HAC-FRL: A learning-driven distributed task allocation framework for large-scale warehouse automation

Social sensing has emerged as a new sensing application paradigm where measurements about the physical world are collected from humans or devices on their behalf. A key limitation in the current social sensing solution space is that data processing and analytics are often done in a "backend" mode (e.g., on dedicated servers or commercial cloud platforms). Such mode ignores the rich processing capability of increasingly powerful edge devices (e.g., mobile phones and nodes in the Internet of Things). Exploiting such edge devices in the social sensing setting introduces new challenges to real-time resource management. In this work, we develop a Bottom-up Game-theoretic Task Allocation (BGTA) framework to solve the critical problem of allocating real-time social sensing tasks to self-aware and non-cooperative edge computing nodes. In particular, we address two important challenges in solving this problem. The first one is "conflicting interest" where the objectives of applications and edge nodes may be at odds with each other. The second challenge is "asymmetric and incomplete information" where the application is often unaware of the detailed status (e.g., energy profile, utilization, CPU frequency) and compliance level of the edge nodes. To address these challenges, we first design a non-cooperative task allocation game model to address the conflicting objectives of the applications and edge nodes. We then develop a decentralized Fictitious Play scheme to allow each edge node to make its own decision on which task to execute in a non-cooperative context. Finally, we design a dynamic incentive mechanism to ensure the decisions made by the edge nodes meet objectives of the application. We implement a system prototype deployed on Nvidia Jetson TX1 and Jetson TK1 boards and evaluate our task allocation framework using two real-world social sensing applications. The results show that our scheme can well satisfy Quality of Service (QoS) requirement of the applications while providing optimized payoffs to edge nodes compared to the state-of-the-art baselines.

In the realm of sociotechnical systems, Level of Automation (LoA) frameworks are commonly used to determine adequate types of automation support for tasks in which human operators are involved. This paper introduces the General Automation Level Allocation (GALA) framework in response to recognized limitations in existing LoA frameworks. While these frameworks have contributed significantly to the formalization of human-automation interaction for the systems they were designed for, they often struggle when dealing with new sociotechnical systems. Some of the main limitations recognized for existing LoA frameworks include: (1) Lack of versatility in terms of missing levels for some “automated functions”, since they are designed with specific systems in mind; (2) Limited precision in the definition of the categories for assigning LoA to specific functions and complex technologies; (3) Limited support in the identification of outcomes of human-automation Interaction at different LoA (e.g. in terms of emerging behaviors or in terms of safety-related implications); (4) Limitations regarding characterization human cognitive processing in off-nominal or complex conditions; (5) Not fully addressing the dynamic allocation of tasks and responsibilities based on changing conditions and real-time priorities. Because of these limitations, some researchers are not satisfied with existing LoA taxonomies and believe that there is even no need to think deeper about LoA taxonomies as basis for or input to design of complex sociotechnical systems.To address the stated issues, GALA offers a two-dimensional approach aiming at being compatible with other previous LoA frameworks and applicable to the design of future systems. It is designed to analyze and classify the appropriate levels of automation for different information processing stages (e.g. information acquisition, information analysis, decision making, action implementation) involved in a task based upon the results of a hierarchical task analysis. GALA is compatible to established state-of-the art methods (Parasuraman, Sheridan & Wickens, 2000; Save, Feuerberg & Avia, 2012; Kaber, 2018) applied to study specific aspects of human-system collaboration in more depth, such as the coactive design method (Johnson et al., 2014). Finally, plans for GALA validation will be presented aiming to ensure that it provides sufficient applicability to various sociotechnical systems of diverse domains, each with its unique requirements and challenges. Further, an outlook on an alternative more compact version of the framework is provided which addresses the specific needs of dynamic task allocation in real-life situation.

Fundamental problem in human-robot teams is to find a set of heterogeneous robots that have to cooperate to execute a complex mission. This paper describes the Shared Knowledge Interaction Modelling (SKIM) framework for task allocation and how it is used to: 1) evaluate the performance of finding a set of robots to execute a certain task, and 2) model shared knowledge as a basis for adaptive autonomy in mixed human-robot teams. The shared knowledge is described by means of two ontologies: SKIM Resource Ontology (SKIM-RO) and SKIM Coordination Ontology (SKIM-CO). SKIM-RO describes resources, including robot capabilities and task requirements, and SKIM-CO describes coordination constraints for robot-robot and robot-human interactions. SKIM-CO is a basis for reasoning which enables the task allocation, and captures the concept of adaptive autonomy. This paper illustrates how framework is used to model and evaluate performance of task allocation in three use cases with the different level of task complexity. The results indicate how adaptive autonomy and shared knowledge improve performance in task allocation in complex missions.

To address interoperability and scalability issues for cloud computing, in our previous paper, we presented a novel cloud market model called CACM that enables a dynamic collaboration (DC) platform among different Cloud providers. As the initiator of dynamic collaboration, primary Cloud provider (pCP) needs an efficient local task selection and allocation algorithm to partition the whole tasks and allocate those tasks to be executed locally. Existing task allocation algorithms cannot be directly applicable in a DC environment since they may cause low resource utilization of local resources. So in this paper we propose a general task selection and allocation framework to improve resource utilization for pCP. The framework utilizes an adaptive filter to select tasks and a modified heuristic algorithm to allocate tasks. Moreover, a trade-off metric is developed as the optimization goal of heuristic algorithm, so that it is able to manage and optimize the trade-off between QoS of tasks and utilization of resources.

The Global Software Development (GSD) promises high-quality software at low cost. It enables round-the-clock development to achieve maximum production in a short period by utilizing expertise around the globe. GSD is only possible if tasks are effectively distributed among sites to ensure smooth development. Therefore, one of the key challenges of GSD is designing a task allocation (TA) strategy. The main objective of the present research is to develop a framework that takes into account important factors, while allocating tasks to distributed sites involved in GSD. The current allocation in plan-based software development is done on ad-hoc basis and does not follow any systematic approach or framework. The framework facilitates decision-makers in allocation of tasks in a manner that controls delay and re-allocation. The study uses a mixed method approach, where the data used to create the framework is acquired via an industrial survey (58 participants) and interviews (10 participants) with GSD practitioners. The developed task allocation framework is validated with the help of an online focus group with participants (7 participants) from around the globe. The ability of the framework to be applicable in real-world scenarios is assessed from the feedback of industry practitioners. They have highlighted the usefulness of the framework to both, practitioners involved in task allocation decision as well as researchers working in the area. The automation and validation of the framework in real-world GSD scenarios is part of future work of this research.

In this article, we study a problem of dynamic task allocation with multiple agent responsibilities in distributed multi-agent systems. Agents in the research have two responsibilities, communication and task execution. Movements in agent task execution bring changes to the system network structure, which will affect the communication. Thus, agents need to be autonomous on communication network reconstruction for good performance on task execution. First, we analyze the relationships between the two responsibilities of agents. Then, we design a multi-responsibility–oriented coalition formation framework for dynamic task allocation with two parts, namely, task execution and self-adaptation communication. For the former part, we integrate our formerly proposed algorithm in the framework for task execution coalition formation. For the latter part, we develop a constrained Bayesian overlapping coalition game model to formulate the communication network. A task-allocation efficiency–oriented communication coalition utility function is defined to optimize a coalition structure for the constrained Bayesian overlapping coalition game model. Considering the geographical location dependence between the two responsibilities, we define constrained agent strategies to map agent strategies to potential location choices. Based on the abovementioned design, we propose a distributed location pruning self-adaptive algorithm for the constrained Bayesian overlapping coalition formation. Finally, we test the performance of our framework, multi-responsibility–oriented coalition formation framework, with simulation experiments. Experimental results demonstrate that the multi-responsibility oriented coalition formation framework performs better than the other two distributed algorithms on task completion rate (by over 9.4% and over 65% on average, respectively).

The just-in-time concept, mass customization, omnichannel distribution, and the rising global population have all fueled the logistics sector. Consequently, using automation inside the warehouses to make them more dynamic and sustainable for the future is one of the crucial components to adapt to this quick shift. Giants in the industry and technology are becoming more interested in the “smart warehouse” system, built with innovative warehousing technologies, as an achievable solution for the development of warehouses in the future. To contextualize the past and provide light on prospective future directions, a study of current articles in the literature is important. This study evaluates works published in the previous 32 years related to flexible automation in warehouses to create a framework that future academics might use to guide them in establishing an original conceptual model that might be implemented at warehouses. One hundred eleven selected, examined, and categorized publications were published between 1990 and 2022 to establish a significant foundation for earlier research. Results indicated that combining automated machinery, collecting data technologies, and management systems are essential to creating a flexible automated warehouse. Finally, based on the examined literature, a flexible automated warehouse framework is suggested to direct future research in this field.

This paper presents an automatic task allocation framework for multi-robot systems (MRS) based on automaton parallel decomposition techniques. Given a synthesized global task automaton for a MRS, an iterative parallel decomposition framework is developed by decomposing this automaton projecting this automaton into a set of parallel decomposable event sets; thus into a set of smallest parallel subtask automata. Furthermore, an enhanced parallel decomposition strategy is presented by extracting the strictly decomposable automaton from a more general task automaton. Next, a task allocation automaton is synthesized for each subtask automaton to determine the robot assignment to tasks in a consecutive way. Through parallel executions of all these subtask allocation automata, a parallel task allocation automaton is obtained, which guarantees the completeness of the solution while reducing the search space. An optimal task allocation solution can be found from this parallel task allocation automaton by taking into account both concurrency and costs of multi-robot tasking. After the task allocation, symbolic motion planning (SMP) is performed for each individual robot. When intermittent communications exist among neighboring robots, task redecomposition and reallocation are triggered to update the optimal task allocation and SMP. This process continues until all the events in each subtask automaton are completed. The overall strategy is demonstrated by a simulation.

ContextGlobal software development (GSD) promises high‐quality software at low cost. GSD enables around‐the‐clock development to achieve maximum production in a short period of time by using expertise around the globe. This development is only possible if tasks are effectively distributed among sites to ensure smooth development. Therefore, one of the key challenges of GSD is to design a task allocation strategy.ObjectiveThe objective of this study is to identify various factors that influence task allocation decisions in GSD and to assess their relative importance. We also aim to determine the interrelationship between the factors along with role played by product architecture and communication and coordination needs during task allocation.MethodsWe used multiple methods to collect data about the task allocation factors and process. A web‐based survey of 54 GSD practitioners from around the globe was conducted to identify the factors and their relative importance for task allocation decision. The selection of the sample was performed via the snowball sampling technique. To increase the sample size, the survey was also posted on social media, that is, Facebook, LinkedIn, and Twitter. Nonparametric statistical tests were applied on the response data to identify correlations and significance. Interviews were conducted from 11 project managers having 10 to 30 years GSD experience to gain insight into the dynamics of task allocation process.ResultsThe survey results highlight “expertise,” “site characteristics,” and “task site dependency” as the most important factors for a task allocation decision. The interview study has highlighted the importance of situation‐specific decision making during task allocation. The significance of factors varies with the characteristics of task, characteristics of organization, type of GSD, and objective of doing GSD. The culture and time differences between distributed sites have been assigned a low priority by the majority of the practitioners. The most common way of distributing task is functional area of expertise and phase‐based division, where detailed architecture is not considered. Interdependent modules are not allocated to distributed sites because of communication and coordination overhead. Our results also demonstrate a correlation between various factors and support Conway's law.ConclusionsWe have interesting results in which certain factors are ranked differently from the prevalent views in the GSD literature. The survey results have also confirmed the application of Conway's law in practice for task allocation, where interdependent modules are not allocated to distributed sites. The significance of factors varies with characteristics of task, characteristics of organization, type of GSD, and objective of GSD, which require trade‐off between factors. The need of a well‐defined situation‐specific task allocation framework is evident from the results of survey and interview study. The outline of a task allocation framework for GSD is presented.

From the last two decades, Software agents are playing an important role in the field of Artificial intelligence and the Distributed Problem Solving. The properties of software agents like autonomy, reactivity, pro-activity and their social ability make them more center of focus for the real world problems. The accomplishment of any complex task is required to be done by the agents autonomously without any user intervention in order to achieve high reliability and adaptability. This paper mainly concentrates on the task allocation problem in multi-agent systems. Task Allocation is an important and challenging problem. This can be defined as the problem of allocating tasks among agents within a multi-agent system. Main objective of the task allocation problem is to maximize the number of successfully completed task and overall system utility without any conflict. To accomplish any complex task, agents negotiate, cooperate and coordinate with each other. Many researchers are working in the field of task allocation in multiagent systems. In this paper, various approaches of task allocation in multi-agent systems are discussed. The comparison of these approaches is also providing that lead to a discussion and motivation of the task allocation problem for multi-agent systems. This paper presents a hybrid approach for task allocation in dynamic multi-agent systems. The conclusion drawn from this survey is that, for dynamic multi-agent systems, the distributed task allocation is a better approach.

This paper presents a human-robot trust integrated task allocation and motion planning framework for multi-robot systems (MRS) in performing a set of parallel subtasks. Parallel subtask specifications are conjuncted with MRS to synthesize a task allocation automaton. Each transition of the task allocation automaton is associated with the total trust value of human in corresponding robots. A dynamic Bayesian network (DBN) based human-robot trust model is constructed considering individual robot performance, safety coefficient, human cognitive workload and overall evaluation of task allocation. Hence, a task allocation path with maximum encoded human-robot trust can be searched based on the current trust value of each robot in the task allocation automaton. Symbolic motion planning (SMP) is implemented for each robot after they obtain the sequence of actions. The task allocation path can be intermittently updated with this DBN based trust model. The overall strategy is demonstrated by a simulation with 5 robots and 3 parallel subtask automata.

In social insects, workers perform a multitude of tasks, such as foraging, nest construction, and brood rearing, without central control of how work is allocated among individuals. It has been suggested that workers choose a task by responding to stimuli gathered from the environment. Response-threshold models assume that individuals in a colony vary in the stimulus intensity (response threshold) at which they begin to perform the corresponding task. Here we highlight the limitations of these models with respect to colony performance in task allocation. First, we show with analysis and quantitative simulations that the deterministic response-threshold model constrains the workers' behavioral flexibility under some stimulus conditions. Next, we show that the probabilistic response-threshold model fails to explain precise colony responses to varying stimuli. Both of these limitations would be detrimental to colony performance when dynamic and precise task allocation is needed. To address these problems, we propose extensions of the response-threshold model by adding variables that weigh stimuli. We test the extended response-threshold model in a foraging scenario and show in simulations that it results in an efficient task allocation. Finally, we show that response-threshold models can be formulated as artificial neural networks, which consequently provide a comprehensive framework for modeling task allocation in social insects.

We present novel techniques for simultaneous task allocation and planning in multi-robot systems operating under uncertainty. By performing task allocation and planning simultaneously, allocations are informed by individual robot behaviour, creating more efficient team behaviour. We go beyond existing work by planning for task reallocation across the team given a model of partial task satisfaction under potential robot failures and uncertain action outcomes. We model the problem using Markov decision processes, with tasks encoded in co-safe linear temporal logic, and optimise for the expected number of tasks completed by the team. To avoid the inherent complexity of joint models, we propose an alternative model that simultaneously considers task allocation and planning, but in a sequential fashion. We then build a joint policy from the sequential policy obtained from our model, thus allowing for concurrent policy execution. Furthermore, to enable adaptation in the case of robot failures, we consider replanning from failure states and propose an approach to preemptively replan in an anytime fashion, replanning for more probable failure states first. Our method also allows us to quantify the performance of the team by providing an analysis of properties, such as the expected number of completed tasks under concurrent policy execution. We implement and extensively evaluate our approach on a range of scenarios. We compare its performance to a state-of-the-art baseline in decoupled task allocation and planning: sequential single-item auctions. Our approach outperforms the baseline in terms of computation time and the number of times replanning is required on robot failure.

The complexity of urban combat environments, the coupling of task allocation and path planning, and the existence of dynamic targets significantly increased the complexity of coordinated UAV swarm attack tasks. In this work, we proposed a multi-UAV task allocation and path planning framework inspired by the collaborative hunting behaviour of wolf packs. This framework was based on a multi-target k-winner-take-all (k-WTA) algorithm and an improved grey wolf optimization (GWO) algorithm. Firstly, for the multi-task allocation problem in unknown environments, a multi-objective k-WTA algorithm was used for task allocation based on the competition mechanism, which realized fast task allocation in dynamic environments. Then, the advantages of GWO and genetic algorithm (GA) were combined through GA-WPO to overcome the random initialization problem of GWO by using GA as an initialization generator. A path planner based on GA-WPO was proposed to enable multi-UAVs to reach the target point safely in complex urban environments. Finally, the proposed path planner was used for multi-UAV path planning, and the effectiveness of the method was verified by a set of simulation experiments, which showed that the method better solved the coupling problem of path planning and task assignment for UAV swarm coordinated attack tasks.

Task allocation is an important issue in multi-agent systems, and finding the optimal solution of task allocation has been demonstrated to be an NP-hard problem. In many scenarios, agents are equipped with not only communication resources but also computing resources, so that tasks can be allocated and executed more efficiently in a distributed and parallel manner. Presently, many methods have been proposed for distributed task allocation in multi-agent systems. Most of them are either based on complete/full search or local search, and the former usually can find the optimal solutions but requires high computational cost and communication cost; the latter is usually more efficient but could not guarantee the solution quality. Evolutionary algorithm (EA) is a promising optimization algorithm which could be more efficient than the full search algorithms and might have better search ability than the local search algorithms, but it is rarely applied to distributed task allocation in multi-agent systems. In this paper, we propose a distributed task allocation method based on EA. We choose the many-objective EA called NSGA-III to optimize four objectives (i.e., maximizing the number of successfully allocated and executed tasks, maximizing the gain by executing tasks, minimizing the resource cost, and minimizing the time cost) simultaneously. Experimental results show the effectiveness of the proposed method, and compared with the full search strategy, the proposed method could solve task allocation problems with more agents and tasks.