A Task Allocation Framework for Human Multi-Robot Collaborative Settings (2210.14036v1)

Abstract: The requirements of modern production systems together with more advanced robotic technologies have fostered the integration of teams comprising humans and autonomous robots. However, along with the potential benefits also comes the question of how to effectively handle these teams considering the different characteristics of the involved agents. For this reason, this paper presents a framework for task allocation in a human multi-robot collaborative scenario. The proposed solution combines an optimal offline allocation with an online reallocation strategy which accounts for inaccuracies of the offline plan and/or unforeseen events, human subjective preferences and cost of switching from one task to another so as to increase human satisfaction and team efficiency. Experiments are presented for the case of two manipulators cooperating with a human operator for performing a box filling task.

• The paper introduces an integrated framework that uses a MILP formulation for initial task allocation to optimize makespan and task quality.
• It details a dynamic online reallocation mechanism that adjusts allocations in real-time based on performance metrics and human preferences.
• Experimental results validate the approach, demonstrating improved efficiency and adaptability in human-multi-robot collaborative environments.

A Task Allocation Framework for Human Multi-Robot Collaborative Settings

Introduction

The paper introduces a task allocation framework designed for settings involving human and multi-robot collaboration. It addresses the complex task of distributing responsibilities between humans and autonomous robots in environments where these agents must work together to enhance productivity and efficiency. The proposed framework integrates optimal offline allocation strategies with dynamic online reallocation capabilities to adapt to evolving task parameters and human preferences.

Framework Components

Offline Task Allocation

The solution divides tasks into clusters based on similarity, where each task is characterized by its spatial location, execution time, switching costs, quality indices, and workload estimates. It introduces a Mixed Integer Linear Programming (MILP) formulation that optimizes the allocation of tasks to agents by minimizing the makespan and workload while maximizing task quality. The approach considers supervision responsibilities for human operators, ensuring that if a task allocation does not meet quality requirements, human supervision can provide the necessary adjustments.

Online Reallocation Strategy

In scenarios where task demands or agent capabilities change, the paper proposes an online reallocation mechanism to maintain optimal performance. Monitoring systems track performance metrics and human preferences, enabling re-planning when deviations from the initial plan occur or when human agents express new task preferences. This adaptive framework is critical for maintaining operational efficiency and human satisfaction in dynamic environments.

Experimental Setup and Results

Validation with Robotic Manipulators

The experimental setup includes two robotic manipulators (Kinova Jaco$^2$) and a human operator tasked with arranging grape bunches and paper cloths into a box. The tasks are organized into clusters that reflect varying requirements for precision and dexterity. A Graphical User Interface (GUI) assists the human operator, providing real-time feedback on task assignments and allowing for preference updates.

Numerical and Experimental Analysis

The framework's efficacy is demonstrated through a validation campaign, comparing scenarios with and without switching cost optimization. The results consistently show improvements in the makespan when switching costs are considered. The experiments reveal the adaptability of the system to human preferences, including scenarios where reallocation is necessary due to dynamic updates in task assignments.

Conclusion

The proposed task allocation framework successfully integrates offline planning with dynamic online adjustments to manage human-robot collaborations effectively. By leveraging MILP for initial planning and incorporating real-time monitoring for adaptability, the framework addresses both efficiency and flexibility. Future directions might include incorporating proactive behaviors through predictive human activity models to further enhance collaboration in mixed human-robot teams.