Human-Robot Anomaly Detection for Adaptive Manufacturing Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning technologies face challenges in controlling robotic systems due to complex physical laws like Rigid Body Dynamics, limited sensor measurements, and the need for human-robot collaboration in manufacturing processes, where human variability affects product quality and speed.
Innovation Solution
Developing advanced model learning technologies that learn typical human worker performance, detect anomalies, and adjust robotic assistance based on human condition, using predictive and classification models to optimize human-robot interaction and manufacturing processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional machine learning technologies are used to control robotic systems, then the system can operate with basic sensor measurements, but the control precision and ability to handle complex physical laws deteriorates
Solution Approach 1:
The patent introduces an intermediary learning module that acts as a mediator between the control system and the complex physical laws. This module learns the relationship between robot states and control inputs through data-driven approaches, translating complex dynamics into learnable patterns without requiring explicit mathematical models of all physical constraints.
Solution Approach 2:
The patent replaces traditional mechanics-based control approaches with data-driven machine learning models. Instead of relying on explicit mechanical models and complex calculations of physical laws, the system uses learned representations from training data to predict optimal control actions, substituting mechanical reasoning with statistical learning.
2Loss of information
If sensors are added to measure velocity and acceleration, then the measurement completeness improves, but the device complexity and cost increases
Solution Approach 1:
The patent replaces physical sensors with computational methods for deriving velocity and acceleration. Instead of using additional hardware sensors, the system computes these derivatives from position data through numerical differentiation or learning-based estimation, substituting mechanical measurement with computational processing.
Solution Approach 2:
The patent introduces a computational intermediary layer that processes position measurements to infer velocity and acceleration information. This learning module acts as a mediator that transforms basic position data into comprehensive state information without requiring direct physical measurement of all state variables.
3Adaptability or versatility
If human variability is accommodated in the manufacturing process, then the adaptability to human workers improves, but the manufacturing speed and consistency deteriorates
Solution Approach 1:
The patent implements dynamic adaptability where the robot's control parameters are continuously adjusted based on real-time detection of human worker states. The system transitions from static pre-programmed sequences to dynamic adaptive control, modifying robot speed, position, and actions according to detected human fatigue, skill level, and performance variations.
Solution Approach 2:
The patent incorporates feedback loops where the robot continuously monitors human worker performance and adjusts its actions accordingly. Detection data about human state feeds back to the control system, which modifies robot behavior in real-time, creating a closed-loop adaptive system that balances human accommodation with manufacturing efficiency.
4Reliability
If the robot adjusts its actions to accommodate human worker condition, then the human-robot collaboration quality improves, but the manufacturing speed deteriorates
Solution Approach 1:
The patent dynamically changes robot operational parameters such as speed, acceleration, and position based on detected human worker conditions. When human fatigue or skill variations are detected, the system adjusts robot parameters to provide appropriate assistance or accommodation, optimizing the collaboration quality for each specific human state.
Solution Approach 2:
The patent implements dynamic parameter adjustment where robot speed and actions are not fixed but continuously adapted based on real-time human state detection. This dynamic approach allows the system to optimize collaboration quality by matching robot behavior to human capabilities while maintaining overall manufacturing productivity through intelligent parameter selection.
Data Source
AI summary
A system for detecting an anomaly in an execution of a task in mixed human-robot processes. Receiving human worker (HW) signals and robot signals. A processor to extract from the HW signals, task information, measurements relating to a state of the HW, and input into a Human Performance (HP) model, to obtain a state of the HW based on previously learned boundaries of the state of the HW, the state of the HW is then inputted into a Human-Robot Interaction (HRI) model, to determine a classification of an anomaly or no anomaly. Update HRI model with robot operation signals, HW signals and classified anomaly, determine a control action of a robot interacting with the HW or a type of an anomaly alarm using the updated HRI model and classified anomaly. Output the control action of the robot to change a robot action or output the type of the anomaly alarm.