HuMachineSensing: A Novel Mobile Crowdsensing Framework with Robust Task Allocation Algorithm

Mobile crowdsensing (MCS) is a novel and innovative sensing model. Instead of traditional methods of sensing, MCS uses intelligent mobile devices carried by people to perform different sensing tasks. With the development of robotics and artificial intelligence technologies, many MCS studies with pervasive machines [1] (e.g., robots, smart vehicles, drones, etc.) as participants have emerged in recent years. Robot participants can perform dangerous, tedious tasks with a high degree of control. In contrast, human participants can perform various complex tasks flexibly with human intelligence. Therefore, this paper proposes a novel framework of MCS, i.e., HuMachineSensing, which combines human participant and robot participant, and investigates the task assignment problem of this system. The key to the combined human-robot task assignment problem is how to determine whether a task is assigned to a human participant or a robot participant. In this paper, we propose the concept of sense information map (SIM), which can reflect the performance of different participants performing sensing tasks at different locations and times, and provide guidance for task assignment. The SIM is a real-time optimized model that can become more accurate as participants perform tasks, as we propose a map-based Gaussian process algorithm to continuously update the map. To further improve the robustness of the system, we propose a self-repairing task assignment algorithm, which can realize the self-repairing and reassignment of the task assignment scheme after the participants encounter abnormal situations. Through experiments, it is demonstrated that the sensing information graph and the map-based self-repairing task assignment algorithm can effectively improve the task coverage.