Hypergraph MRTA Search for Real-Time Smart Factory Reallocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Multi-Robot Task Allocation (MRTA) systems in smart factories face challenges in dynamically optimizing throughput for production robots and energy efficiency for Autonomous Mobile Robots (AMRs) in real-time, especially due to the NP-hard nature of the dual-objective and dual-agent optimization problem.
Innovation Solution
The implementation of a hypergraph-based MRTA system that uses a nested hierarchy hypergraph search algorithm to generate real-time task assignments and transportation paths for production robots and AMRs, respectively, while optimizing productivity and energy efficiency with a tradeoff between solution optimality and time complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional LP and exhaustive search methods are used to solve MRTA, then global optimal solution is achieved, but time complexity increases dramatically leading to time-lag in dynamic operation
Solution Approach 1:
The patent segments the MRTA problem into two separate optimization objectives: production robot throughput optimization and AGV energy consumption optimization. This segmentation allows the system to solve two simpler problems independently rather than one complex dual-objective problem, reducing computational complexity while maintaining solution quality
Solution Approach 2:
The patent implements dynamic task allocation that adapts to changing production conditions in real-time. The system continuously re-evaluates and re-allocates tasks based on current system state, enabling responsive decision-making without requiring exhaustive search of all possible allocations
2Productivity
If dual-objective optimization is implemented for both production robot throughput and AGV energy consumption, then system performance is improved, but the problem becomes NP-hard increasing computational difficulty
Solution Approach 1:
The patent divides the dual-objective optimization into separate single-objective optimization problems: one for production robot throughput and another for AGV energy consumption. Each sub-problem can be solved independently using appropriate algorithms, avoiding the NP-hard complexity of simultaneous dual-objective optimization
Solution Approach 2:
The patent transforms the optimization problem by changing parameters and variables to simplify the mathematical formulation. By redefining decision variables and objective functions separately for each robot type, the system makes the problems tractable while still achieving dual optimization goals
3Adaptability or versatility
If frequent adjustment of production flow is made to respond to market requests, then production flexibility is improved, but computation time for task allocation increases
Solution Approach 1:
The patent implements a dynamic task allocation system that can rapidly re-compute optimal assignments when production conditions change. The segmented optimization approach enables quick recalculation of task assignments in response to new market requests or production changes
Solution Approach 2:
The system performs preliminary optimization by pre-calculating task allocation strategies and maintaining ready-to-execute assignment plans. When changes occur, the system can quickly adjust from pre-computed solutions rather than starting from scratch, reducing computation time for frequent adjustments
Data Source
AI summary
Task assignment for multi-robot systems (MRSs) in a smart factory (Industry 4.0) is described. Aspects are directed to a hypergraph based MRS and production model facilitating the cooperation among robots and serving frequent reconfiguration desired in Industry 4.0. Aspects are directed to a time complexity friendly search algorithm for real-time application using a hypergraph model to get task assignment(s). Parameters are provided for a tradeoff between solution optimality and time complexity. In an implementation, an example system can include a MRS including robots, wherein the MRS is configured to perform a manufacturing task, and a computing device configure to perform a multi-robot task allocation (MRTA) for the MRS. In an implementation, an example method can include generating task assignments, using MRTA, for robots of a MRS including the robots, wherein the MRS is configured to perform a manufacturing task, and providing the task assignments to the MRS.