Deep Learning for Automated Fish Detection from Sonar Images
Sonar image classification with Convolution Neural Network (CNN)
Engineering topics
Publications and source records attributed to Mueller, Robert.
Sonar image classification with Convolution Neural Network (CNN)
Abstract Objective Study the effects of downstream passage through a novel turbine designed for fish safety, the Restoration Hydro Turbine (RHT), on American Eels Anguilla rostrata in a recirculating turbine test facility. Methods A 55‐cm‐diameter RHT was operated under 10 m of hydraulic head and 667 revolutions/min. In total, 131 eels were passed through the turbine and 43 eels were used as experimental controls (length = 33.9–65.5 cm). High‐speed video of passage through the runner region was captured for 89% of turbine‐passed eels, and injury and behavioral effects were recorded immediately before and after passage, as well as after a 48‐h holding period. A subset of 37 eels was additionally examined with X‐ray imaging for internal injuries. Result The 48‐h survival rate for both treatment and control groups was 100%, with no major internal or external injuries present after the holding period. Conclusion This is a substantial improvement over eel survival rates through conventional Kaplan and Francis turbines, which may range from 40% to 95%, and suggests that hydropower turbines designed for safe downstream fish passage could be implemented without major impacts to eels.
There are two strategies to lower overall project costs to an extent that will make many potential sustainable hydropower sites economically viable: (1) design standardized/modular components; (2) use advanced tools to reduce environmental evaluation costs. In this study an autonomous sensor device (Sensor Fish) was used to study a Francis turbine retrofitted with a modular guide vane. The median nadir pressures measured were 74.7, 66.6, and 56.6 kPaA for 90-, 190-, and 380-kW operating conditions respectively. These nadir pressures were compared to other Francis turbines studied using Sensor Fish and were found to be within the same range. The proportion of Sensor Fish releases with severe acceleration events (acceleration = 95G) was also investigated. The proportion ranged from 73 to 80% (runner region), 50 to 64% (guide vane region), and 9 to 28% (draft tube region), which was within the range of the other turbines used for comparison. The Sensor Fish testing that was conducted at Hurley Dam demonstrates that the modular guide vane that was retrofitted to the existing Francis turbine is potentially a suitable replacement that can provide biological performance similar to the guide vane used with other existing Francis turbines, but with the benefit of reduced fabrication costs.
Freshwater eels, such as the American eel (Anguilla rostrata) present numerous challenges related to safe downstream fish passage at hydroelectric facilities. One of those challenges is effective monitoring of their abundance, movements, and behavior to facilitate design and operation of eel protection and passage facilities. A previous EPRI study documented the ability of human analysts to reliably identify American eels in data obtained with a 1100/1800 kHz, multibeam sonar. This report describes a project to develop deep learning (a subset of artificial intelligence) tools to automate the time-consuming, subjective process of eel identification in multibeam sonar data. The project exploited new data collected in the laboratory and the existing data from the prior EPRI field study to develop and test deep learning and other data analytic tools, including wavelet filtering, differencing for static object removal, and convolutional neural network analysis. The analysis of the laboratory data demonstrated feasibility of the approach, revealed object characteristics observed with the sonar that distinguish eels from similarly sized and shaped acoustic targets, and provided additional data for algorithm selection and training. Deep learning algorithms trained and tested on the laboratory data alone achieved accuracy rates of greater than 98% when classifying acoustic images of eels and similar-sized neutrally buoyant sticks. The algorithm trained and tested on the pre-existing field data alone, and yielded classification accuracy of 9.3% false positives and 13.3% false negatives when distinguishing between eels and sticks/PVC pipes based on video clips (i.e., multiple, consecutive images). This performance is comparable to the classification accuracy achieved by human analysts in the prior study. The deep learning algorithm trained on a combination of video clips obtained in the laboratory and the field and tested on video clips from the field, was able to distinguish eels from sticks and PVC pipes (a river debris analog) of similar size with 100% accuracy. Outreach to the hardware, software, and end-user communities early in the project helped to identify needs and specify the application space. Outreach to those communities at the end of the project communicated project results and opportunities for further development. The project achieved proof of concept for automated identification of eel in multibeam sonar data. Future work should focus on acquisition of additional data for more robust algorithm training and testing; modification of the software tools to accommodate multiple acoustic targets in the acoustic field at a given time; identification of additional object classes; incorporation of motion in the object identification and classification algorithms; operationalizing the software tools, including integration with other existing sonar data analysis tools; and partnering with hardware and software providers for distribution of the software tools with their commercial products.
Mars water resources mining and processingLunar ISRU feed forward/risk reduction to MarsMars in-situ constructionLunar ISRU feed forward/risk reduction to MarsWhat does ISRU still need from Mars science missions?
NASA's strategic goal is to put humans on Mars in the 2030s. The NASA Human Spaceflight Architecture Team (HAT) and NASA Mars Design Reference Architecture (DRA) 5.0 has determined that in-situ resource utilization (ISRU) is an essential technology to accomplish this mission. Additive construction technology using in-situ materials from planetary surfaces will reduce launch mass, allow structures to be three dimensionally (3D) printed on demand, and will allow building designs to be transmitted digitally from Earth and printed in space. This will ultimately lead to elimination of reliance on structural materials launched from Earth (zero launch mass of construction consumables). The zero launch mass (ZLM) 3D print head project addressed this need by developing a system that 3D prints using a mixture of in-situ regolith and polymer as feedstock, determining the optimum mixture ratio and regolith particle size distribution, developing software to convert g-code into motion instructions for a FANUC robotic arm, printing test samples, performing materials testing, and printing a reduced scale habitable structure concept. This paper will focus on the ZLM 3D Print Head design, materials selection, software development, and lessons learned from operating the system in the NASA KSC Swamp Works Granular Mechanics & Regolith Operations (GMRO) Laboratory.
We investigated mercury accumulation in Pacific lamprey and sediments in the Columbia River basin. Mercury concentrations in larval lamprey differed significantly among sample locations (P < 0.001) and were correlated with concentrations in sediments (r 2 = 0.83), whereas adult concentrations were highly variable (range 0.1–9.5 µg/g) and unrelated to holding time after collection. The results suggest that Pacific lamprey in the Columbia River basin may be exposed to mercury levels that have adverse ecological effects.
ISRU of Mars resources was baselined in 2009 Design Reference Architecture (DRA) 5.0, but only for Oxygen production using atmospheric CO2. The Methane (LCH4) needed for ascent propulsion of the Mars Ascent Vehicle (MAV) would need to be brought from Earth. However: Extracting water from the Martian Regolith enables the production of both Oxygen and Methane from Mars resources: Water resources could also be used for other applications including: Life support, radiation shielding, plant growth, etc. Water extraction was not baselined in DRA5.0 due to perceived difficulties and complexity in processing regolith. The NASA Evolvable Mars Campaign (EMC) requested studies to look at the quantitative benefits and trades of using Mars water ISRUPhase 1: Examined architecture scenarios for regolith water retrieval. Completed October 2015. Phase 2: Deep dive of one architecture concept to look at end-to-end system size, mass, power of a LCH4/LO2 ISRU production system
Integrate In Situ Resource Utilization (ISRU) sub-systems and examine advanced capabilities and technologies to verify Mars 2024 Forward architecture precursor pathfinder options: Integrated spacecraft/surface infrastructure fluid architecture: propulsion, power, life support center dot Power system feed and propellant scavenging from propulsion system center dot High quality oxygen for life support and EVA Fluid/cryogenic zero-loss transfer and long-term storage center dot Rapid depot-to-rover/spacecraft center dot Slow ISRU plant-to-ascent vehicle Integration of ISRU consumable production center dot Oxygen only from Mars atmosphere carbon dioxide center dot Oxygen, fuel, water, from extraterrestrial soil/regolith Test bed to evaluate long duration life, operations, maintenance on hardware, sensors, and autonomy
Current space telerobotic systems are constrained to only operating in bright light and dust-free conditions. This project will study the effects of difficult lighting and dust conditions on telerobotic perception systems to better assess and refine regolith operations on other neighboring celestial bodies. In partnership with Embry-Riddle Aeronautical University and Caterpillar, Inc., optical, LiDAR and RADAR sensing equipment will be used in performing the study. This project will create a known dust environment in the Swamp Works Granular Mechanics & Regolith Operations (GMRO) Laboratory regolith test bin to characterize the behavior of the sensing equipment in various calibrated lighting and dust conditions. It will also identify potential methods for mitigating the impacts of these undesirable conditions on the performance of the sensing equipment. Enhancing the capability of telerobotic perception systems will help improve life on earth for those working in dangerous, dusty mining conditions, as well as help advance the same technologies used for safer self-driving automobiles in various lighting and weather conditions. It will also prove to be a critical skill needed for advancing robotic and human exploration throughout our solar system, for activities such as mining on an asteroid or pioneering the first colony on Mars.
This project seeks to develop a small free flyer that can be used to safely and effectively prospect on an Asteroid while being controlled by the crew. This will enable the characterization of the Asteroid for the In Situ Resource Utilization (ISRU). Lava tubes can be explored remotely from the outside Asteroids can contain vast amounts of resources such as water for propellants and metals for feed stocks. Lava Tubes on Mars and the Moon may contain frozen volatile resources. Before the resources can be used, they must be found with a prospecting method. The NASA Agency Asteroid Grand Challenge seeks new ideas for Asteroid retrieval mission technologies for exploration and utilization of asteroids in a Distant Retrograde Orbit (DRO). This project will develop a small free flying platform that can be used to safely and effectively prospect on an Asteroid with limited autonomy while being controlled by the crew. This will enable the characterization of the Asteroid for ISRU. Lava tubes can be explored remotely from the outside as well using this same technology.
Providing routine, affordable access to a variety of orbital and deep space destinations requires an intricate network of ground, planetary surface, and space-based spaceports like those on Earth (land and sea), in various Earth orbits, and on other extraterrestrial surfaces. Advancements in technology and international collaboration are critical to establish a spaceport network that satisfies the requirements for private and government research, exploration, and commercial objectives. Technologies, interfaces, assembly techniques, and protocols must be adapted to enable mission critical capabilities and interoperability throughout the spaceport network. The conceptual space mission architecture must address the full range of required spaceport services, from managing propellants for a variety of spacecraft to governance structure. In order to accomplish affordability and sustainability goals, the network architecture must consider deriving propellants from in situ planetary resources to the maximum extent possible. Water on the Moon and Mars, Mars' atmospheric CO2, and O2 extracted from lunar regolith are examples of in situ resources that could be used to generate propellants for various spacecraft, orbital stages and trajectories, and the commodities to support habitation and human operations at these destinations. The ability to use in-space fuel depots containing in situ derived propellants would drastically reduce the mass required to launch long-duration or deep space missions from Earth's gravity well. Advances in transformative technologies and common capabilities, interfaces, umbilicals, commodities, protocols, and agreements will facilitate a cost-effective, safe, reliable infrastructure for a versatile network of Earth- and extraterrestrial spaceports. Defining a common infrastructure on Earth, planetary surfaces, and in space, as well as deriving propellants from in situ planetary resources to construct in-space propellant depots to serve the spaceport network, will reduce exploration costs due to standardization of infrastructure commonality and reduction in number and types of interfaces and commodities.
No abstract available
Faults in wiring systems are a serious concern for the aerospace and aeronautic (commercial, military, and civilian) industries. Circuit failures and vehicle accidents have occurred and have been attributed to faulty wiring created by open and/or short circuits. Often, such circuit failures occur due to vibration during vehicle launch or operation. Therefore, developing non-intrusive fault-tolerant techniques is necessary to detect circuit faults and automatically route signals through alternate recovery paths while the vehicle or lunar surface systems equipment is in operation. Electrical connector concepts combining dust mitigation strategies and cable diagnostic technologies have significant application for lunar and Martian surface systems, as well as for dusty terrestrial applications. The dust-tolerant intelligent electrical connection system has several novel concepts and unique features. It combines intelligent cable diagnostics (health monitoring) and automatic circuit routing capabilities into a dust-tolerant electrical umbilical. It retrofits a clamshell protective dust cover to an existing connector for reduced gravity operation, and features a universal connector housing with three styles of dust protection: inverted cap, rotating cap, and clamshell. It uses a self-healing membrane as a dust barrier for electrical connectors where required, while also combining lotus leaf technology for applications where a dust-resistant coating providing low surface tension is needed to mitigate Van der Waals forces, thereby disallowing dust particle adhesion to connector surfaces. It also permits using a ruggedized iris mechanism with an embedded electrodynamic dust shield as a dust barrier for electrical connectors where required.
High launch costs and mission requirements drive the need for low mass excavators with mobility platforms, which in turn have little traction and excavation reaction capacity in low gravity environments. This presents the need for precursor and long term future missions with low mass robotic mining technology to perform In-Situ Resource Utilization (ISRU) tasks. This paper discusses a series of experiments that investigate the effectiveness of a percussive digging device to reduce excavation loads and thereby the mass of the excavator itself. The goal of percussive excavation is to fluidize dry regolith in front of the leading edge of the tool by mechanically separating the microscopic interlocking grains resulting in a reduced force needed to shear the soil. There are several variables involved with this technique; this experiment varied: Impact energy, frequency, and excavation speed and held constant: impact direction, depth of cut, angle of tool, and soil bulk density. The test apparatus consisted of an aluminum truss bridge with a central pivoting arm. Attached to the arm was a winch with a load cell in line that recorded the tension in the cable and therefore the excavation load. The arm could be adjusted for excavation depth which was recorded along with the arm angle relative to the bridge. A percussive mechanism and 30" wide pivoting bucket were attached at the end of the arm simulating a basic backhoe with a percussion direction tangent to the direction of . movement. Internally the mechanism used a set of die springs and barrel cam to produce the percussive blow. By changing the springs and the speed of the motor the impact energy and frequency of percussion could be varied independently. Impact energies from 11.2J to 30.5J and frequencies from 0 BPM to 700 BPM were investigated. A reduction in excavation force of as much as 51% was achieved in this experimental investigation. Smaller percussive digging implements, tested by others, have achieved a reduction of as much as 72%. This paper will examine the effects of impact energy, frequency, scaling and their effect on excavation forces in a dry granular material such as lunar regolith. The past several years have shown an increasing interest in mining space resources both for exploration and commercial enterprises. This work studied the benefits and risks of percussive excavation and prelimin~ry results indicate that this technique may become an enabling technology for extra-terrestrial excavation of regolith and ice.
High launch costs and mission requirements drive the need for low mass excavators with mobility platforms, which in turn have little traction and excavation reaction capacity in low gravity environments. This presents the need for precursor and long term future missions with low mass robotic mining technology to perform In-Situ Resource Utilization (ISRU) tasks. This paper discusses a series of experiments that investigate the effectiveness of a percussive digging device to reduce excavation loads and thereby the mass of the excavator itself. A percussive mechanism and 30" wide pivoting bucket were attached at the end of the arm simulating a basic backhoe with a percussion direction tangent to the direction of movement. Impact energies from 13.6J to 30.5J and frequencies from 0 BPM to 700 BPM were investigated. A reduction in excavation force of as much as 50% was achieved in this experimental investigation.
A lightweight bulldozer blade prototype has been designed and built to be used as an excavation implement in conjunction with the NASA Chariot lunar mobility platform prototype. The combined system was then used in a variety of field tests in order to characterize structural loads, excavation performance and learn about the operational behavior of lunar excavation in geotechnical lunar simulants. The purpose of this effort was to evaluate the feasibility of lunar excavation for site preparation at a planned NASA lunar outpost. Once the feasibility has been determined then the technology will become available as a candidate element in the NASA Lunar Surface Systems Architecture. In addition to NASA experimental testing of the LANCE blade, NASA engineers completed analytical work on the expected draft forces using classical soil mechanics methods. The Colorado School of Mines (CSM) team utilized finite element analysis (FEA) to study the interaction between the cutting edge of the LANCE blade and the surface of soil. FEA was also used to examine various load cases and their effect on the lightweight structure of the LANCE blade. Overall it has been determined that a lunar bulldozer blade is a viable technology for lunar outpost site preparation, but further work is required to characterize the behavior in 1/6th G and actual lunar regolith in a vacuum lunar environment.
This paper gives an update on the performance of the Mars Exploration Rovers (MER) which have been continually performing for more than 3 years beyond their original 90-day missions. The paper also gives the latest results on the optimization of a multijunction solar cell that is optimized to give more power on the surface of Mars.