Coupled Temporal and Spatial Environment Monitoring for Multi-Agent Teams in Precision Farming
Jun Liu, Ryan K. Williams · 2020 IEEE Conference on Control Technology and Applications (CCTA) · 2020
In this paper, we propose a method for solving a multi-agent monitoring problem, both temporally and spatially. If a large and slowly evolving environmental process needs to be monitored by a multi-agent team, as is the case in our motivating precision farming application, we argue this team must make decisions on both sensing times and sensing locations. As the process evolution is slow, agents must make plans for future deployment times to gather as much information as possible (e.g., mutual information) within a reasonable budget. At the same time, the locations of where to collect information when the team is deployed are critical as different locations may possess widely varying information content. Multi-agent teams, therefore, need to make plans for future deployment locations as well as deployment times. In this paper, we combine these two sub-problems and model them as a submodular maximization problem with matroid and knapsack budgets. We propose a method with improved performance guarantees to solve our problem, as well as a broader class of combinatorial optimization problems. Finally, simulations are provided to demonstrate the effectiveness of the proposed method.