Conference Agenda

Nowadays, truck-and-drone problems represent one of the most studied classes of vehicle routing problems. The Flying Sidekick Traveling Salesman Problem (FS-TSP) is the first optimization problem defined in this class. Since its definition, several variants have been proposed differing for the side constraints related to the operating conditions and for the structure of the hybrid truck-and-drone delivery system. However, regardless the specific problem under investigation, determining the optimal solution of most of these routing problems is a very challenging task, due to the vehicle synchronization issue and drone hovering.

On this basis, this work provides a new arc-based integer linear programming formulation for the FS-TSP. The proposed formulation is quite general and can be easily extended to deal with two interesting variants of the FS-TSP, namely: the Traveling Salesman Problem with Drone and Lockers (TSP-DL), where customers are served either by truck or a drone directly at their houses or use self-pickup facility (i.e., lockers); the Truck-Drone Team Logistics Problem (TDTLP ), where a drone can serve multiple customers during in each flight.

The solution of the proposed formulation and its variants required the development of a branch-and-cut solution approach based on new families of valid inequalities and variable fixing strategies.

We tested the proposed approach on different sets of benchmark instances. The experimentation shows that the proposed method is competitive or outperforms the state-of-the-art approaches, providing either the optimal solution or improved bounds for several instances unsolved before.

Robotic applications in production and logistics generally require collaboration and coordination between robots and humans. Smooth human-robot interaction is essential to ensure efficient operations and requires integrated planning of both actors. This contribution investigates robot and human collaboration planning for unmanned aerial vehicle (UAV) applications requiring the overflight of multiple, spatially distributed regions, where each region contains a dense set of target locations. As the human operator must maintain a continuous visual line of sight (VLOS) to the UAV in flight, its freedom of movement is restricted. The operator must also ferry the UAV between a separate set of potential takeoff and landing locations. Such problems can be found, among others, in last-mile logistics, survey missions, or agricultural settings for the deployment of biological pest control. Considering VLOS in these planning problems is paramount to allow the implementation of solution approaches under the current strict legal and operational restrictions currently in force in a wide range of jurisdictions, including the European Union.

We present a mathematical model formulation of the problem and develop a tailored solution approach to generate feasible, high-quality solutions in reasonable computation times. Algorithmic performance and solution quality are evaluated for several scenarios derived from real-world data.