We [1] propose a reinforcement learning approach for distributed collision avoidance and an optimization-based approach that efficiently obtains a trajectory for data collection in unmanned aerial vehicle (UAV) communication networks. Specifically, each UAV takes charge of delivering physical objects in the forward path and collecting data from terrestrial IoT devices in the backward path. To obtain an optimal visiting order for IoT devices, we formulate and solve a no-return traveling salesman problem. Given a visiting order, we formulate and solve a sequence of convex optimization problems to obtain line segments of an optimal backward path for heterogeneous ground IoT devices. Simulation results show that the proposed approach is superior to a number of alternative approaches.

We [2] propose a novel approach of designing the velocity function and the 3D trajectory for a UAV to efficiently attain location-dependent visual coverage. Specifically, the UAV dynamically adjusts its altitude to photograph terrestrial polygons with unequal image resolution requirements. Unlike prior work that assumes the UAV speed is constant, the proposed approach allows the UAV to change its speed. To minimize the task completion time, we put forward a novel approach that is composed of three algorithmic components. The first component uses an aggressive method for selecting the UAV photographing altitudes, designs the UAV velocity functions, and derives the UAV flying times for all pairs of regions. Based on the UAV flying times rather than the distances, the second component utilizes an auxiliary traveling salesman problem to select the visited order of terrestrial regions. For each region, the third component creates candidate coverage paths and chooses the coverage path based on the UAV flying time. Simulation results indicate that the proposed approach outperforms a two-dimensional trajectory planning algorithm and a greedy algorithm for planning a three-dimensional trajectory in terms of the UAV task completion time. 

[1] Y.-H. Hsu and R.-H. Gau, "Reinforcement Learning-Based Collision Avoidance and Optimal Trajectory Planning in UAV Communication Networks," in IEEE Transactions on Mobile Computing, vol. 21, no. 1, pp. 306-320, Jan. 2022. (Link)
[2] Y.-C. Ko and R.-H. Gau, "UAV Velocity Function Design and Trajectory Planning for Heterogeneous Visual Coverage of Terrestrial Regions," in IEEE Transactions on Mobile Computing, vol. 22, no. 10, pp. 6205-6222, Oct. 2023. (Link)