Robot Arm Fluid Transfer Control Using Posture and Weight Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning methods for controlling robot arms to transfer fluids lack efficiency and accuracy, particularly in configuring systems that do not require image information, leading to complex setups and reduced performance.
Innovation Solution
A method and system that generate a learning model using time-series information about the robot arm's posture and the weight of a container, eliminating the need for image data and enabling automation of fluid transfer with improved accuracy and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If conventional machine learning methods are used for controlling robot arms to transfer fluids, then the system can perform automated fluid transfer, but the system configuration becomes complex and processing time increases due to requirement of image data
Solution Approach 1:
The patent extracts and eliminates the image data input requirement from the machine learning system. By using only numerical data (robot arm posture and container weight) instead of image data, the system removes the optical imaging system and related processing components, significantly simplifying the system configuration while maintaining automated fluid transfer capability
Solution Approach 2:
The patent replaces the optical imaging system (mechanical/optical hardware) with a purely data-based input system. Instead of using cameras and image processing, the system directly uses numerical measurements of robot arm posture and container weight, substituting physical imaging infrastructure with computational data inputs
2Loss of information
If conventional machine learning methods requiring image data are used, then the system can capture visual information about the first container, but the processing time increases and system complexity increases
Solution Approach 1:
The patent substitutes optical imaging and image processing with direct numerical data input. By using sensor data for robot arm posture and weight sensor data for container weight, the system eliminates the time-consuming processes of image capture, processing, and interpretation, achieving faster processing while still capturing all necessary information for fluid transfer control
3Loss of information
If image data is used as input for machine learning in fluid transfer control, then the system can obtain visual information about container state, but the system configuration becomes more complex and requires optical imaging systems
Solution Approach 1:
The patent extracts only the essential numerical parameters (robot arm posture and container weight) needed for fluid transfer control, eliminating the need for optical imaging systems. This selective data extraction approach maintains all necessary information about container state while removing complex optical hardware and image processing requirements
Solution Approach 2:
The patent changes the input parameters from image data (spatial, visual information) to numerical data (posture angles and weight measurements). This parameter transformation simplifies the data structure, reduces computational complexity, and eliminates the need for optical imaging systems while preserving the essential information needed for control
Data Source
AI summary
A system for controlling a robot arm, a fluid contained in a container is poured into another container. A learning model is generated by a machine learning with teaching data. Practically, a plurality of sets of learning data are acquired, each set including i) time-series information showing a posture of a robot arm which holds a first container holding therein a target fluid and pouring the target fluid from the first container to a second container and ii) a weight of the second container which changes time serially. This learning model is used such that only two types of information consisting of the information showing the posture of the robot arm and the weight of the second container at a first time are inputted to the learning model and information showing the posture of the robot arm at a second time is outputted from the learning model.