Abstract:The object recognition technology of unmanned aerial vehicles (UAVs) equipped with “You Only Look Once” (YOLO) has been validated in actual flights. However, here, the challenge lies in efficiently utilizing camera gimbal control technology to swiftly capture images of YOLO-identified target objects in aerial search missions. Enhancing the UAV’s energy efficiency and search effectiveness is imperative. This study aims to establish a simulation environment by employing the Unity simulation software for target tracking by controlling the gimbal. This approach involves the development of deep deterministic policy-gradient (DDPG) reinforcement-learning techniques to train the gimbal in executing effective tracking actions. The outcomes of the simulations indicate that when actions are appropriately rewarded or penalized in the form of scores, the reward value can be consistently converged within the range of 19–35. This convergence implies that a successful strategy leads to consistently high rewards. Consequently, a refined set of training procedures is devised, enabling the gimbal to accurately track the target. Moreover, this strategy minimizes unnecessary tracking actions, thus enhancing tracking efficiency. Numerous benefits arise from training in a simulated environment. For instance, the training in this simulated environment is facilitated through a dataset composed of actual flight photographs. Furthermore, offline operations can be conducted at any given time without any constraint of time and space. Thus, this approach effectively enables the training and enhancement of the gimbal’s action strategies. The findings of this study demonstrate that a coherent set of action strategies can be proficiently cultivated by employing DDPG reinforcement learning. Furthermore, these strategies empower the UAV’s gimbal to rapidly and precisely track designated targets. Therefore, this approach provides both convenience and opportunities to gather more flight-scenario training data in the future. This gathering of data will lead to immediate training opportunities and help improve the system’s energy consumption.

Imitation Learning-Based Algorithm for Drone Cinematography System

Can We Enable the Drone to be a Filmmaker?

Manipulating Camera Gimbal Positioning by Deep Deterministic Policy Gradient Reinforcement Learning for Drone Object Detection

Onboard View Planning of a Flying Camera for High Fidelity 3D Reconstruction of a Moving Actor

Towards a Drone Cinematographer: Guiding Quadrotor Cameras using Visual Composition Principles

Auto Filmer: Autonomous Aerial Videography Under Human Interaction

Learning Deep Sensorimotor Policies for Vision-based Autonomous Drone Racing

Active Human Pose Estimation via an Autonomous UAV Agent

Vision-based drone control for autonomous UAV cinematography

Optimal Trajectory Planning for Autonomous Drone Cinematography

Example-driven Virtual Cinematography by Learning Camera Behaviors

Learning Autonomous Viewpoint Adjustment from Human Demonstrations for Telemanipulation

Automatic View Finding for Drone Photography Based on Image Aesthetic Evaluation

Keyframe-Focused Visual Imitation Learning

A Method For Automated Drone Viewpoints to Support Remote Robot Manipulation

Automated Cinematography Motion Planning for UAVs

Do You See What I See? Coordinating Multiple Aerial Cameras for Robot Cinematography

Optimal Trajectory Planning for Cinematography with Multiple Unmanned Aerial Vehicles

Vision-based Robotic Arm Imitation by Human Gesture

Demonstrating Agile Flight from Pixels without State Estimation

Learning Vision-Based Flight in Drone Swarms by Imitation