution Approach 2:
The patent transforms the high-dimensional anomaly detection problem into a two-dimensional problem by projecting measurements onto the insertion line dimension. This dimensionality reduction converts complex multi-dimensional force and position data into simple force-position relationships along a single line, making the model computationally tractable while preserving detection accuracy.
2Reliability
If model-based methods with parameter bounds are used, then reliability is improved, but device complexity increases
Solution Approach 1:
The patent replaces complex physics-based models with simple statistical models that estimate mean and standard deviation from data. These lightweight statistical models are computationally inexpensive and easy to implement, providing robust fault detection without requiring complex domain knowledge or extensive model development.
Solution Approach 2:
The system continuously monitors force and position measurements during insertion and compares them against statistically derived bounds. This feedback mechanism automatically adjusts to normal variations in the insertion process while detecting anomalies, providing reliable fault detection without complex predetermined models.
3Ease of operation
If threshold-based anomaly detection is used, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The patent replaces static thresholds with dynamic statistical bounds that are estimated from actual insertion data. The mean and standard deviation are continuously updated based on observed measurements, allowing the detection system to adapt to variations in the insertion process while maintaining simple operation. This dynamic approach significantly improves detection accuracy over fixed thresholds.
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
A system for controlling a robotic arm performing insertion of a component along an insertion line accepts measurements of force experienced by the wrist of robotic arm at current position along insertion line and determines probability of value of the force conditioned on the current value of the position according to a probabilistic relationship for the force experienced by the wrist of the robotic arm along the insertion line as a probabilistic function of the positions of the wrist of the robotic arm along the line of insertion. The probabilistic function is learned from measurements of the operation repeatedly performed by one or multiple robotic arms having the configuration of the robotic arm under the control. The system determines a result of anomaly detection based on the probability of the current value of the force and controls the robotic arm based on the result of anomaly detection.