A framework for concurrent design and route planning optimization of unmanned aerial vehicle based urban delivery systems
MetadataShow full item record
With the emergence of new technologies for small Unmanned Aircraft Systems (sUAS), such as lightweight sensors and high-efficiency batteries, the operation of small Unmanned Aerial Vehicles (sUAVs) has expanded from military use to commercial use. A promising commercial application of sUAS is package delivery because of its potential to reduce acquisition and operating costs of the last-mile delivery system while enabling new services such as same-day delivery. Furthermore, in urban areas, sUAVs can deliver packages to customers without negatively affecting street traffic. This thesis addresses an extended sizing and synthesis process that considers the performance of a sUAS in its total cost optimized routing to size the vehicles for sUAS-based delivery systems. This problem is called the concurrent aircraft design and routing problem. Based on decomposition approaches, this problem can be divided into three parts: the sizing, the route planning, and the integration of the two. However, the existing methods have mainly focused either UAV design or optimization of operations of UAVs. Although a concurrent UAV design and routing problem is addressed, only simple routing problems are studied without considering the obstructed environment like an urban area. To fully address the concurrent aircraft design and routing problem for sUAS-based delivery systems, this thesis presents a novel modular framework including all three parts of this problem: the UAV design module, the UAV routing module, and the integration method. First, for the UAV routing module, this thesis presents an endurance-constrained Multi-Trip Vehicle Routing Problem with time windows (MTVRPTW) optimization model that is an extension of the MTVRPTW optimization model. The MTVRPTW model builds a vehicle's schedule including a reuse plan for an on-demand delivery system. However, the MTVRPTW model does not consider the property of sUAV's limited endurance. To alleviate the limitation of the MTVRPTW model, the endurance-constrained MTVRPTW model employs maximum endurance constraints that trace flight time of each vehicle and restrict flight time to vehicle's maximum endurance. Moreover, to address the urban environment with the optimization model, this thesis presents a framework for creating a two-layered urban flight network as an input graph of a Vehicle Routing Problem (VRP) optimization model. The urban flight network is built by feeding airborne Light Detection And Ranging (LiDAR) sensor data into an algorithm that uses a Voronoi diagram to create collision-free paths. By integrating the endurance-constrained MTVRPTW model with the two-layered urban flight network, vehicle's schedule for sUAS-based urban delivery is created. Second, for the UAV design module, a component-based sizing and synthesis process for small fixed-wing VTOL UAVs is implemented. The sizing and synthesis process is an extension of traditional fixed-wing aircraft sizing and synthesis tool. The implemented process can consider vertical flight capacity and provide a take-off weight optimized combination of components of the propulsion system. The main intent of implementing the sizing and synthesis process is to make the framework for the concurrent UAV design and routing problem. Thus, the framework can be extended by integrating other sizing and synthesis tools if the interface is matched. Lastly, to integrate the UAV design module with the UAV routing module, existing methods have used a sequential approach; after conducting the sizing module, the vehicle routing problem is solved. However, the input and output of the two modules are coupled each other. Thus, the methods cannot address a converged solution for both modules. To alleviate the limitation, this thesis presents a novel modular framework for the concurrent aircraft design and routing problem for sUAS-based delivery systems, which is based on a Fixed Point Iteration (FPI) method to find a converged solution of the coupled problem. The presented framework can provide an optimal vehicle design and routing for the sUAS-based delivery system concurrently. This thesis uses the developed framework for concurrent UAV design and routing to study a possible package delivery using sUAS in San Diego, CA. The result shows that the developed framework can take into account both planning vehicle operation on the flight network in which it will be operating and designing the flight network capable of addressing the obstructed environment as part of the vehicle design process.