Robot Transition Control Between AI Track Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI-controlled robots exhibit unstable behavior when switching between different machine learning models due to discontinuities in physical quantities like position, speed, and acceleration, leading to potential delays in operation.
Innovation Solution
A control device for robots that includes base track information and transition section information, using processors to determine whether the actual or predicted track is within a preset error range, and employing specific transition control methods to stabilize the switch between AI models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If different machine learning models are used in connected tracks, then the robot can perform diverse tasks with optimized performance for each track section, but the robot behavior becomes unstable when switching between models at connection points
Solution Approach 1:
A transition section is introduced as an intermediary between two base tracks with different machine learning models. This transition section contains a predetermined control method that acts as a mediator to smoothly connect the two different AI models, ensuring stable robot behavior during the switch while maintaining the ability to use optimized models for specific tasks.
2Reliability
If the robot temporarily stops at the connection point to settle before transitioning to the next track, then the endpoint condition matches the start point condition, but the operation is delayed
Solution Approach 1:
The transition section is predetermined and prepared in advance with a specific control method designed to ensure smooth connection between tracks. This preliminary preparation allows the robot to transition directly without temporary stopping, as the transition section already contains the necessary control logic to match endpoint conditions with start point conditions of the next track.
3Device complexity
If AI control is used without guaranteeing all physical quantities, then the control system is simpler and more flexible, but the endpoint condition may not match the start point condition of the connected next track
Solution Approach 1:
The track is segmented into base tracks and transition sections. Base tracks use flexible AI control for simplicity and adaptability, while transition sections are specifically designed to ensure track continuity and match conditions between connected tracks. This segmentation allows AI control to remain simple in most sections while ensuring reliability at critical connection points.
Data Source
AI summary
A first transition section is set in a first base track and includes a connection point between the first base track and a second base track. A first machine learning model and a second machine learning model are set for the first base track and the second base track, respectively. When the actual track or the predicted track at the predetermined point is within the error range, the control device controls the automatic machine using the first machine learning model set for the first base track in the transition section. When the actual track or the predicted track at the predetermined point is outside the error range, the control device controls the automatic machine in the transition section by a control method different from that of the first machine learning model.