Learning to Predict the Wind for Safe Aerial Vehicle Planning

Achermann, Florian; Lawrance, Nicholas; Ranftl, René; Dosovitskiy, Alexey; Chung, Jen Jen; Siegwart, Roland

doi:10.1109/icra.2019.8793547

Cited by 16 publications

(11 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Regardless, many existing datasets are taken from ground vehicles which makes the transfer to an aerial view challenging. Algorithms developed on the visual-thermal dataset presented in [10] which was taken from a UAV flying over Swiss farmland in broad daylight could be useful for WiSAR settings, but would not be very robust to varied weather and lighting conditions that are characteristic of WiSAR operations. Datasets for search and rescue: Unfortunately, there are not many existing datasets targeted for search and rescue applications.…”

Section: B Related Datasetsmentioning

confidence: 99%

“…Recent work leverages deep neural networks to address the cross-spectral alignment problem [10]. This method significantly outperforms classical methods for estimating the homography between multi-spectral image pairs in baseline tests, however the imagery used for training the neural network is qualitatively more benign (i.e., mostly farmland) compared to WiSARD.…”

Section: E Visual-thermal Image Registrationmentioning

confidence: 99%

See 1 more Smart Citation

WiSARD: A Labeled Visual and Thermal Image Dataset for Wilderness Search and Rescue

Broyles

Hayner

Leung

2022

2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

View full text Add to dashboard Cite

Sensor-equipped unoccupied aerial vehicles (UAVs) have the potential to help reduce search times and alleviate safety risks for first responders carrying out Wilderness Search and Rescue (WiSAR) operations, the process of finding and rescuing person(s) lost in wilderness areas. Unfortunately, visual sensors alone do not address the need for robustness across all the possible terrains, weather, and lighting conditions that WiSAR operations can be conducted in. The use of multi-modal sensors, specifically visual-thermal cameras, is critical in enabling WiSAR UAVs to perform in diverse operating conditions. However, due to the unique challenges posed by the wilderness context, existing dataset benchmarks are inadequate for developing vision-based algorithms for autonomous WiSAR UAVs. To this end, we present WiSARD, a dataset with roughly 56,000 labeled visual and thermal images collected from UAV flights in various terrains, seasons, weather, and lighting conditions. To the best of our knowledge, WiSARD is the first large-scale dataset collected with multi-modal sensors for autonomous WiSAR operations. We envision that our dataset will provide researchers with a diverse and challenging benchmark that can test the robustness of their algorithms when applied to real-world (life-saving) applications. Link to dataset: https://sites.google.com/uw.edu/wisard/ ⋆ Both authors contributed equally.

show abstract

Section: B Related Datasetsmentioning

confidence: 99%

Section: E Visual-thermal Image Registrationmentioning

confidence: 99%

WiSARD: A Labeled Visual and Thermal Image Dataset for Wilderness Search and Rescue

Broyles

Hayner

Leung

2022

2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

View full text Add to dashboard Cite

show abstract

“…Among the non-parametric methods, Gaussian Processes (GPs) have found applications to estimate the mean wind field [44], [45] and to make forecasts [46]. Regarding supervised methods, [47] uses a convolutional neural network (CNN) to predict the wind field given the 3D geometry of the environment; [48] proposes a shallow neural network to estimate time-varying flows. Approaches that more explicitly attempt to model the spatial distribution of the wind are presented by [34], where the relationship between wind speed and altitude is assumed to follow a Weibull distribution, while [49] employs a polynomial model whose parameters can be estimated online.…”

Section: Related Workmentioning

confidence: 99%

Airflow-Inertial Odometry for Resilient State Estimation on Multirotors

Tagliabue¹,

How²

2021

Preprint

View full text Add to dashboard Cite

We present a dead reckoning strategy for increased resilience to position estimation failures on multirotors, using only data from a low-cost IMU and novel, bio-inspired airflow sensors. The goal is challenging, since low-cost IMUs are subject to large noise and drift, while 3D airflow sensing is made difficult by the interference caused by the propellers and by the wind. Our approach relies on a deep-learning strategy to interpret the measurements of the bio-inspired sensors, a map of the wind speed to compensate for position-dependent wind, and a filter to fuse the information and generate a pose and velocity estimate. Our results show that the approach reduces the drift with respect to IMU-only dead reckoning by up to an order of magnitude over 30 seconds after a position sensor failure in non-windy environments, and it can compensate for the challenging effects of turbulent, and spatially varying wind.

show abstract

“…• CNN to estimate the 3D wind flow for trajectory adjustment [10]. • SVM to classify LoS and NLoS link conditions of the UAV along the flight [11].…”

Section: Situational Awarenessmentioning

confidence: 99%

“…This information is invaluable to identify the appropriate air spaces and times in advance, before such weather complexities take place. The authors of [10] estimate the 3D wind flow using a deep Convolutional Neural Network (CNN) in less than two minutes to enable safer trajectories for the UAV during strong wind conditions. Reference [19] highlights that the time required for safer UAV flight path determination must be done in less than 30 seconds.…”

Section: B Situational Awarenessmentioning

confidence: 99%

Machine Learning-Assisted UAV Operations with the UTM: Requirements, Challenges, and Solutions

Abdalla

Marojevic

2020

2020 IEEE 92nd Vehicular Technology Conference (VTC2020-Fall)

View full text Add to dashboard Cite

Unmanned aerial vehicles (UAVs) are emerging in commercial spaces and will support many applications and services, such as smart agriculture, dynamic network deployment, and network coverage extension, surveillance and security. The unmanned aircraft system (UAS) traffic management (UTM) provides a framework for safe UAV operation integrating UAV controllers and central data bases via a communications network. This paper discusses the challenges and opportunities for machine learning (ML) for effectively providing critical UTM services. We introduce the four pillars of UTM-operation planning, situational awareness, status and advisors and security-and discuss the main services, specific opportunities for ML and the ongoing research. We conclude that the multi-faceted operating environment and operational parameters will benefit from collected data and data-driven algorithms, as well as online learning to face new UAV operation situations.

show abstract

Learning to Predict the Wind for Safe Aerial Vehicle Planning

Cited by 16 publications

References 30 publications

WiSARD: A Labeled Visual and Thermal Image Dataset for Wilderness Search and Rescue

WiSARD: A Labeled Visual and Thermal Image Dataset for Wilderness Search and Rescue

Airflow-Inertial Odometry for Resilient State Estimation on Multirotors

Machine Learning-Assisted UAV Operations with the UTM: Requirements, Challenges, and Solutions

Contact Info

Product

Resources

About