Recurrent Schedule Propagation for Human-Robot Task Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing task allocation and scheduling systems for human-robot teams are inefficient due to their reliance on time and computationally resource-intensive search-based algorithms and the need for sequential interaction with the environment, which limits their scalability and speed.
Innovation Solution
A deep learning-based system utilizing a heterogeneous graph-based encoder and a recurrent schedule propagator to generate task-agent assignments efficiently, allowing for fast schedule generation without the need for continuous environmental interaction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If search-based algorithms are used for task allocation and scheduling, then exact results can be provided, but the system becomes time and computationally resource-intensive
Solution Approach 1:
The patent replaces traditional search-based mechanical algorithms with a neural network-based learning system. The neural network is trained offline to learn optimal scheduling patterns, then rapidly generates schedules during operation without requiring intensive search computations, thus maintaining accuracy while reducing computation time
Solution Approach 2:
The system performs preliminary training offline where the neural network learns from extensive scheduling data and search-based solutions beforehand. This preliminary learning enables the model to make accurate scheduling decisions rapidly during actual operation without repeating the computationally intensive search process
2Productivity
If learning-based models utilizing Graph Neural Networks are used, then task allocation can be improved, but sequential interaction with the environment is required after each decision
Solution Approach 1:
The patent creates a simplified virtual environment that copies the essential characteristics of the real environment. The neural network learns to operate within this virtual copy, which allows for rapid scheduling decisions without requiring repeated interactions with the actual physical environment, thus reducing time loss while maintaining allocation efficiency
3Loss of energy
If heuristics are used for task allocation, then computational resources are reduced, but subject-matter experts must account for the problem
Solution Approach 1:
The system employs self-service through automated neural network training and operation. The model automatically learns optimal scheduling strategies from data without requiring continuous expert intervention or hand-crafted heuristics. Once trained, the system autonomously handles task allocation, reducing both computational resource usage and dependence on expert knowledge
Data Source
AI summary
An exemplary deep learning-based system and method are disclosed for human-robot coordination under temporal constraints that has a Heterogeneous Graph-based encoder and a Recurrent Schedule Propagator. The encoder extracts relevant information about the initial environment, while the Propagator generates the consequential models of each task-agent assignments based on the initial model. Inspired by the sensory encoding and recurrent processing of the brain, the approach allows for fast schedule generation, removing the need to interact with the environment between every task-agent pair selection.