Reward optimization of spatial crowdsourcing for coalition-based maintenance task.

Spatial crowdsourcing (SC) can use the moving workers to achieve location-based tasks. It has been widely used in takeout, data labeling, organizing activities, security and artificial intelligence. Currently, many manufacturers wish to explore SC in maintenance business, as SC can reduce maintenance time, reduce labor costs, and improve customer satisfaction. In maintenance business scenarios, there are two kinds of cooperated participators. One is freedom workers who are sensitive to distance, and the other is employed by manufacturers who are insensitive to distance. Therefore, some methods of SC with one type worker are challenging to apply to maintenance business scenarios directly. To match this scenario, we model the maintenance scenario and prove that this scenario is an NP-hard problem; then, both greedy and Nash equilibrium methods are proposed to complete the tasks for making a high total reward. The greedy algorithm (GA) first assigns the nearest available workers to each task, and the employee will try to join the task to help GM to complete more tasks, on the condition that the task cannot be completed and reaches a particular proportion. The Nash equilibrium algorithm (NA) is used to find a Nash equilibrium for all the workers and employees. The experiments demonstrate the efficiency and effectiveness of the synthetic data set of gMission in small and large data sets. The finished task number of NA is about 5% more than that of GA, and the reward of NA is about 10% more than that of GA.