Autonomous Vision-based Target Detection Using Unmanned Aerial Vehicle

Deeds, Jeff; Engstrom, Zach; Gill, Caleb; Wood, Zack; Wang, Jing; Ahn, In Soo; Lu, Yufeng

doi:10.1109/mwscas.2018.8623940

Cited by 7 publications

(2 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These smart machines are capable to accomplish more than simply completing tasks. With the power of AI, they learn, evolve, and interact with the environment surrounding them [22], [23]. In [24], a framework for solving a general VRP using RL is proposed.…”

Section: A Related Workmentioning

confidence: 99%

A Generic Spatiotemporal Scheduling for Autonomous UAVs: A Reinforcement Learning-Based Approach

Bouhamed

Ghazzai

Besbes

et al. 2020

IEEE Open J. Veh. Technol.

View full text Add to dashboard Cite

Considerable attention has been given to leverage a variety of smart city applications using unmanned aerial vehicles (UAVs). The rapid advances in artificial intelligence can empower UAVs with autonomous capabilities allowing them to learn from their surrounding environment and act accordingly without human intervention. In this paper, we propose a spatiotemporal scheduling framework for autonomous UAVs using reinforcement learning. The framework enables UAVs to autonomously determine their schedules to cover the maximum of pre-scheduled events spatially and temporally distributed in a given geographical area and over a predetermined time horizon. The designed framework has the ability to update the planned schedules in case of unexpected emergency events. The UAVs are trained using the Q-learning (QL) algorithm to find effective scheduling plan. A customized reward function is developed to consider several constraints especially the limited battery capacity of the flying units, the time windows of events, and the delays caused by the UAV navigation between events. Numerical simulations show the behavior of the autonomous UAVs for various scenarios and corroborate the ability of QL to handle complex vehicle routing problems with several constraints. A comparison with an optimal deterministic solution is also provided to validate the performance of the learning-based solution.

show abstract

Section: A Related Workmentioning

confidence: 99%

A Generic Spatiotemporal Scheduling for Autonomous UAVs: A Reinforcement Learning-Based Approach

Bouhamed

Ghazzai

Besbes

et al. 2020

IEEE Open J. Veh. Technol.

View full text Add to dashboard Cite

show abstract

“…Beyond the precise landing, several state-of-the-art solutions take advantage of the vision-based approach to deal with target identification. For example, in [ 31 , 32 ] UAV landing is preceded by the localization of a target associated with a certain fiducial tag. Similarly, in [ 33 , 34 ], the UAV detection of (and eventually the reaction to) multiple targets is taken into account, based on the exploitation of different fiducial markers.…”

Section: Introductionmentioning

confidence: 99%

Indoor Visual-Based Localization System for Multi-Rotor UAVs

Bertoni

Michieletto

Oboe

et al. 2022

Sensors

View full text Add to dashboard Cite

Industry 4.0, smart homes, and the Internet of Things are boosting the employment of autonomous aerial vehicles in indoor environments, where localization is still challenging, especially in the case of close and cluttered areas. In this paper, we propose a Visual Inertial Odometry localization method based on fiducial markers. Our approach enables multi-rotor aerial vehicle navigation in indoor environments and tackles the most challenging aspects of image-based indoor localization. In particular, we focus on a proper and continuous pose estimation, working from take-off to landing, at several different flying altitudes. With this aim, we designed a map of fiducial markers that produces results that are both dense and heterogeneous. Narrowly placed tags lead to minimal information loss during rapid aerial movements while four different classes of marker size provide consistency when the camera zooms in or out according to the vehicle distance from the ground. We have validated our approach by comparing the output of the localization algorithm with the ground-truth information collected through an optoelectronic motion capture system, using two different platforms in different flying conditions. The results show that error mean and standard deviation can remain constantly lower than 0,11m, so not degrading when the aerial vehicle increases its altitude and, therefore, strongly improving similar state-of-the-art solutions.

show abstract