SQUEEZEPOSENET: IMAGE BASED POSE REGRESSION WITH SMALL CONVOLUTIONAL NEURAL NETWORKS FOR REAL TIME UAS NAVIGATION

The number of unmanned aerial vehicles (UAVs) is increasing since low-cost airborne systems are available for a wide range of users. The outdoor navigation of such vehicles is mostly based on global navigation satellite system (GNSS) methods to gain the vehicles trajectory. The drawback of satellite-based navigation are failures caused by occlusions and multi-path interferences. Beside this, local image-based solutions like Simultaneous Localization and Mapping (SLAM) and Visual Odometry (VO) can e.g. be used to support the GNSS solution by closing trajectory gaps but are computationally expensive. However, if the trajectory estimation is interrupted or not available a re-localization is mandatory. In this paper we will provide a novel method for a GNSS-free and fast image-based pose regression in a known area by utilizing a small convolutional neural network (CNN). With on-board processing in mind, we employ a lightweight CNN called SqueezeNet and use transfer learning to adapt the network to pose regression. Our experiments show promising results for GNSS-free and fast localization.

Download Full-text

GNSS-ISE: Instruction Set Extension for GNSS Baseband Processing

Sensors ◽

10.3390/s20020465 ◽

2020 ◽

Vol 20 (2) ◽

pp. 465 ◽

Cited By ~ 1

Author(s):

Krzysztof Marcinek ◽

Witold A. Pleskacz

Keyword(s):

Flash Memory ◽

Low Cost ◽

Satellite System ◽

Microprocessor System ◽

Instruction Set ◽

Advantages And Disadvantages ◽

Wide Range ◽

Instruction Set Extension ◽

Baseband Processing ◽

Global Navigation Satellite

This work presents the results of research toward designing an instruction set extension dedicated to Global Navigation Satellite System (GNSS) baseband processing. The paper describes the state-of-the-art techniques of GNSS receiver implementation. Their advantages and disadvantages are discussed. Against this background, a new versatile instruction set extension for GNSS baseband processing is presented. The authors introduce improved mechanisms for instruction set generation focused on multi-channel processing. The analytical approach used by the authors leads to the introduction of a GNSS-instruction set extension (ISE) for GNSS baseband processing. The developed GNSS-ISE is simulated extensively using PC software and field-programmable gate array (FPGA) emulation. Finally, the developed GNSS-ISE is incorporated into the first-in-the-world, according to the authors’ best knowledge, integrated, multi-frequency, and multi-constellation microcontroller with embedded flash memory. Additionally, this microcontroller may serve as an application processor, which is a unique feature. The presented results show the feasibility of implementing the GNSS-ISE into an embedded microprocessor system and its capability of performing baseband processing. The developed GNSS-ISE can be implemented in a wide range of applications including smart IoT (internet of things) devices or remote sensors, fostering the adaptation of multi-frequency and multi-constellation GNSS receivers to the low-cost consumer mass-market.

Download Full-text

Control System Architecture for Automatic Recovery of Fixed-Wing Unmanned Aerial Vehicles in a Moving Arrest System

Journal of Intelligent & Robotic Systems ◽

10.1007/s10846-021-01521-z ◽

2021 ◽

Vol 103 (4) ◽

Author(s):

Kristoffer Gryte ◽

Martin L. Sollie ◽

Tor Arne Johansen

Keyword(s):

Unmanned Aerial Vehicles ◽

Low Cost ◽

Field Testing ◽

Satellite System ◽

Base Station ◽

Research Literature ◽

Aerial Vehicles ◽

Global Navigation Satellite ◽

The Right ◽

Automatic Recovery

AbstractAutomatic recovery is an important step in enabling fully autonomous missions using fixed-wing unmanned aerial vehicles (UAVs) operating from ships or other moving platforms. However, automatic recovery in moving arrest systems is only briefly studied in the research literature, and is not yet an option when using low-cost, commercial off-the-shelf (COTS) autopilots. Acknowledging the reliability and low cost of COTS avionics, this paper adds recovery functionality as a modular extension based on non-intrusive additions to an autopilot with very general assumptions on its interface. This is achieved by line-of-sight guidance, which sends an augmented desired position to the autopilot, to ensure line-following along a virtual runway that guides the UAV into the arrest system. The translation and rotation of this line is determined by the pose of the arrest system, determined using two Global Navigation Satellite System (GNSS) receivers, where one is configured as a Real-Time Kinematic (RTK) base station. The relative position of the UAV and arrest system is also precisely estimated using RTK GNSS. Through extensive field testing, on two different fixed-wing UAVs, the system has shown its performance and reliability; 43 recovery attempts in a stationary net hit 0.01 ± 0.25m to the right and 0.07 ± 0.20m below the target in calm conditions. Further, 15 recoveries in a barge-mounted, ship-towed net hit 0.06 ± 0.53m to the right and 0.98 ± 0.27m below the target in winds up to 4 m/s. The remaining error is largely systematic, caused by communication delays, and could be reduced with more integral effect or through direct compensation.

Download Full-text

Determination of Cultivated Area, Field Boundary and Overlapping for A Plowing Operation Using ISO 11783 Communication and D-GNSS Position Data

Agriculture ◽

10.3390/agriculture9020038 ◽

2019 ◽

Vol 9 (2) ◽

pp. 38 ◽

Cited By ~ 6

Author(s):

Andreas Heiß ◽

Dimitrios Paraforos ◽

Hans Griepentrog

Keyword(s):

Crop Production ◽

Low Cost ◽

Satellite System ◽

Position Information ◽

Cost Differential ◽

Main Challenge ◽

Wide Range ◽

Position Data ◽

Machine Speed ◽

Global Navigation Satellite

Easily available and detailed area-related information is very valuable for the optimization of crop production processes in terms of, e.g., documentation and invoicing or detection of inefficiencies. The present study dealt with the development of algorithms to gain sophisticated information about different area-related parameters in a preferably automated way. Rear hitch position and wheel-based machine speed were recorded from ISO 11783 communication data during plowing with a mounted reversible moldboard plow. The data were georeferenced using the position information from a low-cost differential global navigation satellite system (D-GNSS) receiver. After the exclusion of non-work sequences from continuous data logs, single cultivated tracks were reconstructed, which represented as a whole the cultivated area of a field. Based on that, the boundary of the field and the included area were automatically detected with a slight overestimation of 1.4%. Different field parts were distinguished and single overlaps between the cultivated tracks were detected, which allowed a distinct assessment of the lateral and headland overlapping (2.05% and 3.96%, respectively). Incomplete information about the work state of the implement was identified as the main challenge to get precise results. With a few adaptions, the used methodology could be transferred to a wide range of mounted implements.

Download Full-text

Precise Single-Frequency Positioning Using Low-Cost Receiver with the Aid of Lane-Level Map Matching for Land Vehicle Navigation

Journal of Navigation ◽

10.1017/s0373463320000375 ◽

2020 ◽

pp. 1-14

Author(s):

Fei Liu ◽

Yue Liu ◽

Zhixi Nie ◽

Yang Gao

Keyword(s):

Low Cost ◽

Digital Camera ◽

Satellite System ◽

Single Frequency ◽

Map Matching ◽

Digital Map ◽

The Road ◽

Ambiguity Fixing ◽

Wide Range ◽

Global Navigation Satellite

Precise positioning with low-cost single-frequency global navigation satellite system (GNSS) receivers has great potential in a wide range of applications because of its low price and improved accuracy. However, challenges remain in achieving reliable and accurate solutions using low-cost receivers. For instance, the successful ambiguity fixing rate could be low for real-time kinematic (RTK) while large errors may occur in precise point positioning (PPP) in some scenarios (e.g., trees along the road). To solve the problems, this paper proposes a method with the aid of additional lane-level digital map information to improve the accuracy and reliability of RTK and PPP solutions. In the method, a digital camera will be applied for lane recognition and the positioning solution from a low-cost receiver will be projected to the digital map lane link. With the projected point position as a constraint, the RTK ambiguity fixing rate and PPP performance can be enhanced. A field kinematic test was conducted to verify the improvement of the RTK and PPP solutions with the aid of map matching. The results show that the RTK ambiguity fixing rate can be increased and the PPP positioning error can be reduced by map matching.

Download Full-text

INS-AIDED 3D LIDAR SEAMLESS MAPPING IN CHALLENGING ENVIRONMENT FOR FUTURE HIGH DEFINITION MAP

ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences ◽

10.5194/isprs-archives-xlii-2-w16-251-2019 ◽

2019 ◽

Vol XLII-2/W16 ◽

pp. 251-257

Author(s):

G. J. Tsai ◽

K. W. Chiang ◽

N. El-Sheimy

Keyword(s):

Low Cost ◽

Autonomous Driving ◽

Satellite System ◽

Global Coordinate System ◽

Lower Grade ◽

High Definition ◽

Position Information ◽

Localization And Mapping ◽

Parking Lot ◽

Global Navigation Satellite

<p><strong>Abstract.</strong> With advances in computing and sensor technologies, onboard systems can deal with a large amount of data and achieve real-time process continuously and accurately. In order to further enhance the performance of positioning, high definition map (HD map) is one of the game changers for future autonomous driving. Instead of directly using Inertial Navigation System and Global Navigation Satellite System (INS/GNSS) navigation solutions to conduct the Direct Geo-referencing (DG) and acquiring 3D mapping information, Simultaneous Localization and Mapping (SLAM) relies heavily on environmental features to derive the position and attitude as well as conducting the mapping at the same time. In this research, the new structure is proposed to integrate the INS/GNSS into LiDAR Odometry and Mapping (LOAM) algorithm and enhance the mapping performance. The first contribution is using the INS/GNSS to provide the short-term relative position information for the mapping process when the LiDAR odometry process is failed. The checking process is built to detect the divergence of LiDAR odometry process based on the residual from correspondences of features and innovation sequence of INS/GNSS. More importantly, by integrating with INS/GNSS, the whole global map is located in the standard global coordinate system (WGS84) which can be shared and employed easily and seamlessly. In this research, the designed land vehicle platform includes commercial INS/GNSS integrated product as a reference, relatively low-cost and lower grade INS system and Velodyne LiDAR with 16 laser channels, respectively. The field test is conducted from outdoor to the indoor underground parking lot and the final solution using the proposed method has a significant improvement as well as building a more accurate and reliable map for future use.</p>

Download Full-text

Design of a Multiband Global Navigation Satellite System Radio Frequency Interference Monitoring Front-End with Synchronized Secondary Sensors

Sensors ◽

10.3390/s18082594 ◽

2018 ◽

Vol 18 (8) ◽

pp. 2594

Author(s):

Aiden Morrison ◽

Nadezda Sokolova ◽

James Curran

Keyword(s):

Radio Frequency ◽

Global Navigation Satellite System ◽

Low Cost ◽

Satellite System ◽

Radio Frequency Interference ◽

Navigation Satellite ◽

Inertial Measurement ◽

Front End ◽

Global Navigation ◽

Global Navigation Satellite

This paper investigates the challenges of developing a multi-frequency radio frequency interference (RFI) monitoring and characterization system that is optimized for ease of deployment and operation as well as low per unit cost. To achieve this, we explore the design and development of a multiband global navigation satellite system (GNSS) front-end which is intrinsically capable of synchronizing side channel information from non-RF sensors, such as inertial measurement units and integrated power meters, to allow the simultaneous production of substantial amounts of sampled spectrum while also allowing low-cost, real-time monitoring and logging of detected RFI events. While the inertial measurement unit and barometer are not used in the RFI investigation discussed, the design features that provide for their precise synchronization with the RF sample stream are presented as design elements worth consideration. The designed system, referred to as Four Independent Tuners with Data-packing (FITWD), was utilized in a data collection campaign over multiple European and Scandinavian countries in support of the determination of the relative occurrence rates of L1/E1 and L5/E5a interference events and intensities where it proved itself a successful alternative to larger and more expensive commercial solutions. The dual conclusions reached were that it was possible to develop a compact low-cost, multi-channel radio frequency (RF) front-end that implicitly supported external data source synchronization, and that such monitoring systems or similar capabilities integrated within receivers are likely to be needed in the future due to the increasing occurrence rates of GNSS RFI events.

Download Full-text

Kinematic PPP Using Mixed GPS/Glonass Single-Frequency Observations

Artificial Satellites ◽

10.2478/arsa-2019-0008 ◽

2019 ◽

Vol 54 (3) ◽

pp. 97-112

Author(s):

Mostafa Hamed ◽

Ashraf Abdallah ◽

Ashraf Farah

Keyword(s):

Low Cost ◽

Satellite System ◽

Single Frequency ◽

Future Research ◽

Frequency Data ◽

Dual Frequency ◽

Trajectory Data ◽

Average Improvement ◽

Kinematic Ppp ◽

Global Navigation Satellite

Abstract Nowadays, Precise Point Positioning (PPP) is a very popular technique for Global Navigation Satellite System (GNSS) positioning. The advantage of PPP is its low cost as well as no distance limitation when compared with the differential technique. Single-frequency receivers have the advantage of cost effectiveness when compared with the expensive dual-frequency receivers, but the ionosphere error makes a difficulty to be completely mitigated. This research aims to assess the effect of using observations from both GPS and GLONASS constellations in comparison with GPS only for kinematic purposes using single-frequency observations. Six days of the year 2018 with single-frequency data for the Ethiopian IGS station named “ADIS” were processed epoch by epoch for 24 hours once with GPS-only observations and another with GPS/GLONASS observations. In addition to “ADIS” station, a kinematic track in the New Aswan City, Aswan, Egypt, has been observed using Leica GS15, geodetic type, dual-frequency, GPS/GLONASS GNSS receiver and single-frequency data have been processed. Net_Diff software was used for processing all the data. The results have been compared with a reference solution. Adding GLONASS satellites significantly improved the satellite number and Position Dilution Of Precision (PDOP) value and accordingly improved the accuracy of positioning. In the case of “ADIS” data, the 3D Root Mean Square Error (RMSE) ranged between 0.273 and 0.816 m for GPS only and improved to a range from 0.256 to 0.550 m for GPS/GLONASS for the 6 processed days. An average improvement ratio of 24%, 29%, 30%, and 29% in the east, north, height, and 3D position components, respectively, was achieved. For the kinematic trajectory, the 3D position RMSE improved from 0.733 m for GPS only to 0.638 m for GPS/GLONASS. The improvement ratios were 7%, 5%, 28%, and 13% in the east, north, height, and 3D position components, respectively, for the kinematic trajectory data. This opens the way to add observations from the other two constellations (Galileo and BeiDou) for more accuracy in future research.

Download Full-text

Aerial and Ground Robot Collaboration for Autonomous Mapping in Search and Rescue Missions

Drones ◽

10.3390/drones4040079 ◽

2020 ◽

Vol 4 (4) ◽

pp. 79

Author(s):

Dimitrios Chatziparaschis ◽

Michail G. Lagoudakis ◽

Panagiotis Partsinevelos

Keyword(s):

Humanoid Robot ◽

Satellite System ◽

Search And Rescue ◽

Localization Algorithm ◽

Localization And Mapping ◽

Planning Algorithm ◽

Crisis Scenarios ◽

Trained Neural Network ◽

Unknown Area ◽

Global Navigation Satellite

Humanitarian Crisis scenarios typically require immediate rescue intervention. In many cases, the conditions at a scene may be prohibitive for human rescuers to provide instant aid, because of hazardous, unexpected, and human threatening situations. These scenarios are ideal for autonomous mobile robot systems to assist in searching and even rescuing individuals. In this study, we present a synchronous ground-aerial robot collaboration approach, under which an Unmanned Aerial Vehicle (UAV) and a humanoid robot solve a Search and Rescue scenario locally, without the aid of a commonly used Global Navigation Satellite System (GNSS). Specifically, the UAV uses a combination of Simultaneous Localization and Mapping and OctoMap approaches to extract a 2.5D occupancy grid map of the unknown area in relation to the humanoid robot. The humanoid robot receives a goal position in the created map and executes a path planning algorithm in order to estimate the FootStep navigation trajectory for reaching the goal. As the humanoid robot navigates, it localizes itself in the map while using an adaptive Monte-Carlo Localization algorithm by combining local odometry data with sensor observations from the UAV. Finally, the humanoid robot performs visual human body detection while using camera data through a Darknet pre-trained neural network. The proposed robot collaboration scheme has been tested under a proof of concept setting in an exterior GNSS-denied environment.

Download Full-text

Terrain changes from images acquired on opportunistic flights by SfM photogrammetry

The Cryosphere ◽

10.5194/tc-11-827-2017 ◽

2017 ◽

Vol 11 (2) ◽

pp. 827-840 ◽

Cited By ~ 12

Author(s):

Luc Girod ◽

Christopher Nuth ◽

Andreas Kääb ◽

Bernd Etzelmüller ◽

Jack Kohler

Keyword(s):

Low Cost ◽

Rapid Change ◽

Satellite System ◽

Light Transport ◽

Area Of Interest ◽

Accuracy And Precision ◽

Global Navigation Satellite ◽

The Cost ◽

Gnss Data

Abstract. Acquiring data to analyse change in topography is often a costly endeavour requiring either extensive, potentially risky, fieldwork and/or expensive equipment or commercial data. Bringing the cost down while keeping the precision and accuracy has been a focus in geoscience in recent years. Structure from motion (SfM) photogrammetric techniques are emerging as powerful tools for surveying, with modern algorithm and large computing power allowing for the production of accurate and detailed data from low-cost, informal surveys. The high spatial and temporal resolution permits the monitoring of geomorphological features undergoing relatively rapid change, such as glaciers, moraines, or landslides. We present a method that takes advantage of light-transport flights conducting other missions to opportunistically collect imagery for geomorphological analysis. We test and validate an approach in which we attach a consumer-grade camera and a simple code-based Global Navigation Satellite System (GNSS) receiver to a helicopter to collect data when the flight path covers an area of interest. Our method is based and builds upon Welty et al. (2013), showing the ability to link GNSS data to images without a complex physical or electronic link, even with imprecise camera clocks and irregular time lapses. As a proof of concept, we conducted two test surveys, in September 2014 and 2015, over the glacier Midtre Lovénbreen and its forefield, in northwestern Svalbard. We were able to derive elevation change estimates comparable to in situ mass balance stake measurements. The accuracy and precision of our DEMs allow detection and analysis of a number of processes in the proglacial area, including the presence of thermokarst and the evolution of water channels.

Download Full-text

Unsupervised Human Detection with an Embedded Vision System on a Fully Autonomous UAV for Search and Rescue Operations

Sensors ◽

10.3390/s19163542 ◽

2019 ◽

Vol 19 (16) ◽

pp. 3542 ◽

Cited By ~ 15

Author(s):

Eleftherios Lygouras ◽

Nicholas Santavas ◽

Anastasios Taitzoglou ◽

Konstantinos Tarchanidis ◽

Athanasios Mitropoulos ◽

...

Keyword(s):

Embedded System ◽

Emergency Services ◽

Vision System ◽

Open Water ◽

Satellite System ◽

Search And Rescue ◽

Human Detection ◽

Primary Role ◽

Wide Range ◽

Global Navigation Satellite

Unmanned aerial vehicles (UAVs) play a primary role in a plethora of technical and scientific fields owing to their wide range of applications. In particular, the provision of emergency services during the occurrence of a crisis event is a vital application domain where such aerial robots can contribute, sending out valuable assistance to both distressed humans and rescue teams. Bearing in mind that time constraints constitute a crucial parameter in search and rescue (SAR) missions, the punctual and precise detection of humans in peril is of paramount importance. The paper in hand deals with real-time human detection onboard a fully autonomous rescue UAV. Using deep learning techniques, the implemented embedded system was capable of detecting open water swimmers. This allowed the UAV to provide assistance accurately in a fully unsupervised manner, thus enhancing first responder operational capabilities. The novelty of the proposed system is the combination of global navigation satellite system (GNSS) techniques and computer vision algorithms for both precise human detection and rescue apparatus release. Details about hardware configuration as well as the system’s performance evaluation are fully discussed.

Download Full-text