A Survey on Task Allocation and Scheduling in Robotic Network Systems

TL;DR

This survey addresses the allocation and scheduling problem in robotic network systems spanning cloud, fog, and edge infrastructures. It clusters existing work into three mathematical paradigms—optimization, combinatorial methods, and reinforcement learning—and further differentiates studies by static versus dynamic task allocation and cloud involvement. The authors highlight strengths and limitations across these categories, including scalability, real-time adaptability, and integration across computing layers, and they emphasize the growing role of hybrid cloud-edge solutions. Key takeaways include the effectiveness of combinatorial techniques for certain problem classes, the rising prominence of RL in dynamic environments, and the need for unified frameworks that balance latency, energy, and resource utilization in realistic, noisy settings.

Abstract

Cloud Robotics is helping to create a new generation of robots that leverage the nearly unlimited resources of large data centers (i.e., the cloud), overcoming the limitations imposed by on-board resources. Different processing power, capabilities, resource sizes, energy consumption, and so forth, make scheduling and task allocation critical components. The basic idea of task allocation and scheduling is to optimize performance by minimizing completion time, energy consumption, delays between two consecutive tasks, along with others, and maximizing resource utilization, number of completed tasks in a given time interval, and suchlike. In the past, several works have addressed various aspects of task allocation and scheduling. In this paper, we provide a comprehensive overview of task allocation and scheduling strategies and related metrics suitable for robotic network cloud systems. We discuss the issues related to allocation and scheduling methods and the limitations that need to be overcome. The literature review is organized according to three different viewpoints: Architectures and Applications, Methods and Parameters. In addition, the limitations of each method are highlighted for future research.

Figures (15)

• Figure 1: In static task allocation, we determine the set of all algorithms required to perform any task for which the system is designed. The goal is to determine which unit should be assigned to each algorithm for execution so that each individual task can be optimally executed by each unit. $\mathbf{A}$ is the set of all algorithms required by the system to perform all the tasks for which the system is designed, and each of the tasks is denoted by a DAG with precedence order. $A_i$ denotes a single algorithm.
• Figure 2: In dynamic task allocation (task scheduling), a set of tasks arrives to the system at time $t$ to be executed. The goal is to determine to which unit each task should be assigned for execution in order to optimally complete all requested tasks. $\mathbf{Task_t}$ is the set of tasks that have arrived to the system at time $t$, and each of the newly arrived tasks in $\mathbf{Task_t}$ is denoted by a DAG with precedence order. $A_i$ and $B_i$ are used to denote algorithms of tasks.
• Figure 3: Overview of approaches to task allocation and scheduling. We exemplify each with two references. $T$ is a finite set of tasks, $\mathbf{A}$ is the set of all algorithms that the system needs to perform all tasks in $T$ for which the system is designed. All tasks in $T$ can be completed by executing some algorithms from $\mathbf{A}$ with a certain order of precedence. $T_i\subseteq T$ is the set of tasks arriving in the system at time step $t_i$, where $t_1 < t_2 < \ldots$ is the time at which a new set of tasks arrives in the system.
• Figure 4: Different categorization of all contributions.
• Figure 5: Environment abstraction. The imaginary vertical dashed lines are some borders of regions, $v_1$ to $v_5$. The filled object in the middle of the environment is the obstacle.