No abstract

No abstract

In this paper, a distributed algorithm with obstacle avoidance capability is presented to deploy a group of ground robots for field-based agriculture applications. To this end, the field (consisting of many plots) is first modeled as a directed graph, and the robots are deployed to collect data from some important areas of the field (e.g., areas with high water stress or biotic stress). The key idea is to formulate the underlying problem as a locational optimization problem and then find the optimal solution based on the Voronoi partitioning of the associated graph. The proposed partitioning method is validated through simulation studies, as well as experiments using a group of mobile robots.

In this work, we develop a system that can be used for real-time monitoring of multiple important areas in controlled environment agriculture (and in particular greenhouses) using an autonomous ground vehicle (AGV). To model the greenhouse layout, as well as the tasks that should be accomplished by the AGV, we generate two weighted directed graphs. Based on those graphs, an algorithm is then proposed for finding the optimal (in the sense of traveled distance) trajectory of the vehicle with the goal of precisely monitoring important areas in the greenhouse. Furthermore, a data collection system and image processing algorithm is proposed and implemented so that the vehicle: (i) can capture images and detect changes that have occurred on the crops in real time, and (ii) construct (if needed) a map of the plant rows, when arriving at each one of the important areas. Based on this work, the images can either be stitched onboard the vehicle and then sent to a server or be sent directly to the server and then processed (stitched) there. Both simulation and experimental results are provided to demonstrate the effectiveness and performance of the proposed system.

In this work, we investigate the problem of finding the minimum coverage time of an agricultural field using a team of heterogeneous unmanned aerial vehicles (UAVs). The aerial robotic system is assumed to be heterogeneous in terms of the equipped cameras’ field of view, flight speed, and battery capacity. The coverage problem is formulated as a vehicle routing problem (VRP) [1] with two significant extensions. First, the field is converted into a graph, including nodes and edges generated based on sweep direction and the minimum length of UAVs’ footprints. Second, the underlying optimization problem accounts for aerial vehicles having different sensor footprints. A series of simulation experiments are carried out to demonstrate that the proposed strategy can yield a satisfactory monitoring performance and offer promise to be used in practice.