Filters








1,922 Hits in 6.9 sec

Vision-Based Autonomous Drone Control using Supervised Learning in Simulation [article]

Max Christl
2020 arXiv   pre-print
We address these challenges in the context of autonomous navigation and landing of MAVs in indoor environments and propose a vision-based control approach using Supervised Learning.  ...  To achieve this, we collected data samples in a simulation environment which were labelled according to the optimal control command determined by a path planning algorithm.  ...  A solution to the autonomous control of MAVs are vision-based systems that only rely on data captured by a camera and basic onboard sensors.  ... 
arXiv:2009.04298v1 fatcat:6ocygp25qvci5psecvasf7oxei

Visual Attention Prediction Improves Performance of Autonomous Drone Racing Agents [article]

Christian Pfeiffer, Simon Wengeler, Antonio Loquercio, Davide Scaramuzza
2022 arXiv   pre-print
We then use this visual attention prediction model to train an end-to-end controller for vision-based autonomous drone racing using imitation learning.  ...  We hypothesize that gaze-based attention prediction can be an efficient mechanism for visual information selection and decision making in a simulator-based drone racing task.  ...  Acknowledgments We thank Yunlong Song for help with the Flightmare simulator configuration.  ... 
arXiv:2201.02569v2 fatcat:scc33lp2ovcrdkov6uris2w7fy

Visual attention prediction improves performance of autonomous drone racing agents

Christian Pfeiffer, Simon Wengeler, Antonio Loquercio, Davide Scaramuzza, Sathishkumar V E
2022 PLoS ONE  
We then use this visual attention prediction model to train an end-to-end controller for vision-based autonomous drone racing using imitation learning.  ...  We hypothesize that gaze-based attention prediction can be an efficient mechanism for visual information selection and decision making in a simulator-based drone racing task.  ...  Acknowledgments We thank Yunlong Song for help with the Flightmare simulator configuration.  ... 
doi:10.1371/journal.pone.0264471 pmid:35231038 pmcid:PMC8887736 fatcat:h2slgyjxq5bs7koyibvda4gil4

Learning to Fly by Crashing [article]

Dhiraj Gandhi, Lerrel Pinto, Abhinav Gupta
2017 arXiv   pre-print
One approach is to use a small dataset collected by human experts: however, high capacity learning algorithms tend to overfit when trained with little data. An alternative is to use simulation.  ...  We use all this negative flying data in conjunction with positive data sampled from the same trajectories to learn a simple yet powerful policy for UAV navigation.  ...  AG was supported in part by Sloan Research Fellowship.  ... 
arXiv:1704.05588v2 fatcat:m47mx7fvuvgqbpzfvx2rmrrm2e

Open-Source Drone Programming Course for Distance Engineering Education

José M. Cañas, Diego Martín-Martín, Pedro Arias, Julio Vega, David Roldán-Álvarez, Lía García-Pérez, Jesús Fernández-Conde
2020 Electronics  
This "drone programming" course is open-access and ready-to-use for any teacher/student to teach/learn drone programming with it for free.  ...  Its educational contents are built upon robot operating system (ROS) middleware (de facto standard in robot programming), the powerful 3D Gazebo simulator, and the widely used Python programming language  ...  Acknowledgments: The authors thank Google for funding the JdeRobot non-profit organization in its calls for Google Summer of Code 2015, 2017, 2018, 2019, and 2020.  ... 
doi:10.3390/electronics9122163 fatcat:4xhbfba57fbc3nm6iw3u6hhfkq

Persistent self-supervised learning principle: from stereo to monocular vision for obstacle avoidance [article]

Kevin van Hecke, Guido de Croon, Laurens van der Maaten, Daniel Hennes, Dario Izzo
2016 arXiv   pre-print
Both simulations and real-world experiments with a stereo vision equipped AR drone 2.0 show the feasibility of this approach, with the robot successfully using monocular vision to avoid obstacles in a  ...  Self-Supervised Learning (SSL) is a reliable learning mechanism in which a robot uses an original, trusted sensor cue for training to recognize an additional, complementary sensor cue.  ...  Setup We simulate a 'flying' drone with stereo vision camera in SmartUAV [31] , an in-house developed simulator that allows for 3D rendering and simulation of the sensors and algorithms used on board  ... 
arXiv:1603.08047v1 fatcat:zn6aoeyjqrhsfmeq3uda66mnqa

Persistent self-supervised learning: From stereo to monocular vision for obstacle avoidance

Kevin van Hecke, Guido de Croon, Laurens van der Maaten, Daniel Hennes, Dario Izzo
2018 International Journal of Micro Air Vehicles  
Self-supervised learning (SSL) does not learn a control policy as LfD and RL, but rather focuses on improving the sensory inputs used in control.  ...  Setup We simulate a "flying" drone with stereo vision camera in SmartUAV, 37 an in-house developed simulator that allows for 3D rendering and simulation of the sensors and algorithms used on board the  ... 
doi:10.1177/1756829318756355 fatcat:ate5sovpirblnakqjn4sjacttm

Learning Vision-based Cohesive Flight in Drone Swarms [article]

Fabian Schilling, Julien Lecoeur, Fabrizio Schiano, Dario Floreano
2018 arXiv   pre-print
We simulate a swarm of quadrotor drones and formulate the controller as a regression problem in which we generate 3D velocity commands directly from raw camera images.  ...  This paper presents a data-driven approach to learning vision-based collective behavior from a simple flocking algorithm.  ...  Acknowledgements We thank Enrica Soria for the feedback and helpful discussions, as well as Olexandr Gudozhnik and Przemyslaw Kornatowski for their contributions to the drone hardware.  ... 
arXiv:1809.00543v1 fatcat:lkizzqursnejndc3muovz6riri

Vision Based Drone Obstacle Avoidance by Deep Reinforcement Learning

Zhihan Xue, Tad Gonsalves
2021 AI  
Reinforcement learning can overcome this problem by using drones to learn data in the environment.  ...  Among them, an increasing number of researchers are using machine learning to train drones. These studies typically adopt supervised learning or reinforcement learning to train the networks.  ...  In fact, the training of endto-end vision-based DRL navigation strategy is very time-consuming, because the CNN used to learn vision-based functions involves multiple matrix operations.  ... 
doi:10.3390/ai2030023 fatcat:qeq3tq5a7ngmtacbrx5cuazw5e

Accuracy Improvement of Autonomous Straight Take-off, Flying Forward, and Landing of a Drone with Deep Reinforcement Learning

Che-Cheng Chang, Jichiang Tsai, Peng-Chen Lu, Chuan-An Lai
2020 International Journal of Computational Intelligence Systems  
In this paper, we present a novel method that uses ArUco markers as a reference to improve the accuracy of a drone on autonomous straight take-off, flying forward, and landing based on Deep Reinforcement  ...  Learning (DRL).  ...  One remarkable advantage is without the need of human supervision, as well as allowing the drone to learn how to use high-level actions autonomously.  ... 
doi:10.2991/ijcis.d.200615.002 fatcat:pjei3qqednenpgo4xq3pktjo24

A Vision-Based Neural Network Controller for the Autonomous Landing of a Quadrotor on Moving Targets

Abdullah Almeshal, Mohammad Alenezi
2018 Robotics  
Simulations of the proposed controller are presented using ROS Gazebo environment and are validated experimentally in the laboratory using a Parrot AR Drone system.  ...  This paper presents a design of vision-based neural network controller for the autonomous landing of a quadrotor on fixed and moving targets for Maritime Search and Rescue applications.  ...  Design In this section, a vision-based hybrid intelligent controller is developed for the autonomous landing on targets.  ... 
doi:10.3390/robotics7040071 fatcat:vjx4f3aupreyvjvj56q6whatde

Evaluation of Reinforcement and Deep Learning Algorithms in Controlling Unmanned Aerial Vehicles

Yalew Zelalem Jembre, Yuniarto Wimbo Nugroho, Muhammad Toaha Raza Khan, Muhammad Attique, Rajib Paul, Syed Hassan Ahmed Shah, Beomjoon Kim
2021 Applied Sciences  
In this work, we configure the quadrotor to fly autonomously by using agents (the machine learning schemes being used to fly the quadrotor autonomously) to learn about the virtual physical environment.  ...  However, stability, path planning, and control remain significant challenges in autonomous quadrotor flights.  ...  Supervised learning is one of the most used methods in attempting UAV control, but the training dataset has been problematic in this regard.  ... 
doi:10.3390/app11167240 fatcat:qblc3c6io5gqhfefmfauszkfmq

CAD2RL: Real Single-Image Flight Without a Single Real Image

Fereshteh Sadeghi, Sergey Levine
2017 Robotics: Science and Systems XIII  
In [25] , laser range scanned real images are used to estimate depth in a supervised learning approach and then the output is used to learn control policies.  ...  LEARNING FROM SIMULATION Conventionally, learning-based approaches to autonomous flight have relied on learning from demonstration [2, 1, 30, 34] .  ... 
doi:10.15607/rss.2017.xiii.034 dblp:conf/rss/SadeghiL17 fatcat:meuops6hdzb4pbfluhw2hgv5gu

Obstacle Avoidance Drone by Deep Reinforcement Learning and Its Racing with Human Pilot

Sang-Yun Shin, Yong-Won Kang, Yong-Guk Kim
2019 Applied Sciences  
The former has the advantage in optimization for vision datasets, but such actions can lead to unnatural behavior.  ...  They typically adopt either supervised learning or reinforcement learning (RL) for training their networks.  ...  Besides, a study shows that a depth map generated from an RGB image using a CNN-based model can be used for training a network controlling a drone [38] by utilizing supervised learning.  ... 
doi:10.3390/app9245571 fatcat:hozuaj44avekhnb6nat3kghwxa

CAD2RL: Real Single-Image Flight without a Single Real Image [article]

Fereshteh Sadeghi, Sergey Levine
2017 arXiv   pre-print
In this paper, we explore the following question: can we train vision-based navigation policies entirely in simulation, and then transfer them into the real world to achieve real-world flight without a  ...  We propose a learning method that we call CAD^2RL, which can be used to perform collision-free indoor flight in the real world while being trained entirely on 3D CAD models.  ...  In [25] , laser range scanned real images are used to estimate depth in a supervised learning approach and then the output is used to learn control policies.  ... 
arXiv:1611.04201v4 fatcat:7w3vienpabgu7ejsfsq2qkqesm
« Previous Showing results 1 — 15 out of 1,922 results