Visual Navigation for UAV with Map References Using ConvNets

2016 ◽  
pp. 13-22 ◽  
Author(s):  
Fidel Aznar ◽  
Mar Pujol ◽  
Ramón Rizo
Keyword(s):  
Visual Navigation

A Visual Navigation Method Using a Hand-Drawn-Route-Map in Dynamic Environments

ROBOT ◽  
2011 ◽  
Vol 33 (4) ◽  
pp. 490-501 ◽  
Author(s):  
Xinde LI ◽  
Xuejian WU ◽  
Bo ZHU ◽  
Xianzhong DAI
Keyword(s):  
Dynamic Environments ◽  
Visual Navigation ◽  
Navigation Method

Visual Navigation With Multiple Goals Based on Deep Reinforcement Learning

2021 ◽  
pp. 1-11
Author(s):  
Zhenhuan Rao ◽  
Yuechen Wu ◽  
Zifei Yang ◽  
Wei Zhang ◽  
Shijian Lu ◽  
...  
Keyword(s):  
Reinforcement Learning ◽  
Visual Navigation ◽  
Multiple Goals

scholarly journals Air Learning: a deep reinforcement learning gym for autonomous aerial robot visual navigation

2021 ◽  
Author(s):  
Srivatsan Krishnan ◽  
Behzad Boroujerdian ◽  
William Fu ◽  
Aleksandra Faust ◽  
Vijay Janapa Reddi
Keyword(s):  
Reinforcement Learning ◽  
Embedded System ◽  
Broad Class ◽  
Visual Navigation ◽  
Raspberry Pi ◽  
Latency Distribution ◽  
Hardware In The Loop ◽  
Resource Constrained ◽  
Aerial Robot ◽  
Policy Optimization

AbstractWe introduce Air Learning, an open-source simulator, and a gym environment for deep reinforcement learning research on resource-constrained aerial robots. Equipped with domain randomization, Air Learning exposes a UAV agent to a diverse set of challenging scenarios. We seed the toolset with point-to-point obstacle avoidance tasks in three different environments and Deep Q Networks (DQN) and Proximal Policy Optimization (PPO) trainers. Air Learning assesses the policies’ performance under various quality-of-flight (QoF) metrics, such as the energy consumed, endurance, and the average trajectory length, on resource-constrained embedded platforms like a Raspberry Pi. We find that the trajectories on an embedded Ras-Pi are vastly different from those predicted on a high-end desktop system, resulting in up to $$40\%$$ 40 % longer trajectories in one of the environments. To understand the source of such discrepancies, we use Air Learning to artificially degrade high-end desktop performance to mimic what happens on a low-end embedded system. We then propose a mitigation technique that uses the hardware-in-the-loop to determine the latency distribution of running the policy on the target platform (onboard compute on aerial robot). A randomly sampled latency from the latency distribution is then added as an artificial delay within the training loop. Training the policy with artificial delays allows us to minimize the hardware gap (discrepancy in the flight time metric reduced from 37.73% to 0.5%). Thus, Air Learning with hardware-in-the-loop characterizes those differences and exposes how the onboard compute’s choice affects the aerial robot’s performance. We also conduct reliability studies to assess the effect of sensor failures on the learned policies. All put together, Air Learning enables a broad class of deep RL research on UAVs. The source code is available at: https://github.com/harvard-edge/AirLearning.

scholarly journals On Embodied Visual Navigation in Real Environments Through Habitat

2021 ◽  
Author(s):  
Marco Rosano ◽  
Antonino Furnari ◽  
Luigi Gulino ◽  
Giovanni Maria Farinella
Keyword(s):  
Visual Navigation

scholarly journals Mediastinal anatomical landmarks, their variants and tips for video-assisted thoracoscopic navigation during oesophageal extirpation

2021 ◽  
Author(s):  
Sergey Dydykin ◽  
Friedrich Paulsen ◽  
Tatyana Khorobykh ◽  
Natalya Mishchenko ◽  
Marina Kapitonova ◽  
...  
Keyword(s):  
Prone Position ◽  
Anatomical Study ◽  
Structural Features ◽  
Posterior Mediastinum ◽  
Visual Navigation ◽  
Anatomical Landmarks ◽  
Video Assisted ◽  
Before And After ◽  
Human Cadavers ◽  
The Right

Abstract Purpose There is no systematic description of primary anatomical landmarks that allow a surgeon to reliably and safely navigate the superior and posterior mediastinum’s fat tissue spaces near large vessels and nerves during video-assisted endothoracoscopic interventions in the prone position of a patient. Our aim was to develop an algorithm of sequential visual navigation during thoracoscopic extirpation of the esophagus and determine the most permanent topographic and anatomical landmarks allowing safe thoracoscopic dissection of the esophagus in the prone position. Methods The anatomical study of the mediastinal structural features was carried out on 30 human cadavers before and after opening the right pleural cavity. Results For thoracoscopic extirpation of the esophagus in the prone position, anatomical landmarks are defined, their variants are assessed, and an algorithm for their selection is developed, allowing their direct visualization before and after opening the mediastinal pleura. Conclusion The proposed algorithm for topographic and anatomical navigation based on the key anatomical landmarks in the posterior mediastinum provides safe performance of the video-assisted thoracoscopic extirpation of the esophagus in the prone position.

Smartphone-Based Indoor Visual Navigation with Leader-Follower Mode

2021 ◽  
Vol 17 (2) ◽  
pp. 1-22
Author(s):  
Jingao Xu ◽  
Erqun Dong ◽  
Qiang Ma ◽  
Chenshu Wu ◽  
Zheng Yang
Keyword(s):  
Real Time ◽  
Environmental Changes ◽  
State Of The Art ◽  
Visual Navigation ◽  
Indoor Navigation ◽  
Location Services ◽  
Localization And Mapping ◽  
Leaders And Followers ◽  
Indoor Navigation System ◽  
Free Pair

Existing indoor navigation solutions usually require pre-deployed comprehensive location services with precise indoor maps and, more importantly, all rely on dedicatedly installed or existing infrastructure. In this article, we present Pair-Navi, an infrastructure-free indoor navigation system that circumvents all these requirements by reusing a previous traveler’s (i.e., leader) trace experience to navigate future users (i.e., followers) in a Peer-to-Peer mode. Our system leverages the advances of visual simultaneous localization and mapping ( SLAM ) on commercial smartphones. Visual SLAM systems, however, are vulnerable to environmental dynamics in the precision and robustness and involve intensive computation that prohibits real-time applications. To combat environmental changes, we propose to cull non-rigid contexts and keep only the static and rigid contents in use. To enable real-time navigation on mobiles, we decouple and reorganize the highly coupled SLAM modules for leaders and followers. We implement Pair-Navi on commodity smartphones and validate its performance in three diverse buildings and two standard datasets (TUM and KITTI). Our results show that Pair-Navi achieves an immediate navigation success rate of 98.6%, which maintains as 83.4% even after 2 weeks since the leaders’ traces were collected, outperforming the state-of-the-art solutions by >50%. Being truly infrastructure-free, Pair-Navi sheds lights on practical indoor navigations for mobile users.

TIME-TO-CONTACT INFORMATION ESTIMATION FOR MONOCULAR MOBILE ROBOTS

2008 ◽  
Vol 05 (03) ◽  
pp. 223-233 ◽  
Author(s):  
RONG LIU ◽  
MAX Q. H. MENG
Keyword(s):  
Mobile Robots ◽  
Optical Flow ◽  
Active Contour Model ◽  
Visual Navigation ◽  
Moving Object ◽  
Time To Contact ◽  
Temporal Derivative ◽  
Novel Method ◽  
Derivatives Of ◽  
Flow Experiments

Time-to-contact (TTC) provides vital information for obstacle avoidance and for the visual navigation of a robot. In this paper, we present a novel method to estimate the TTC information of a moving object for monocular mobile robots. In specific, the contour of the moving object is extracted first using an active contour model; then the height of the motion contour and its temporal derivative are evaluated to generate the desired TTC estimates. Compared with conventional techniques employing the first-order derivatives of optical flow, the proposed estimator is less prone to errors of optical flow. Experiments using real-world images are conducted and the results demonstrate that the developed method can successfully achieve TTC with an average relative error (ARVE) of 0.039 with a single calibrated camera.

Sign in / Sign up

Export Citation Format

Share Document