Lv
L.F.J. van der Heijden
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
2 records found
1
Missions to small bodies are increasingly gaining interest as they might hold the secrets to our solar system’s origin while some are also posing a threat to life on Earth. The small size and irregular shape result in complex dynamics complicating the close-proximity operations. Furthermore, due to the long round-trip time communication delays of up to 20 minutes can exists, excluding any required computation time on Earth. Currently used approaches either rely on detecting pre-defined landmarks on the target, detecting features and matching them to a database, or tracking craters or unknown features across images (relative navigation). However, these methods rely heavily on a-priori information, suffer from computationally intensive matching steps, or depend on the accuracy of the initial state estimate.
This work investigated the usage of a novel CNN-based pipeline that can be used to autonomously navigate accurately around asteroids. A top-down CNN-based feature detector is developed consisting of an object and keypoint detection network in sequence, which detects n pre-defined keypoints designated on the target’s 3D model using the 3D SIFT algorithm. These 2D detections alongside their 2D-3D correspondences are sent to an Efficient Perspective-n-Point (EPnP) solver that solves the Perspective-n-Points (PnP) problem. The CNN-based feature detector replaces traditional hand-engineered Image Processing (IP) algorithms as it is more robust to illumination conditions and image noise. Furthermore, the use of a CNN facilitates an offline feature selection step and as such avoid the cumbersome and computationally intensive 2D-3D matching step of the detected 2D feature to their location on the 3D model, plaguing traditional approaches. This pose estimation pipeline can be used to navigate around the asteroid up until it covers the full field of view of the camera, and it can be used to(re)-initialize the navigation filter for a relative navigation approach.
The networks have been selected based on their applicability to embedded devices and this resulted in the use of the SSD-MobileNetV2-FPN-Lite as the object detection network and the Lightweight Pose Network (LPN) model as the keypoint detection network. This lightweight CNN-based pipeline has a fraction of the parameters and Floating Point Operations (FLOPs) compared to state-of-the-art deep-learning networks and pipelines. These networks have been trained and evaluated on synthetic datasets created in this work, consisting of 32,352 images with a variety of poses for a distance of 4.5 km to 9 km from the asteroid for different illumination conditions, asteroid orientations, and image corruptions that emulate real sensor artifacts.
The pipeline could achieve a mean and median line-of-sight distance estimate of around 42 m and 30 m, respectively, at a confidence level of 90% for the large relative range, while satisfying the accuracy requirement of a maximum 10% knowledge error for 99.979% of the cases. Furthermore, the pipeline has been proven to be robust against illumination conditions, occlusions, textures, and image corruptions, mimicking effects of real sensors and the space environment. Demonstrating the efficacy of this CNN-based approach for autonomous navigation around asteroids.
...
This work investigated the usage of a novel CNN-based pipeline that can be used to autonomously navigate accurately around asteroids. A top-down CNN-based feature detector is developed consisting of an object and keypoint detection network in sequence, which detects n pre-defined keypoints designated on the target’s 3D model using the 3D SIFT algorithm. These 2D detections alongside their 2D-3D correspondences are sent to an Efficient Perspective-n-Point (EPnP) solver that solves the Perspective-n-Points (PnP) problem. The CNN-based feature detector replaces traditional hand-engineered Image Processing (IP) algorithms as it is more robust to illumination conditions and image noise. Furthermore, the use of a CNN facilitates an offline feature selection step and as such avoid the cumbersome and computationally intensive 2D-3D matching step of the detected 2D feature to their location on the 3D model, plaguing traditional approaches. This pose estimation pipeline can be used to navigate around the asteroid up until it covers the full field of view of the camera, and it can be used to(re)-initialize the navigation filter for a relative navigation approach.
The networks have been selected based on their applicability to embedded devices and this resulted in the use of the SSD-MobileNetV2-FPN-Lite as the object detection network and the Lightweight Pose Network (LPN) model as the keypoint detection network. This lightweight CNN-based pipeline has a fraction of the parameters and Floating Point Operations (FLOPs) compared to state-of-the-art deep-learning networks and pipelines. These networks have been trained and evaluated on synthetic datasets created in this work, consisting of 32,352 images with a variety of poses for a distance of 4.5 km to 9 km from the asteroid for different illumination conditions, asteroid orientations, and image corruptions that emulate real sensor artifacts.
The pipeline could achieve a mean and median line-of-sight distance estimate of around 42 m and 30 m, respectively, at a confidence level of 90% for the large relative range, while satisfying the accuracy requirement of a maximum 10% knowledge error for 99.979% of the cases. Furthermore, the pipeline has been proven to be robust against illumination conditions, occlusions, textures, and image corruptions, mimicking effects of real sensors and the space environment. Demonstrating the efficacy of this CNN-based approach for autonomous navigation around asteroids.
...
Missions to small bodies are increasingly gaining interest as they might hold the secrets to our solar system’s origin while some are also posing a threat to life on Earth. The small size and irregular shape result in complex dynamics complicating the close-proximity operations. Furthermore, due to the long round-trip time communication delays of up to 20 minutes can exists, excluding any required computation time on Earth. Currently used approaches either rely on detecting pre-defined landmarks on the target, detecting features and matching them to a database, or tracking craters or unknown features across images (relative navigation). However, these methods rely heavily on a-priori information, suffer from computationally intensive matching steps, or depend on the accuracy of the initial state estimate.
This work investigated the usage of a novel CNN-based pipeline that can be used to autonomously navigate accurately around asteroids. A top-down CNN-based feature detector is developed consisting of an object and keypoint detection network in sequence, which detects n pre-defined keypoints designated on the target’s 3D model using the 3D SIFT algorithm. These 2D detections alongside their 2D-3D correspondences are sent to an Efficient Perspective-n-Point (EPnP) solver that solves the Perspective-n-Points (PnP) problem. The CNN-based feature detector replaces traditional hand-engineered Image Processing (IP) algorithms as it is more robust to illumination conditions and image noise. Furthermore, the use of a CNN facilitates an offline feature selection step and as such avoid the cumbersome and computationally intensive 2D-3D matching step of the detected 2D feature to their location on the 3D model, plaguing traditional approaches. This pose estimation pipeline can be used to navigate around the asteroid up until it covers the full field of view of the camera, and it can be used to(re)-initialize the navigation filter for a relative navigation approach.
The networks have been selected based on their applicability to embedded devices and this resulted in the use of the SSD-MobileNetV2-FPN-Lite as the object detection network and the Lightweight Pose Network (LPN) model as the keypoint detection network. This lightweight CNN-based pipeline has a fraction of the parameters and Floating Point Operations (FLOPs) compared to state-of-the-art deep-learning networks and pipelines. These networks have been trained and evaluated on synthetic datasets created in this work, consisting of 32,352 images with a variety of poses for a distance of 4.5 km to 9 km from the asteroid for different illumination conditions, asteroid orientations, and image corruptions that emulate real sensor artifacts.
The pipeline could achieve a mean and median line-of-sight distance estimate of around 42 m and 30 m, respectively, at a confidence level of 90% for the large relative range, while satisfying the accuracy requirement of a maximum 10% knowledge error for 99.979% of the cases. Furthermore, the pipeline has been proven to be robust against illumination conditions, occlusions, textures, and image corruptions, mimicking effects of real sensors and the space environment. Demonstrating the efficacy of this CNN-based approach for autonomous navigation around asteroids.
This work investigated the usage of a novel CNN-based pipeline that can be used to autonomously navigate accurately around asteroids. A top-down CNN-based feature detector is developed consisting of an object and keypoint detection network in sequence, which detects n pre-defined keypoints designated on the target’s 3D model using the 3D SIFT algorithm. These 2D detections alongside their 2D-3D correspondences are sent to an Efficient Perspective-n-Point (EPnP) solver that solves the Perspective-n-Points (PnP) problem. The CNN-based feature detector replaces traditional hand-engineered Image Processing (IP) algorithms as it is more robust to illumination conditions and image noise. Furthermore, the use of a CNN facilitates an offline feature selection step and as such avoid the cumbersome and computationally intensive 2D-3D matching step of the detected 2D feature to their location on the 3D model, plaguing traditional approaches. This pose estimation pipeline can be used to navigate around the asteroid up until it covers the full field of view of the camera, and it can be used to(re)-initialize the navigation filter for a relative navigation approach.
The networks have been selected based on their applicability to embedded devices and this resulted in the use of the SSD-MobileNetV2-FPN-Lite as the object detection network and the Lightweight Pose Network (LPN) model as the keypoint detection network. This lightweight CNN-based pipeline has a fraction of the parameters and Floating Point Operations (FLOPs) compared to state-of-the-art deep-learning networks and pipelines. These networks have been trained and evaluated on synthetic datasets created in this work, consisting of 32,352 images with a variety of poses for a distance of 4.5 km to 9 km from the asteroid for different illumination conditions, asteroid orientations, and image corruptions that emulate real sensor artifacts.
The pipeline could achieve a mean and median line-of-sight distance estimate of around 42 m and 30 m, respectively, at a confidence level of 90% for the large relative range, while satisfying the accuracy requirement of a maximum 10% knowledge error for 99.979% of the cases. Furthermore, the pipeline has been proven to be robust against illumination conditions, occlusions, textures, and image corruptions, mimicking effects of real sensors and the space environment. Demonstrating the efficacy of this CNN-based approach for autonomous navigation around asteroids.
Clip-on Wings
Final Design report CHESTA
Bachelor thesis
(2018)
-
M. Blanke, M.E. Huistra, W.J.T. Keijer, M.M. Nelissen, P.P. van Zelst, S. Gudaal, L.C. Veldkamp, Julienne Jongbloed, M.L. Ponson, L.F.J. van der Heijden, M.M.D.J. Gillis, M.M. van Paassen, H.G. Visser, F. Avallone, E. Mooij, W.J.C. Verhagen
Multiple concepts for hybrid vehicles capable of both road and flight transport are becoming a reality. However, through incorporation of both flight and road hardware into one vehicle these designs become inefficient. The aim of this project is to design an alternative strategy for realising optimised hybrid vehicles. The objective of this project is to "Develop a personal transport vehicle suitable for commuter use on which flight hardware can be attached within 5 minutes, by 10 students within 10 weeks". This was derived from the mission statement which was to "Develop a faster and cheaper personal transport vehicle for the European market with detachable flight/road hardware that combines the advantages of road transport with flight capability". From this objective, the three most key requirements are that the vehicle shall have a road and flight configuration, that the flight hardware shall be separable from the road hardware and that the flight hardware shall be attached/detached from the road hardware within 5 minutes. When analysing the performance of the vehicle, typical flight stages and stalling configurations were investigated in terms of speed and power setting. At cruise speed, the 12 DEP (Distributed Electric Propulsion) propellers in the midsection of the wing are turned off and folded . However, for take-off and landing configuration, all the propellers are switched on. For fuel use two options are present. E10 gasoline for maximum range, or E85 for decent range, but a 70% lower eco-impact. 122 kW of power is supplied to the 6 outboard motors during cruise. The take-off distance is 450 m and the landing distance is 462 m. Two different propellers were designed and optimized for DEP and cruise conditions. To verify the noise requirements, a propeller noise analysis of the vehicle was carried out, since that is the largest contributer to overall vehicle noise. From that analysis it was concluded that the propeller noise level of the vehicle is 52.3 dBA at 1000 ft, which is low compared to other general aviation aircraft. The airfoil of the wing was selected using the design lift coefficient of 0.56. The airfoils chosen for the wing and tail are, respectively, theNACA4418 andNACA0012. Thewingwas designed to have an aspect ratio of 17, a surface area of 8.414 m2 and a span of 11.96 m. Using these values, the induced drag and pitching moment coefficients were determined. To investigate the overall efficiency of the aerodynamics of the wing, the lift over drag ratio (L/D) was calculated for each configuration. For cruise, landing and take-off, the L/D is 12.93, 7.74 and 9.48, respectively. The packaging of the vehicle resulted in the centre of gravity of the operative empty weight of the full configuration to act at 37.4% of the fuselage length at an empty mass of 1174.7 kg. The longitudinal and lateral stability and control of the vehicle were assessed. Due to the large downwash caused by the DEP propellers of the wing, a T-tail configuration was designed to move the tail away from the downwash and make it more effective. A fully-movable horizontal tail was necessary to counter the large lift coefficient and, consequently, the moment created by distributed electrical propulsion. The horizontal and vertical stabiliser were designed with a surface area of 2.42m2 and 1.1m2 respectively. After analysing the strengths and manufacturing methods of several materials it was found that the wingbox would be made using aluminium (AL7075-T6) with taper in the sheet thickness. Due to the slender wing, the weight of the wing became relatively high at 200 kg. For the skin, the most suitable material was polyester with glass fibres for its specific strength, price and compatibility with a foam or balsa core. For the linkage system, the location where the linkage is established needed to be considered. After investigating this, it was determined that the best place for the linkage would be the rear of the road hardware and the front of the flight hardware. For the linkage the Scharfenberg train coupling was used together with safety pins. The Scharfenberg connection was scaled down to suit the vehicle as it was originally designed for trains. After the final design is concluded, it is necessary to plan for future research in order to make this concept a reality. There are still uncertainties in how DEP affects the performance and aerodynamics of the concept. Therefore, it is recommended to conduct more lowspeed tests to properly establish those effects. Computational FluidDynamics (CFD) and wing tunnel testing is also recommended to visualise and analyse the fluid flow of the wing. A structural analysis using the finite elementmethod and full scale tests of themajor components are also recommended. As the coupling mechanism has never been used in aerospace application, further research into the adaptability and the integration of this mechanism should be performed. As a single-engine Private Pilot License is preferred, recommendations were also made on tests that need to be performed in order to make a case for the airworthiness authorities.
...
Multiple concepts for hybrid vehicles capable of both road and flight transport are becoming a reality. However, through incorporation of both flight and road hardware into one vehicle these designs become inefficient. The aim of this project is to design an alternative strategy for realising optimised hybrid vehicles. The objective of this project is to "Develop a personal transport vehicle suitable for commuter use on which flight hardware can be attached within 5 minutes, by 10 students within 10 weeks". This was derived from the mission statement which was to "Develop a faster and cheaper personal transport vehicle for the European market with detachable flight/road hardware that combines the advantages of road transport with flight capability". From this objective, the three most key requirements are that the vehicle shall have a road and flight configuration, that the flight hardware shall be separable from the road hardware and that the flight hardware shall be attached/detached from the road hardware within 5 minutes. When analysing the performance of the vehicle, typical flight stages and stalling configurations were investigated in terms of speed and power setting. At cruise speed, the 12 DEP (Distributed Electric Propulsion) propellers in the midsection of the wing are turned off and folded . However, for take-off and landing configuration, all the propellers are switched on. For fuel use two options are present. E10 gasoline for maximum range, or E85 for decent range, but a 70% lower eco-impact. 122 kW of power is supplied to the 6 outboard motors during cruise. The take-off distance is 450 m and the landing distance is 462 m. Two different propellers were designed and optimized for DEP and cruise conditions. To verify the noise requirements, a propeller noise analysis of the vehicle was carried out, since that is the largest contributer to overall vehicle noise. From that analysis it was concluded that the propeller noise level of the vehicle is 52.3 dBA at 1000 ft, which is low compared to other general aviation aircraft. The airfoil of the wing was selected using the design lift coefficient of 0.56. The airfoils chosen for the wing and tail are, respectively, theNACA4418 andNACA0012. Thewingwas designed to have an aspect ratio of 17, a surface area of 8.414 m2 and a span of 11.96 m. Using these values, the induced drag and pitching moment coefficients were determined. To investigate the overall efficiency of the aerodynamics of the wing, the lift over drag ratio (L/D) was calculated for each configuration. For cruise, landing and take-off, the L/D is 12.93, 7.74 and 9.48, respectively. The packaging of the vehicle resulted in the centre of gravity of the operative empty weight of the full configuration to act at 37.4% of the fuselage length at an empty mass of 1174.7 kg. The longitudinal and lateral stability and control of the vehicle were assessed. Due to the large downwash caused by the DEP propellers of the wing, a T-tail configuration was designed to move the tail away from the downwash and make it more effective. A fully-movable horizontal tail was necessary to counter the large lift coefficient and, consequently, the moment created by distributed electrical propulsion. The horizontal and vertical stabiliser were designed with a surface area of 2.42m2 and 1.1m2 respectively. After analysing the strengths and manufacturing methods of several materials it was found that the wingbox would be made using aluminium (AL7075-T6) with taper in the sheet thickness. Due to the slender wing, the weight of the wing became relatively high at 200 kg. For the skin, the most suitable material was polyester with glass fibres for its specific strength, price and compatibility with a foam or balsa core. For the linkage system, the location where the linkage is established needed to be considered. After investigating this, it was determined that the best place for the linkage would be the rear of the road hardware and the front of the flight hardware. For the linkage the Scharfenberg train coupling was used together with safety pins. The Scharfenberg connection was scaled down to suit the vehicle as it was originally designed for trains. After the final design is concluded, it is necessary to plan for future research in order to make this concept a reality. There are still uncertainties in how DEP affects the performance and aerodynamics of the concept. Therefore, it is recommended to conduct more lowspeed tests to properly establish those effects. Computational FluidDynamics (CFD) and wing tunnel testing is also recommended to visualise and analyse the fluid flow of the wing. A structural analysis using the finite elementmethod and full scale tests of themajor components are also recommended. As the coupling mechanism has never been used in aerospace application, further research into the adaptability and the integration of this mechanism should be performed. As a single-engine Private Pilot License is preferred, recommendations were also made on tests that need to be performed in order to make a case for the airworthiness authorities.