Dynamic Task Allocation for Heterogeneous Agents in Disaster Environments Under Time, Space and Communication Constraints

Abstract

Task allocation for heterogeneous agents in disaster environments under time, space and communication constraints is a challenging issue in both theory and practice. This paper presents a dynamic task allocation approach for such situations. The proposed approach consists of an information collection mechanism, a group task allocation mechanism and a group coordination mechanism. Initially, the information collection mechanism is applied to help agents in communication networks to reduce their communication connections and select one agent in each network as the network leader in a decentralised manner so as to facilitate the collection of information for task allocation under communication constraints. Then, the group task allocation mechanism is employed by each network leader to allocate tasks and agents in its network to groups with suitable space ranges by considering time, space and communication constraints as well as the differing capabilities of agents. During task execution, due to the dynamics of disaster environments, the original allocation (by the group task allocation mechanism) of tasks and agents in groups may be unsuitable. In order to achieve continuous coordination of the heterogeneous agents among groups under communication constraints, the group coordination mechanism is employed. Experimental results demonstrate that the proposed approach can have better performance than many existing approaches in terms of information collection and dynamic task allocation in disaster environments under time, space and communication constraints.Keywords: Dynamic task allocation; Disaster environments; Time, space and communication constraints a task, it first needs to move to the location of the task [7,6,8]; 3) Communication constraints. In disaster environments, due to the destruction of local communication facilities, the amount of information transferred among agents (i.e., the constraint of communication capacities) and the communication distances of agents (i.e., the constraint of communication ranges) are limited [9, 10]. 4) Dynamics of environments. In disaster environments, agents can enter and leave the environments and tasks can be continuously discovered and finished in the environments [11,12]; 5) Differing capabilities of agents. In disaster environments, each agent has its own capabilities, which determine what kind of tasks it can conduct [7]. 6) Local views of agents. Due to the communication constraints of disaster environments, each agent can only acquire the information about tasks close to its location and the information of agents with which it can directly communicate [13];Various approaches have been proposed to achieve efficient task allocation in disaster environments [14,11,15]. Some approaches deal with task allocation through a central point [7,6,8]. In such approaches, the central controller (i.e., the agent in charge of task allocation) can create an optimal solution for task allocation by considering time and space constraints as well as the differing capabilities of agents ...

Searching and Rescuing Victims in Emergency: A Comprehensive Survey

“…the distance between them), the probability that a rescuer fails to save a victim, and the associated penalty that the rescuer fails to rescue the victim are taken into consideration, and are transformed into a goal function which reflects the trade-off between the total successful rescue costs and the failure penalties; the RNN is utilised as a fast optimisation algorithm to minimise the goal function with a gradient descent learning procedure; in the RNN, each possible rescuer-victim pair is considered as a neuron, and the neuron with the highest excitation probability after the training process is selected as the decision. The study in [19] proposes a task allocation algorithm for heterogeneous agents in hazard environments with respect to time, space and communication restrictions. As aforementioned in section 2.2, the work in [7] employs an ordinary differential equation model to redistribute rescue robots among various tasks in a decentralised manner without inter-robot communications.…”

Emergency Management Systems and Algorithms: a Comprehensive Survey

“…the distance between the rescuer and the victim), the probability that a rescuer fails to save a victim, and the associated penalty that the rescuer fails to rescue the victim are taken into consideration, and are transformed into a goal function which reflects the trade-off between the total successful rescue costs and the failure penalties; the RNN is utilised as a fast optimisation algorithm to minimise the goal function with a gradient descent learning procedure; in the RNN, each possible rescuer-victim pair is considered as a neuron, and the neuron with the highest excitation probability after the training process is selected as the decision. The research in [186] proposes a task allocation algorithm for heterogeneous agents in hazard environments with respect to time, space and communication restrictions; to reduce the communication load and links among agents, each agent elects the agent with the most direct neighbours within its communication range as the communication network leader; the elected network leader will be responsible for maintaining the communication among intra-network agents and other network leaders; to ensure the appropriate distance among task locations and the agents, each network leader further divides its coverage area into sub-areas by using the mean shift clustering algorithm; the window radius parameter h in the mean shift clustering algorithm is set to the communication range of the network leader to guarantee that the agents in the same sub-area can always communicate with each other; heterogeneous agents are then allocated into these sub-areas in terms of their abilities and the requirements of tasks within the sub-areas: firstly, the "similarity" value of each agent sub-area pair is calculated by the dot product [187] of the vector of the requirements of tasks and the normalised vector of ability of the agent; secondly, the agent will be assigned to the sub-area with the highest similarity value. As aforementioned in subsection 3.1.2, the studies in [167,168] employ an ordinary differential equation model to redistribute rescue robots among various tasks in a decentralised manner without inter-robot communications; the relation between population fractions at tasks and the elapsed time since the start of the process is expressed by a set of ordinary differential equations; by substituting the desired population fraction in each task into the proposed model, each robot can independently calculate the time point when the system reaches the desired distribution.…”