scispace - formally typeset
Search or ask a question

Showing papers presented at "IEEE Aerospace Conference in 2021"


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, a receiver architecture suitable for processing both time division (TDMA) and frequency division multiple access (FMDA) signals from Orbcomm and Iridium NEXT satellites is presented to produce Doppler frequency measurements from multi-constellation LEO satellites.
Abstract: A framework for opportunistic navigation with multi-constellation low Earth orbit (LEO) satellite signals is proposed. A receiver architecture suitable for processing both time division (TDMA) and frequency division multiple access (FMDA) signals from Orbcomm and Iridium NEXT satellites is presented to produce Doppler frequency measurements from multi-constellation LEO satellites. An extended Kalman filter (EKF)-based estimator is formulated to solve for a stationary receiver's position using the resulting Doppler measurements. Experimental results are presented showing receiver positioning with one Orbcomm satellite and four Iridium NEXT satellite with an unprecedented final position error 22.7 m.

46 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, a Doppler tracking and beacon detection framework for blind opportunistic navigation (BON) with low Earth orbit (LEO) satellite signals is proposed, which cognitively deciphers partially known signals of opportunity (SOPs) and exploits them for navigation purposes.
Abstract: A Doppler tracking and beacon detection framework for blind opportunistic navigation (BON) with low Earth orbit (LEO) satellite signals is proposed. The BON framework cognitively deciphers partially known signals of opportunity (SOPs) and exploits them for navigation purposes. When only the bandwidth and length of an Mary phase shift keying (MPSK) SOP beacon is available, the BON framework enables acquisition and tracking of terrestrial and space-based SOPs in a blind fashion. A computationally efficient algorithm is presented to blindly detect the beacon signals and estimate the Doppler frequency. The BON framework is applied to decipher the C/A pseudorandom noise (PRN) sequences from four GPS satellites. The experimental results show that the BON framework is capable of cognitively detecting the PRNs of GPS satellites with a percentage of correctly detected chips ranging between 91 % and 99 %. Another experimental example with signals from two Orbcomm LEO satellites is presented, demonstrating an unmanned aerial vehicle (UAV) navigating via the BON framework. The UAV traversed a total trajectory of 782 m, achieving a position root mean-squared error (RMSE) of 21.2 m.

35 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, a machine learning algorithm for predicting LEO satellite orbits in the STAN framework is developed, which is shown to improve the satellite tracking performance over an extended Kalman filter (EKF)-based satellite tracking approach.
Abstract: Machine learning for orbit determination of low Earth orbit (LEO) satellites in a simultaneous tracking and navigation (STAN) framework is assessed. STAN is a navigation paradigm that aims to exploit LEO satellites, which are not intended for navigation purposes. Since these satellites are not intended as navigation sources, their states (position, velocity, clock bias, and clock drift) cannot be assumed to be transmitted to the navigator. STAN estimates the states of such satellites simultaneously with the states of the navigating vehicle, using Doppler and pseudorange measurements drawn from the LEO satellite signals. This paper proposes a machine learning algorithm for predicting LEO satellite orbits in the STAN framework. A time delay neural network (TDNN) is developed, which is shown to improve the LEO satellite tracking performance over an extended Kalman filter (EKF)-based satellite tracking approach. The proposed EKF-TDNN-STAN is validated experimentally on a ground vehicle, where the Doppler measurements extracted from two Orbcomm LEO satellite signals were used to aid an on-board inertial measurement unit. In the experiment, the vehicle navigated for a total of 258 seconds, the last 30 seconds of which were in the absence of global navigation satellite system (GNSS) signals. The vehicle traversed a distance of 1.1 km during the period of GNSS unavailability. An EKF-STAN achieved a ground vehicle three-dimensional (3-D) position root mean-squared error (RMSE) of 10.6 m, while the two LEO satellites were tracked with 3-D position RMSE of 71 m and 26 m, respectively. In contrast, the proposed EKF-TDNN-STAN framework achieved a ground vehicle 3-D position root RMSE of 6.6 m, while the two LEO satellites were tracked with 3-D position RMSE of 6 m and 26 m, respectively.

18 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the authors provide a system design trade-off analysis for MMSE precoding with adaptive antennas in terms of average spectral efficiency and outage probability, showing that significant gains can be considered when increasing/reducing the overlap of signals from adjacent beams at beam edge.
Abstract: To cope with the ever increasing demand for larger throughput, Satellite Communications (SatCom) systems are evolving towards Very High Throughput Satellites (VHTS). These are mainly characterised by an increased number of beams, 4-colour frequency reuse, and adaptive antennas. Aiming at further improving the system throughput by means of aggressive frequency reuse schemes, namely full frequency reuse, linear precoding techniques have already proven the benefits for both terrestrial and satellite HTS systems. However, precoding has never been considered for implementation with adaptive antennas that are capable of varying the radiation patterns onboard the satellite and, thus, increasing or reducing the overlap of adjacent beams signals at beam edge. In this paper, we provide a system design trade-off analysis for MMSE precoding with adaptive antennas in terms of average spectral efficiency and outage probability. The proposed analysis shows that significant gains can be considered when increasing/reducing the overlap of signals from adjacent beams at beam edge. Moreover, thanks to adaptive antennas, a non-negligible benefit can also be achieved in terms of reduced transmission power on-board the satellite when more directive patterns are generated.

15 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the authors present two heuristics that, given a terrain heightmap around the rover, produce cost estimates that more effectively rank the candidate paths before ACE evaluation.
Abstract: Enhanced AutoNav (ENav), the baseline surface navigation software for NASA's Perseverance rover, sorts a list of candidate paths for the rover to traverse, then uses the Approximate Clearance Evaluation (ACE) algorithm to evaluate whether the most highly ranked paths are safe. ACE is crucial for maintaining the safety of the rover, but is computationally expensive. If the most promising candidates in the list of paths are all found to be infeasible, ENav must continue to search the list and run time-consuming ACE evaluations until a feasible path is found. In this paper, we present two heuristics that, given a terrain heightmap around the rover, produce cost estimates that more effectively rank the candidate paths before ACE evaluation. The first heuristic uses Sobel operators and convolution to incorporate the cost of traversing high-gradient terrain. The second heuristic uses a machine learning (ML) model to predict areas that will be deemed untraversable by ACE. We used physics simulations to collect training data for the ML model and to run Monte Carlo trials to quantify navigation performance across a variety of terrains with various slopes and rock distributions. Compared to ENav's baseline performance, integrating the heuristics can lead to a significant reduction in ACE evaluations and average computation time per planning cycle, increase path efficiency, and maintain or improve the rate of successful traverses. This strategy of targeting specific bottlenecks with ML while maintaining the original ACE safety checks provides an example of how ML can be infused into planetary science missions and other safety-critical software.

14 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the authors present the conceptual design of the Lunar Crater Radio Telescope (LCRT) on the far side of the Moon, where a wire mesh is deployed using wall-climbing DuAxel robots in a 3-5 km diameter crater, with a suitable depth-to-diameter ratio, to form a parabolic reflector with a 1 km diameter.
Abstract: An ultra-long-wavelength radio telescope on the far side of the Moon has significant advantages compared to Earth-based and Earth-orbiting telescopes, including: 1. Enabling observations of the Universe at wavelengths longer than 10 meters (i.e., frequencies below 30 MHz), wavelengths at which critical cosmological or extrasolar planetary signatures are predicted to appear, yet cannot be observed from the ground due to absorption from the Earth's ionosphere; and 2. The Moon acts as a physical shield that isolates a far-side lunar-surface telescope from radio interference from sources on the Earth's surface, the ionosphere, Earth-orbiting satellites, and the Sun's radio emission during the lunar night. In this paper, we present the conceptual design of the Lunar Crater Radio Telescope (LCRT) on the far side of the Moon. We propose to deploy a wire mesh using wall-climbing DuAxel robots in a 3–5 km diameter crater, with a suitable depth-to-diameter ratio, to form a parabolic reflector with a 1 km diameter. LCRT will be the largest filled-aperture radio telescope in the Solar System; larger than the former Arecibo telescope (305 m diameter, 3 cm - 1 m wavelength band, 0.3-10 GHz frequency band) and the Five-hundred-meter Aperture Spherical radio Telescope (FAST) (500 m diameter, 0.1-4.3 m wavelength band, 60–3000 MHz frequency band). LCRT's science objective is to track the evolution of the neutral intergalactic medium before and during the formation of the first stars in the 10–100 m wavelength band (3–30 MHz frequency band), which is consistent with priorities identified in the Astrophysics decadal survey. We describe LCRT's science objectives and the key technology challenges that need to be overcome to make this concept a reality. We envisage that LCRT will open a new window for humanity's exploration of the Universe.

13 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this article, the authors developed a system, with guidance, control, and estimation schemes designed to safely place an active chaser spacecraft in a parking orbit around a passive target spacecraft, where natural motion trajectories are considered to identify a set of passively safe parking orbits under Clohessy-Wiltshire-Hill dynamics.
Abstract: Autonomous rendezvous, proximity operations, and docking are key enablers of missions such as satellite servicing, active debris removal, and in-space assembly. However, errors in the control and estimation systems, or failures to account for off-nominal conditions may result in catastrophic collisions between spacecraft. Safety may potentially be preserved in these cases by switching to a safety-driven backup system. This paper develops such a system, with guidance, control, and estimation schemes designed to safely place an active chaser spacecraft in a parking orbit around a passive target spacecraft. Natural motion trajectories are considered to identify a set of passively safe parking orbits under Clohessy-Wiltshire-Hill dynamics, and a mixed integer programming formulation is developed to find the optimal transfer trajectories to this set. The practicality of the estimation and control schemes is demonstrated though simulated case studies. The guidance algorithm is integrated into a run time assurance framework, which allows real-time enforcement of the safety constraints in a least-intrusive fashion.

13 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, a system that combines a UAV, an RGB camera, an Artificial Intelligence (AI) detector trained using deep learning methods and locally collected images over a runway is presented.
Abstract: There are several ways in which Foreign Object Debris (FOD) are detected on runways. Some of these methods include Radar, infrared technologies, and stationary cameras mounted on the runway and use image processing tools to find these FODs. Radar technology is highly accurate when finding FODs but is highly inaccurate with small FOD items causing a high false-positive rate. Stationary based RGB camera-based methods also have a high false-positive rate prompting the shutdown of runways, creating disruptions for both the airport and the airline carriers. The paper presents a new method of detection by using an Unmanned Aerial Vehicle (UAV) to fly above the runway at a low altitude (e.g. < 30 m) to find FOD in. We developed a system that combines a UAV, an RGB camera, an Artificial Intelligence (AI) detector trained using deep learning methods and locally collected images over a runway. The classes specifically looked at were paper, metal, bolts, plastic, and plastic bottles. Different lighting conditions of both full sunlight and cloudy weather were taken into consideration when the images were collected. The detector was trained with various data augmentation techniques including resize, rotate, and colour augmentation. Results have concluded that there is a potential use for UAV's as a method of FOD detection, with a high rate of accuracy in the detections. This could lead to shorter timeframes and fewer disruptions where runways are closed.

11 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the electrical integrity of wide band-gap gallium nitride (GaN) heterostructure devices is evaluated subject to Venus surface atmospheric conditions, and three unique device architectures were fabricated at Stanford Nanofabrication Facility and exposed in a Venus simulation chamber for 244 hours at the University of Arkansas Center for Space and Planetary Sciences.
Abstract: Further development of harsh environment electronics capable of uncooled operation under Venus surface atmospheric conditions (~460°C, ~92 bar, corrosive) would enable future missions to the surface of Venus to operate for up to a year. Wide band-gap gallium nitride (GaN) heterostructure devices are attractive candidates for Venus lander missions due to their ability to withstand high-temperature exposure. Here, we present the first assessment of the electrical integrity of GaN-based devices subject to Venus surface atmospheric conditions. Three unique device architectures were fabricated at the Stanford Nanofabrication Facility and exposed in a Venus simulation chamber for 244 hours at the University of Arkansas Center for Space and Planetary Sciences. The three device architectures tested were InAlN/GaN high electron mobility transistors (HEMTs), InAlN/GaN Hall-effect sensors, and AlGaN/GaN UV photo detectors, which all have potential applications in the collection and readout of sensor data from Venusian landers. After exposure, HEMT threshold voltage had shifted only ~1% and gate leakage current remained on the same order of magnitude, demonstrating stability of the IrOx gate under supercritical CO2 ambient. Fluctuations in drain current after exposure are attributed to thermal detrapping and electrically-activated trapping processes. Measurements of the InAlN/GaN 2DEG properties in virgin and exposed Hall-effect sensors were comparable. Furthermore, the Hall-effect sensors exhibited a maximum change of only +11.4% in current-scaled sensitivity and −6.6% in voltage-scaled sensitivity post-exposure. The UV photodetectors with 362 nm peak wavelength exhibited an average decrease in responsivity of 38% after exposure, which is thought to be due to strain relaxation or ohmic contact degradation. Similar performance of the InAlN/GaN HEMTs and Hall-effect sensors before and after exposure highlights the viability of this material platform for development of Venus surface electronics, while the decrease in AlGaN/GaN UV photocurrent requires further analysis to assess whether the AlGaN/Ga heterostructure is suitable for robust, Venus-capable electronics.

11 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: The Neutral gas and ion mass spectrometer (NIM) is one of the six instruments of the Particle Environment Package (PEP) on board the JUICE spacecraft as discussed by the authors.
Abstract: The JUpiter ICy moons Explorer (JUICE) of the European Space Agency (ESA) will investigate Jupiter and its icy moons Europa, Ganymede and Callisto, with the aim to better understand the origin and evolution of our Solar System and the emergence of life. The Neutral gas and Ion Mass spectrometer (NIM) is one of six instruments of the Particle Environment Package (PEP) on board the JUICE spacecraft. PEP will measure neutral atoms and molecules, the ion population, and the electron population over an energy range covering from meV to MeV. The NIM instrument is designed to measure the chemical and isotope composition of the exospheres of three of Jupiter's satellites, the icy moons, both, during several flybys and during its final destination in Ganymede orbit. From measurements of the exosphere, we will derive the chemical composition of the surface, which will allow us a better understanding of the icy moons formation processes, interaction processes with the magnetospheric plasma and energetic particles of Jupiter's magnetospheric system. The NIM instrument is a compact time-of-flight mass spectrometer allowing measurements of thermal neutral molecules and ionospheric ions. To minimize the background radiation on the detector and protect electronics against the harsh radiation environment around Jupiter, elaborated radiation shielding was designed. NIM consists of two major subunits, namely, the ion-optical system and the electronics. This study presents details on the technical design and the results obtained from the calibration campaigns of different subsystems of the flight instrument including a mass range of m/z 1 to 650, a mass resolution $m/\Delta m$ of at least 750 (FWHM), and an instantaneous dynamic range of almost 6 decades. These results are discussed in detail with respect to the scientific requirements. This performance in combination with its radiation tolerance allows for both a detailed analysis of the chemical composition of Jupiter's icy moons' exospheres and ionospheres, and to explore environments, where formation of life might be possible.

11 citations


Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the authors present an approach for continuous on-board terrain reconstruction for the purpose of hazard-free landing site detection for the autonomous navigation of a Mars Science Helicopter, a JPL research concept that investigates the feasibility of flying a multikilogram science payload at various Mars science locations, with flight ranges of multiple kilometers per flight.
Abstract: Future Mars Rotorcrafts require advanced navigation capabilities to enable all terrain access for science investigations with long distance flights that are executed fully autonomously. A critical component is the ability to safely land in hazardous terrain as part of a mission, or triggered by an emergency situation. In this paper, we present an advanced navigation system for continuous on-board terrain reconstruction for the purpose of hazard-free landing site detection for the autonomous navigation of a Mars Science Helicopter - a JPL research concept that investigates the feasibility of flying a multi-kilogram science payload at various Mars science locations, with flight ranges of multiple kilometers per flight. Our approach builds on a vision-based perception system that incorporates an on-board visual-inertial state estimator augmented by a laser altimeter (range-VIO), and a structure-from-motion 3D reconstruction approach that uses a single, downward-looking camera to provide dense depth measurements while the vehicle is in motion. Depth measurements are accumulated in a local, robot-centric, multi- resolution elevation map that is analyzed by a landing site detector to extract safe landing areas below the rotorcraft, based on a heuristic that includes slope, roughness and the presence of landing hazards. Detected landing sites are prioritized by an on-board autonomy engine that either selects suitable landing sites for immediate landing maneuvers, or can explore a terrain location as part of a mission in order to find a best landing site in a pre-planned area. We demonstrate and evaluate our approach on simulated data and data acquired with a surrogate unmanned aerial system (UAS) executing flights over relevant terrain.

Proceedings ArticleDOI
Hai Wen Chen1, Neal Gross1, Ravi Kapadia1, Joseph Cheah1, Mo Gharbieh1 
06 Mar 2021
TL;DR: In this article, the authors used the You Only Look Once (YOLOv2) detection model with customized Convolutional Neural Network (CNN) feature extraction layers to extract the key features of vehicle targets from the IR images.
Abstract: Automatic Target Detection (ATD) and Recognition (ATR) are critical for video analysis and image understanding for many military and commercial applications deployed on satellites and UAV platforms. Infrared (IR) sensors can be used to detect targets during day and night time but there are few effective ATR algorithms that can exploit these sensors. Several years ago, Defense Systems Information Analysis Center (DSIAC) released an ATR Algorithm Development Image Database containing a large collection of mid-wave infrared (MWIR) imagery with multiple military and civilian vehicles as labelled targets. With the DSIAC database, we have developed AI models, which combine layers of the open-source You Only Look Once (YOLOv2) detection model with customized Convolutional Neural Network (CNN) feature extraction layers. The CNN layers are trained to extract the key features of the vehicle targets from the IR images. YOLOv2 provides target detection, classification, and localization. We have trained and tested the CNN YOLOv2 models with four different military and civilian vehicles (T72, ZSU -23-4, SUV, pickup truck) at distances ranging from 2,000m to 5,000m. Our ATR results with IR datasets show high mean average precision (mAP) between 97.25%-99.5% for day and night time images at distances of 2,000m to 5,000m. That is, we can both reliably detect and recognize different targets with only a few missed detections, and without a falsely recognized target (e.g., mistakenly classifying a civilian vehicle as a military vehicle) from as far as 5,000m. This work represents significant progress in being able to perform ATR at all times (day and night).

Proceedings ArticleDOI
06 Mar 2021
TL;DR: The Foundation Surface Habitat (FSH) is the current concept in consideration to serve as this initial surface habitat that will extend the crew mission durations and provide 30-60 day habitability for a crew of four allowing for the astronauts to explore farther and longer on each visit to the lunar surface as mentioned in this paper.
Abstract: NASA has been tasked with implementing a bold vision of the future for human spaceflight including expanding the commercial market and operations in low-Earth orbit (LEO); launching the world's most powerful rocket and deep space crew spacecraft; incrementally establishing a sustainable presence on and around the Moon, starting by landing the first woman and the next man on the surface in 2024 while, in parallel, constructing an orbital Gateway in cislunar space. The key piece of establishing a sustainable presence in deep space is the development of habitation systems that will not only extend mission operations, but provide for living quarters that will keep the crew happy and healthy throughout their expeditions. Beyond the Gateway habitation needs, these capabilities will need to be defined and advanced to support the initial lunar surface missions and to prepare for human missions to the Mars system. The Foundation Surface Habitat (FSH) is the current concept in consideration to serve as this initial surface habitat that will extend the crew mission durations. It will provide 30–60 day habitability for a crew of four allowing for the astronauts to explore farther and longer on each visit to the lunar surface. The Transit Habitat is the current concept under study that would be capable of supporting long-duration missions. These missions could include extended operations at the Gateway or as a free-flyer facility in low-Earth orbit but ultimately, the Transit Habitat is envisioned as the crew habitat that would transport humans on long-duration deep space missions. The Transit Habitat integrated with an advanced propulsion system would serve as the in-space transportation system tasked with safely ferrying humans to and from Mars. Each of these habitation concepts is currently under study internal to NASA, but the agency is also working closely with U.S. industry through the Next Space Technologies for Exploration Partnerships (NextSTEP) activity to understand their concepts for a commercially-provided FSH and Transit Habitat as well as close coordination with our international partners to understand their desires for in-space and surface habitation. This paper will provide a status of these concepts and partnership activities as well as potential future development paths and architecture plans.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this article, the authors presented the adaptation of the Kinematic-Model-Free (KMF) robot control method to a three-section, nine degrees-of-freedom continuum manipulator for both planar and spatial task spaces.
Abstract: Continuum robots have strong potential for application in Space environments. However, their modeling is challenging in comparison with traditional rigid-link robots. The Kinematic-Model-Free (KMF) robot control method has been shown to be extremely effective in permitting a rigid-link robot to learn approximations of local kinematics and dynamics (“kinodynamics”) at various points in the robot's task space. These approximations enable the robot to follow various trajectories and even adapt to changes in the robot's kinematic structure. In this paper, we present the adaptation of the KMF method to a three-section, nine degrees-of-freedom continuum manipulator for both planar and spatial task spaces. Using only an external 3D camera, we show that the KMF method allows the continuum robot to converge to various desired set points in the robot's task space, avoiding the complexities inherent in solving this problem using traditional inverse kinematics. The success of the method shows that a continuum robot can “learn” enough information from an external camera to reach and track desired points and trajectories, without needing knowledge of exact shape or position of the robot. We similarly apply the method in a simulated example of a continuum robot performing an inspection task on board the ISS.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this article, a charge management system (CMS) for the Gravitational Reference Sensor (GRS) being used in the LISA mission has been presented, which is capable of generating stable and robust square UV light pulses.
Abstract: Gravitational reference sensors (GRSs) are imperative to Earth geodesy missions and gravitational wave observations in space. A typical GRS consists of a test mass (TM) surrounded by a capacitive electrode housing to perform sensitive relative position measurements and to apply small forces to the TM. This paper specifically discusses advancements in the charge management system (CMS) for the GRS being used in the LISA mission. Space radiation accumulating charge on the TM will eventually generate unwanted forces on the TM due to stray electric fields in the spacecraft. Thus, the TM charge must be kept close to a zero potential. The TM charge will be controlled in a contact-free manner by shining UV light and exploiting the photoelectric effect. A major design improvement for future missions is using UV LEDs, which can be pulsed. This facilitates more advanced charge control schemes for a continuous science measurement. The UV LEDs are housed in an aluminum block and controlled with supporting electronics via the charge management device (CMD). The CMD needs to be integrated with the spacecraft computer and needs to contain redundancy to survive the 10+ year LISA mission. The CMD is a NASA deliverable for the ESA mission and has begun the process of technology advancement and testing. The unit has custom PCBs designed to supply both continuous and pulsed current to the UV LEDs, readback telemetry data, manage CMD power needs, and synchronize with the spacecraft computer to communicate with spacecraft operators. The system achieved TRL 4 at the end of 2018 and surpassed all requirements for performance, redundancy, and lifetime. The system is capable of generating stable and robust square UV light pulses, has the capacity to drive the UV LEDs at their full dynamic range, and meets requirements on power, pulse properties, stability, and commanding speed. This technology features novel discharge methods and improvements on past missions in terms of noise level and continuous science measurements. The CMD design and testing results in this study are critical to the success of LISA, future Earth geodesy missions, and future science missions relying on precision inertial sensor technology.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: Sheaves that can work over directed graphs such as temporal flow networks, a sheaf representation for Dijkstra's algorithm, and a construction for routing sheaves capable of modeling multicast scenarios are developed.
Abstract: The goal of Delay Tolerant Networking (DTN) is to take a collection of heterogeneous, disparate connections between satellites, space assets, ground stations, and ground infrastructure and bring it together into a cohesive, functioning overlay network. Depending on the systems being considered, one can find links with a one-way light time exceeding minutes (and hours), periodic links which can sometimes be predicted by orbital mechanics, and restrictions based on the variety of capabilities built into these systems. These characteristics preclude traditional network models and routing techniques and have classically led to either rigid routing tables or purely probabilistic models. As the deeper underlying structures remain unknown, development of more DTN-optimized algorithms has lacked the necessary foundation. In a continuation of previous work, the goal of this paper is to identify and study these fundamental structures that exist in delay tolerant networks (DTN), with a focus on space networks. The current routing methodology has been to use contact graph routing (CGR) algorithms. CGR models a series of known contacts as a static graph. For CGR to work, this graph must be globally consistent and must have an accurate picture of the network. Because this is a globally controlled structure, there is little room for flexibility in the event of changes to the network which would naturally occur as the network grows. As a response to the desire for flexibility as the network changes, we introduced the mathematical structure known as sheaves to DTNs last year. The tag-line for sheaves is that they are a mathematically precise way of gluing local data together into unique global data. Thus, sheaves lend extra power to traditional models (and routing algorithms) by taking additional information and merging it, in as consistent a manner as possible, with the representation itself. The clearest example of how Earth-bound networks exhibit behavior that is “sheafy” is link state routers, which build a local-to-global picture of their network by gluing local information together into a global network, exactly as a sheaf would do. For routing within delay tolerant networks to truly exploit this structure, a deeper structure than a graph is required. In this paper, we develop sheaves that can work over directed graphs such as temporal flow networks, we construct a sheaf representation for Dijkstra's algorithm, and we outline a construction for routing sheaves capable of modeling multicast scenarios. Finally, there is a section of future work suggesting follow-on research.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: The CORALS (Characterization of Ocean Residues And Life Signatures) instrument as discussed by the authors is a transformative mass spectrometer that comprises a solid-state UV laser source, custom ion transfer optics, and a high performance Orbitrap™ mass analyzer.
Abstract: Europa is a high-priority astrobiology target due to the presence of liquid water, carbon-rich materials, and energy sources that may support complex chemistry and the emergence of biological activity. The CORALS (Characterization of Ocean Residues And Life Signatures) instrument—a transformative mass spectrometer that comprises a solid-state UV laser source, custom ion transfer optics, and a high performance Orbitrap™ mass analyzer—is capable of comprehensive analyses of planetary materials that can provide important context for the origin and evolution of potential biosignatures and geologic icy matrices on Europa. The CORALS laser source, ion inlet system, and mass analyzer constitute a highly versatile and low SWaP (Size: 11,000 cm3; Weight: 8.0 kg; and Power: 41 W peak) mass spectrometer. Here we report on the design of the CORALS engineering test unit (ETU), which will be qualified for spaceflight via random vibration testing and exposure to dry heat microbial reduction, and test results from the two pathfinding prototypes that have informed the development of the ETU. The demonstrated analytical performance of the CORALS instrument supports the wide range of science goals and planetary targets this spectrometer can access, highlighting the instrument's multidimensional strengths in the search for life signatures on Europa or elsewhere in the Solar System.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this article, the effects of a virtual reality intervention on spaceflight-validated exercise protocols using a prototype rowing ergometer designed to operate within the constraints of future long-duration exploration missions (LDEM) were examined.
Abstract: The exploration of space will require ever-increasing exposure to microgravity environments. The human response to this exposure has been categorized and mitigated via countermeasures, principally exercise. However, additional constraints to future mission design minimizes the allotted space and modalities for exercise, creating a risk for psychological fatigue, a reduction in motivation, and a suite of other categorical factors that could, taken together, present a risk for reduced adherence to the countermeasures and/or mission performance. Thus, the current study will examine the effects of a virtual reality (VR) intervention on spaceflight-validated exercise protocols using a prototype rowing ergometer designed to operate within the constraints of future long-duration exploration missions (LDEM). The Integrated Resistance and Aerobic Training (dubbed “SPRINT”) protocol will be used in conjunction with a combination flywheel and resistance training device (M-MED) utilized in prior bedrest studies. The SPRINT protocol trades exercise duration for intensity, providing similar benefits to existing countermeasures while reducing time spent on exercise. The M - MED permits resistance training on the muscles most effected by microgravity exposure on the same device used to train cardiovascular function, thus reducing the volume and weight requirements of the exercise countermeasure. It is upon this framework that we will add the VR rowing simulation. VR has shown to be a lightweight, reliable, and enjoyable technology in numerous studies, while exergaming has been shown to improve measures of motivation and adherence. We will create a rowing simulation that can integrate with a rowing ergometer and any exercise protocol, and then implement it on the M-MED with SPRINT. The simulation will feature virtual teammates, virtual competitors, and other gaming mechanisms that encourage a user to maintain a prescribed heart rate intensity in a way that aims to maximize factors associated with enjoyment and adherence. We plan to conduct a within-subjects experiment on an astronaut-like population. Subjects will be randomly assigned to VR or non-VR in their initial experiment, complete the SPRINT protocol on the M-MED, break for a one-month minimum washout period, then return to complete the protocol again in the other group. As a pilot study, dependent variables have been selected broadly. Physical and psychological outcomes are to be measured alongside adherence to and motivation toward this very challenging protocol. Additional measures are to be made of virtual presence, preexisting bias toward or against VR, and personality traits, which may influence a preference for or against VR. Preliminary data on non-VR subjects shows increasing measures of state-trait anxiety, negative feelings toward the exercise, and amotivation from the start to the end of the protocol. It is hypothesized that overall attitudes toward the protocol will improve with the VR intervention as indicated by metrics of adherence, motivation, affect, and mood restoration. The results of this study will inform future designs of exercise combined with VR applications for implementation during LDEM.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the authors compare results from the fusion of histograms to that of fusion of confusion matrices developed from data of the same modality and that of a cross modality.
Abstract: Data fusion from a variety of sources requires alignment, association, and analysis. One method to determine the relationship between two variables measuring the same information is a correlation analysis. The canonical variates analysis (CVA) supports the assessments of two sets of data. In this paper, we compare results from the fusion of histograms to that of the fusion of confusion matrices developed from data of the same modality and that of a cross modality. We use the Confusion Matrix Fusion (CMF) approach in the analysis and compare the results for EO/RF fusion. In the analysis, the Experiments, Scenarios, Concept of Operations, and Prototype Engineering (ESCAPE) data set is used for comparison to previous aerospace results.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this article, the authors proposed an artificial intelligence technique based on two novel neural networks, the growing neural networks (GNN) and variable sequence LSTM (VarLSTM) model to automate the process of diagnosis, prognosis, and health monitoring (DPHM) for aerospace systems.
Abstract: Due to the increase in complexity in aerospace systems, developing a diagnosis, prognosis, and health monitoring (DPHM) framework is a challenge that must be considered to assure the safety of such systems. This paper discusses this problem by proposing an artificial intelligence technique based on two novel neural networks, the growing neural networks (GNN) and variable sequence LSTM (VarLSTM) model to automate the process of DPHM for aerospace systems. For single-unit datasets, the proposed model estimates a Health Index value using the residuals between the measured telemetry data and the one predicted using the GNN algorithm, and then the HI value is extrapolated for prognostics. For multiple-units datasets, the model makes RUL predictions by directly mapping the RUL of the training units to their corresponding measured features at every measured instant. In this paper, the model optimizes the architecture of a recurrent neural network and was used to make RUL predictions for aircraft engines and detect failure for satellite attitude actuators (Reaction Wheels). It was tested on the CMAPSS and PHM08 aircraft engine datasets (multiple-unit datasets) simulated by NASA, and it was able to make RUL predictions with root mean square errors as low as 14 engine cycles. Another application to test the proposed model was on the Kepler Spacecraft's reaction wheels from which two have failed (single-unit datasets). The model detected the failure of the two failed reaction wheels by estimating a HI value which indicates the probability of failure of the reaction wheels using the residuals between the speed predictions made by the model and measured speed values. Failure was detected using the model almost 105 days and 54 days for reaction wheels two and four respectively. Prognostics were also applied on the Kepler Mission reaction wheels and RUL predictions were made with mean absolute errors ranging between 2–13 days depending on how close the reaction wheel is to fail when the prediction is made. The proposed artificial intelligence algorithm shows promising results in system fault diagnosis and prognosis leading to the development of smart systems for aerospace applications.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: The Framework for Robust Execution and Scheduling of Commands On-Board (FRESCO) as discussed by the authors is the result of lessons learned in developing a software architecture to enable autonomous solar system exploration and generalizes this work to offer a modular, software-agnostic approach to developing verifiable architecture for autonomous space systems.
Abstract: Achieving the science exploration and defense goals of the following decades will require flight systems capable of operations with limited operator contact, system mode changes and retasking based on sensor data, and complex robotic operations. To support these capabilities, increasingly autonomous flight systems are required that can perform dedicated mission functions, e.g. payload targeting and communications, and system-level functions, e.g. planning and goal monitoring. Architecting an autonomous system requires a well-reasoned, self-consistent framework to avoid ad hoc design choices that will introduce complexity and risk. The Framework for Robust Execution and Scheduling of Commands On-Board, FRESCO, is the result of lessons learned in developing a software architecture to enable autonomous solar system exploration. FRESCO generalizes this work to offer a modular, software-agnostic approach to developing verifiable architecture for autonomous space systems. FRESCO specifies guiding principles, functions, interfaces, and interactions from which mission-specific autonomous control architectures can be derived. FRESCO is a principled framework relying on explicit, state-based goal definitions, centralized management of state knowledge, clearly separated control boundaries, and hierarchical reasoning. Using components from FRESCO reference architecture, an autonomous decision-making architecture can be designed for spacecraft which can then be mapped to flight software architecture. FRESCO is flexibly defined to enable autonomous control of flight systems built using extensive software and hardware heritage. Finally, FRESCO-derived architectures support a spectrum of operator/spacecraft interactions, ranging from traditional commanding to goal-driven commanding with the ability to change mission goals autonomously. FRESCO has been used in defining the autonomy architectures for the ASTERIA mission and have been demonstrated in laboratory and software simulation for small body rendezvous and in-space servicing missions.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: The Absolute and Relative Trajectory Measurement System (ARTMS) as mentioned in this paper is a framework for autonomous navigation of spacecraft swarms around planetary bodies using angles-only measurements from onboard cameras.
Abstract: This paper presents and demonstrates an algorithmic framework for autonomous navigation of spacecraft swarms around planetary bodies, using angles-only measurements from onboard cameras. Angles-only methods are compelling as they reduce reliance on external measurement sources. However, prior demonstrations have faced limitations, including 1) inability to treat more than one observer and target in a swarm, 2) lack of autonomy and reliance on external state information, and 3) treatment of only Earth-orbit scenarios. The new Absolute and Relative Trajectory Measurement System (ARTMS) overcomes these challenges and consists of three core modules leveraging novel algorithms: Image Processing, which tracks and identifies targets in images and computes their bearing angles; Batch Orbit Determination, which computes a swarm state initialization from angles-only measurements; and Sequential Orbit Determination, which uses an unscented Kalman filter to refine the swarm state estimate, seamlessly fusing measurements from multiple observers to achieve the autonomy, robustness and distribution needed for deep space navigation. Theoretical performance of ARTMS is investigated through a quantitative observability analysis of multi-observer angles-only navigation in Mars orbit. For swarms of at least 3 spacecraft and at least 2 observers, the complete swarm state is observable. After two orbits, the absolute orbit is estimated to within 1 km, target ranges are estimated to within 0.5%, and other relative state components are estimated to 0.02% of target range. Clock offsets are estimated to within 0.05s. These accuracies are validated with camera-in-the-loop simulations of a four-spacecraft swarm taking distributed measurements in an eccentric Mars orbit. ARTMS displays robust navigation across a variety of formations and under challenging conditions, and achieves the necessary performance to support the proposed objectives.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: The proposed CyberBERT solution is proposed, the first deep session-based recommender system to employ bidirectional transformers to model the intent of anonymous users within a session, and achieves state-of-the-art measured by F1 score on the Windows PE Malware API sequence dataset, and state of theart for P@20 and MRR@20 on YOOCHOOSE 1/64.
Abstract: Session-based recommendation is the task of predicting user actions during short online sessions. The user is considered to be anonymous in this setting, with no past behavior history available. Predicting anonymous users' next actions and their preferences in the absence of historical user behavior information is valuable from a cybersecurity and aerospace perspective, as cybersecurity measures rely on the prompt classification of novel threats. Our offered solution builds upon the previous representation learning work originating from natural language processing, namely BERT, which stands for Bidirectional Encoder Representations from Transformers (Devlin et al., 2018). In this paper we propose CyberBERT, the first deep session-based recommender system to employ bidirectional transformers to model the intent of anonymous users within a session. The session-based setting lends itself to applications in threat recognition, through monitoring of real-time user behavior using the CyberBERT architecture. We evaluate the efficiency of this dynamic state method using the Windows PE Malware API sequence dataset (Catak and Yazi, 2019), which contains behavior for 7107 API call sequences executed by 8 classes of malware. We compare the proposed CyberBERT solution to two high-performing benchmark algorithms on the malware dataset: LSTM (Long Short-term Memory) and transformer encoder (Vaswani et al., 2017). We also evaluate the method using the YOOCHOOSE 1/64 dataset, which is a session-based recommendation dataset that contains 37,483 items, 719,470 sessions, and 31,637,239 clicks. Our experiments demonstrate the advantage of a bidirectional architecture over the unidirectional approach, as well as the flexibility of the CyberBERT solution in modelling the intent of anonymous users in a session. Our system achieves state-of-the-art measured by F1 score on the Windows PE Malware API sequence dataset, and state-of-the-art for P@20 and MRR@20 on YOOCHOOSE 1/64. As CyberBERT allows for user behavior monitoring in the absence of behavior history, it acts as a robust malware classification system that can recognize threats in aerospace systems, where malicious actors may be interacting with a system for the first time. This work provides the backbone for systems that aim to protect aviation and aerospace applications from prospective third-party applications and malware.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the authors present a decoupled approach for the quadrotor and the soft gripper, combining a geometric controller and a minimum-snap trajectory optimization for the drone base, with a quasi-static finite element model and control-space interpolation for the gripper.
Abstract: Manipulation and grasping with unmanned aerial vehicles (UAVs) currently require accurate positioning and are often executed at reduced speed to ensure successful grasps. This is due to the fact that typical UAV s can only accommodate rigid manipulators with few degrees of freedom, which limits their capability to compensate for disturbances caused by the vehicle positioning errors. Moreover, UAV s have to minimize external contact forces in order to maintain stability. Biological systems, on the other hand, exploit softness to overcome similar limitations, and leverage compliance to enable aggressive grasping. This paper investigates control and trajectory optimization for a soft aerial manipulator, consisting of a quadrotor and a tendon-actuated soft gripper, in which the advantages of softness can be fully exploited. To the best of our knowledge, this is the first work at the intersection between soft manipulation and UAV control. We present a decoupled approach for the quadrotor and the soft gripper, combining (i) a geometric controller and a minimum-snap trajectory optimization for the quadrotor (rigid) base, with (ii) a quasi-static finite element model and control-space interpolation for the soft gripper. We prove that the geometric controller asymptotically stabilizes the quadrotor velocity and attitude despite the addition of the soft load. Finally, we evaluate the proposed system in a realistic soft dynamics simulator, and show that: (i) the geometric controller is fairly insensitive to the soft payload, (ii) the platform can reliably grasp unknown objects despite inaccurate positioning and initial conditions, and (iii) the decoupled controller is amenable for real-time execution. Video Attachment: https://youtu.be/NNpQxP0SPFk

Proceedings ArticleDOI
06 Mar 2021
TL;DR: This paper describes the elicitation, formal specification, and analysis of general collision avoidance system requirements for a conceptual spacecraft conducting autonomous close-proximity operations based on a run time assurance construct, the first formally specified and analyzed generalizedrun time assurance architecture for spacecraft that includes a fault monitor, interlock monitor, and human-machine interface.
Abstract: One of the greatest challenges preventing the use of advanced controllers in aerospace is developing methods to verify, validate, and certify them with high assurance. One emerging approach is to push the burden of assurance from offline verification of an autonomous controller at design time, to online verification of safe behavior through a monitor and high assurance backup controller at run time. Run time assurance goes a step beyond alerting systems by detecting imminent unsafe behavior and intervening with a trusted control response. In the spacecraft domain, autonomous operations could be approved if run time assurance systems can provide collision avoidance capabilities. While several approaches to run time assurance have been developed and successfully demonstrated, the design and verification of these systems is ad hoc and specific to the application. This paper describes the elicitation, formal specification, and analysis of general collision avoidance system requirements for a conceptual spacecraft conducting autonomous close-proximity operations based on a run time assurance construct. This includes the first formally specified and analyzed generalized run time assurance architecture for spacecraft that includes a fault monitor, interlock monitor, and human-machine interface. Mathematically precise requirements are elicited through the process of formal specification based on common design elements, spacecraft guidance constraints in the literature, and a structured hazard assessment. Finally, the requirements are analyzed using compositional reasoning and formal model checking verification techniques.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this article, the authors compare the utility of virtual reality interfaces and how they can be employed in human-supervised robot applications so that we may move towards more intuitive and easy-to-use interfaces for control and interaction.
Abstract: In the last decade, there have been great advancements in virtual reality (VR) resulting in its availability for everyday consumers. As VR becomes more ubiquitous, there is an opportunity to utilize this technology to create intuitive operator interfaces for interaction with complex dynamic systems, such as humanoid robots. As evidenced in the DARPA Robotics Challenge (DRC), current interfaces for humanoids primarily use a standard computer setup with monitor, keyboard, and mouse requiring operators to process 3D data with 2D devices. And although these interfaces can be very capable in operating a robot, they are often complex and require expert operators as well as extensive training. However, this paradigm can be changed with VR by allowing operators to visualize and interact with 3D data in a 3D environment, allowing for a more natural interaction. In this paper, we present our work on converting a typical interface to a virtual reality interface for NASA's humanoid robot, Valkyrie. We compare our standard computer interface and our VR interface for Valkyrie, as well as the shared control planners and system architecture that make our interfaces possible. The goal of this work is to better understand the utility of virtual reality interfaces and how they can be employed in human-supervised robot applications so that we may move towards more intuitive and easy-to-use interfaces for control and interaction.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this article, the authors proposed an overall mission architecture for performing multiple on-orbit servicing missions by a fleet of free-flying single-arm space manipulators in the form of single-flyers.
Abstract: Similar to any service or product, industrialization of On-Orbit Servicing (OOS) demands performance enhancement through introducing relevant autonomy elements in planning and executing single and multiple servicing missions. This paper proposes an overall mission architecture for performing multiple on-orbit servicing missions by a fleet of servicers in the form of free-flying single-arm space manipulators. The architecture targets to improve the two key industrialization criteria of resource and service. In the far-range rendezvous with target satellites, the servicers burn most of their fuel. Furthermore, the time that servicers spend in transfer orbits determines the approximate duration of a servicing mission. Hence, as part of resource management, the presented architecture first identifies the main contributors to the fuel consumption and mission duration in far-range rendezvous phase of OOS missions being: (i) the location of the parking orbit, (ii) the type of transfer trajectories, and (iii) the dispatch scheduling. As the result, separate optimization loops are considered for minimizing the mission costs, across the OOS industry. Servicers are suggested to form an equally phased constellation in a parking orbit close to Sun-synchronous orbits in the Low Earth Orbital (LEO) region, where 57.5% of operational LEO satellites reside. A satellite in the parking orbit constellation is named “Administrator”, whose sole purpose is to plan and manage servicing missions. The Administrator determines the optimal number and sequence of servicing missions that must be performed by the available servicers, and the optimal transfer trajectories servicers shall follow to reach the targets. Upon completion of their missions, each servicer returns to the parking orbit and occupies the available position that requires the lowest fuel consumption to enter. In almost 90% of servicers' lifetime, they are in an idle state in the parking orbit awaiting dispatch or in transfer orbits. To enhance quality of the provided service, the proposed architecture suggests effective use of this time to task servicers with performing machine learning that helps improve the functionality of their guidance, navigation and control systems in upcoming missions. The task involves trajectory learning for a servicer's manipulator system in free-floating regime to reach a simulated moving target while avoiding (virtual) obstacles and compensating for environmental disturbances. Both supervised and unsupervised machine learning techniques are considered, and based on a qualitative analysis, the unsupervised DDPG algorithm is deemed most applicable in the free-floating trajectory learning task.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the authors present a detailed look at possible trajectories in launch years from 2030 to 2040 with flyout speeds at least twice that of Voyager 1 and 2, and significant opportunities for tuning the flyout direction to maximize the heliophysics return as well as allow encounters with outer planets or Kuiper Belt Objects.
Abstract: Since the beginning of space exploration, one of the most ambitious goals has been to explore beyond the boundaries of our solar system. Ground and Earth-orbit based systems have given a deep understanding of the overall characteristics of the heliosphere in the local interstellar medium and how the characteristics of our solar system are similar to and different from other systems. Viewing the heliosphere from outside will allow, for the first time, a more complete understanding of how a star system evolves and interacts with the Universe. Interstellar missions have been studied for decades. The primary reasons we have not yet explored this region are critical limitations in technology. These include a lack of propulsion that can achieve the high speeds needed to get to the heliospheric boundary in reasonable time, reliable systems that can function for the long lifetime needed, reasonable communications capabilities at interstellar range, and constraints on mission resources such as power when more than 100 AU from the Earth. Recent developments in launch systems, execution of long-lived missions such as New Horizons, new radioisotope power systems, and advanced communications systems have for the first time allowed for a practical, feasible near-term mission that can achieve the goal of exploring outside the solar system. We present recent results of a concept study that examined possible missions that could be launched as early as 2030 using existing, or near-existing, technology. These possible missions support significant payloads for heliospheric science, with the potential for additions to the payload for planetary investigations or astrophysics. The concept study includes a detailed look at possible trajectories in launch years from 2030 to 2040 with flyout speeds at least twice that of Voyager 1 and 2, and significant opportunities for tuning the flyout direction to maximize the heliophysics return as well as allow encounters with outer planets or Kuiper Belt Objects. We present a summary of trade studies performed to investigate the constraints and design space for an interstellar probe. These trades include a comparison of trajectories that include gravity assists at Jupiter and at the Sun to increase speed, optimization of the telecommunications architecture to balance data downlink rate with power usage, and spacecraft control methods to allow precise pointing for telecommnications while minimizing propellant usage for an extremely long-lived mission. We present a spacecraft design that can support the potential payloads designed to operate reliably for 50 years, while allowing for communications from 1000 AU.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this article, the authors present research findings in the following areas critical to validating the effectiveness of providing required 5G access to the drones with security, reliability, and spectral efficiency: 1) Radio coverage for the drone corridor by adding a separate set of antennas for coverage in the air while the conventional sets of antennas continue to provide coverage on the ground.
Abstract: 5G can provide the multiplicative capacity gains needed to support a large number of drones/UAS (Unmanned Aircraft Systems). 5G cellular networks with newly available millimeter wave (mmWave) frequency bands can provide wireless communication links for control as well as data traffic for drones and drone swarms. Drones are becoming increasingly important for commercial uses such as delivery and transportation as well as for public safety search and rescue of natural disaster victims, surveillance of remote critical infrastructure, surveys of environmental quality in protected regions, and detection of threats during major public events. This paper presents research findings in the following areas critical to validating the effectiveness of providing required 5G access to the drones with security, reliability, and spectral efficiency: 1) Radio coverage for the drone corridor by adding a separate set of antennas for coverage in the air while the conventional set of antennas continues to provide coverage on the ground. Beam transmission and validation with ray-tracing simulations are covered. 2) Optimization of uplink communication from a swarm of drones with a single mmWave beam by grouping the drones with power allocations for non-orthogonal multiple access (NOMA). 3) Optimization of the network lifetime of a swarm of drones resulting in suitable trajectories in the presence of interference. 4) Methods including precoding that can enhance physical layer security with channel information about the interference source. The paper concludes with plans for future research to provide further scientific basis for the proposed cellular drone network.

Proceedings ArticleDOI
06 Mar 2021
TL;DR: In this paper, the authors focus on the fetch part of the Mars Sample Return (MSR) and more specifically the problem of autonomously detecting and localizing sample tubes deposited on the Martian surface.
Abstract: A potential Mars Sample Return (MSR) architecture is being jointly studied by NASA and ESA. As currently envisioned, the MSR campaign consists of a series of 3 missions: sample cache, fetch and return to Earth. In this paper, we focus on the fetch part of the MSR, and more specifically the problem of autonomously detecting and localizing sample tubes deposited on the Martian surface. Towards this end, we study two machine-vision based approaches: First, a geometry-driven approach based on template matching that uses hard-coded filters and a 3D shape model of the tube; and second, a data-driven approach based on convolutional neural networks (CNNs) and learned features. Furthermore, we present a large benchmark dataset of sample-tube images, collected in representative outdoor environments and annotated with ground truth segmentation masks and locations. The dataset was acquired systematically across different terrain, illumination conditions and dust-coverage; and benchmarking was performed to study the feasibility of each approach, their relative strengths and weaknesses, and robustness in the presence of adverse environmental conditions.