Poster Gallery for Session 2 (February 10th): Autonomy

Moderator: John Day, JPL

📑 Combined PDF of posters for this session

Poster 23 Thumbnail

Machine Learning for Detection of Life-Like Motality in Liquid Samples

Authors
Steffen Mauceri, Mark Wronkiewicz, Jake Lee, Jack Lightholder, Gary Doran, Lukas Mandrake
Theme
Autonomy
Poster ID
23

Abstract: Jupiter’s moon Europa and Saturn’s moon Enceladus are expected to contain a liquid water ocean beneath their icy shells. All life we know relies on water, making these moons especially important targets for future missions to explore. To support such a mission, JPL is working on a suite of instruments capable of searching for life in liquid water samples called the Ocean Worlds Life Surveyor (OWLS). Finding life in these oceans could finally answer the question whether we are alone in the universe.

Poster 24 Thumbnail

Construction Rovers

Authors
Kendall Johnson
Theme
Autonomy
Poster ID
24

Abstract: In July 2019, I attended the Lunar Operations and Technologies to Enable Human Exploration of Mars and the Moon Workshop. It is a group of NASA scientists who believe that leveraging lunar technologies will allow us to be more successful in getting to Mars. They also believe in creating devices that can work on both celestial bodies will save time, money, and energy in our inevitable journey into the cosmos. One of the main priorities stated for helping to enable Moon and Mars exploration is specified rovers that can complete tasks involving science, maintenance, transportation, and construction (Thronson, 2019). NASA’s Martian rover team has done a great job concerning a scientific rover, the classic Lunar rover can transport humans on the surface of the Moon, and currently, there is nothing to maintain on any celestial body, other than Earth, for a maintenance rover. The last of this Lunar Operations and Technologies task is a construction rover, and this is where I have started my theories and experiments to demonstrate actions, hard data, possibilities, shortcomings, and abilities of a future construction rover by building and testing a prototype. The first task of a construction rover is to best prepare the celestial body's surface for the arrival of humans. The rover is given the task of preparing the surface that can include clearing away regolith, flattening the ground, and moving equipment and debris. This is a demonstration of the proof of concept for the construction rover as an autonomous platform using Artificial Intelligence (AI) and Machine Learning (ML) for the rover moving debris. Adding AI to the construction rover will best open the door to the system being fully autonomous. The AI of the construction rover will consist of separate trained Artificial Neural Network (ANN) for sets of different tasks. An ANN is an AI technique that uses the concept of a biological neuron to weigh parameters for predicting. The first ANN has been trained for the rover's movement based on incoming sensor data producing states for movement. The next ANN is a Convolution Neural Network (CNN) that I am using for object classification, detection, and recognition. A CNN is an ANN, but with the added benefit of a convolutional layer that can take into account the surrounding pixels along with the original pixel into the training of the neural network. The rover's CNN will be trained on images of which objects to carry and which objects to push regarding the AI platform's proposed tasks. The last ANN will be used for reinforcement learning. This ANN is not very different from the first other than it will be trained in a simulation to carry out the important task of path planning. Each ANN gives the AI platform the ability to make its own decisions from what it sees through its camera and reads through its sensors. In the demonstration of the AI construction rover, I hope to show it successfully moving an object using these three AI algorithms as one. I further hope that NASA can use the concepts of these AI algorithms as a tool for the next step of rover programs, and getting humans more opportunities on other worlds.

Poster 25 Thumbnail

Automating validation of satellite‐derived ice‐cover features: Discriminating ice objects in optical ice images with different degrees of local texture distortions

Authors
Ekaterina Kim, Nabil Panchi, Ole-Magnus Pedersen, Sveinung Løset, Anirban Bhattacharyya
Theme
Autonomy
Poster ID
25

Abstract: The amount of data from satellite missions is expected to increase. In fact, this number increases faster than the capacity of experts to process, adequately validate and evaluate the uncertainty of the results. Despite rapid progress in machine learning, the methods and standards for automated interpretation of sea ice imagery remain underdeveloped. One such field is the automated interpretation of ice imagery from ground operations, especially under poor visibility conditions (e.g., imagery from surface vessels, shore stations, etc.). There is a strong need for robust and efficient methods enabling the automated processing of close‐range sea ice imagery to aid in the derivation of useful characteristics of sea ice cover (ice types, concentration, decay).

Poster 26 Thumbnail

A Machine Learning Approach to Low Earth Orbit Satellite Health and Safety Telemetry

Authors
Zhenping Li
Theme
Autonomy
Poster ID
26

Abstract: (Not available.)

Poster 27 Thumbnail

Absolute Localisation for surface robotics in GPS denied locations using a Neural Network

Authors
AS Chung, P Ludivig, RWK Potter, T Seabrook, B Wu
Theme
Autonomy
Poster ID
27

Abstract: (Not available.)

Poster 29 Thumbnail

Using Planet Cubesat Imagery for a Dynamic Environmental Sensor Web

Authors
James Mason, Steve Chien, Jim Boerkoel, Daniel Wang, Ashley Davies, Joel Mueting, Vivek Vittaldev, Vishwa Shah, Ignacio Zuleta
Theme
Autonomy
Poster ID
29

Abstract: (Not available.)

Poster 30 Thumbnail

Evaluation of Path Planning Algorithms Using a Simulation Platform for Autonomous Surface Vessels

Authors
Anete Vagale, Robin T Bye, Ottar L Osen
Theme
Autonomy
Poster ID
30

Abstract: Problem: Improved safety while navigating on waters and reduction of col- lision risk is a vital part of the guidance, navigation and control system of an autonomous surface vehicle. Another problem is, how to compare the performance of existing path plan- ning and collision avoidance algorithms in a unified way. Solution: To tackle these problems, a novel evaluation simulator platform (ESP) is proposed in this poster for simulation-based testing of algorithms.

Poster 32 Thumbnail

Automated Scheduling for the ECOSTRESS Mission

Authors
Amruta Yelamanchili, Steve Chien, Alan Moy, Elly Shao, Michael Trowbridge, Kerry Cawse-Nicholson, Jordan Padams, Dana Freeborn
Theme
Autonomy
Poster ID
32

Abstract: The ECOSystem Thermal Radiometer Experiment on Space Station (ECOSTRESS) mission seeks to better understand how much water plants need and how they respond to stress. ECOSTRESS measures the temperature of plants to understand combined evaporation and transpiration, known as evapotranspiration.

Poster 33 Thumbnail

Automated Scheduling for the Orbiting Carbon Observatory-3 Mission

Authors
Amruta Yelamanchili, Christopher Wells, Alan Moy, Steve Chien, Annmarie Eldering, Ryan Pavlick
Theme
Autonomy
Poster ID
33

Abstract: The Orbiting Carbon Observatory-3 (OCO-3) is a NASA instrument for measuring atmospheric CO₂. OCO-3 launched to on May 4, 2019 to the ISS (International Space Station) on a SpaceX Falcon 9 rocket as part of a resupply mission. It is mounted on the International Space Station on the Japanese Experiment Module – Exposed Facility (JEM-EF). It is expected to begin nominal science operations in August 2019 and its planned mission duration is three years. OCO-3 will enable identification of CO₂ sources and sinks and study changes in CO₂ levels over time. Automated scheduling is being deployed for operations of OCO-3. The OCO-3 scheduling process begins with a mostly-automated dynamic science priority assignment that is input to an automated scheduling of area targets, calibration targets, nadir, and glint mode. It is also being used to schedule observations for the calibration of the pointing mirror.

Poster 16 Thumbnail

Front Delineation and Tracking with Multiple Underwater Vehicles

Authors
Andrew Branch, Mar M Flexas, Martina Troesch, Brian Claus, Andrew F Thompson, Yanwu Zhang, Evan B. Clark, Steve Chien, David M Fratantoni, James C Kinsey, Brett Hobson, Brian Kieft, Francisco P Chavez
Theme
Autonomy
Poster ID
16

Abstract: Space based remote sensing provides great information about ocean dynamics. However, remote sensing information is generally limited to measuring the ocean surface or the upper layer of the ocean. Ocean models can further augment this information. However, in order to probe the immense volume of the ocean most accurately generally requires marine vehicles such as autonomous underwater vehicles (AUVs), Seagliders, profiling buoys, and surface vehicles sampling in-situ. Deploying and operating these assets is very expensive. This means there is a very limited number of marine vehicles compared to the massive size of the ocean. Knowing where the assets should be deployed and operated is very difficult

Poster 17 Thumbnail

Autonomous Nested Search for Hydrothermal Venting

Authors
Andrew Branch, James Mason, Guangyu Xu, Michael V Jakuba, Christopher R German, James McMahon, Steve Chien, James C Kinsey, Andrew D Bowen, Kevin P Hand, Jeffrey S Seewald
Theme
Autonomy
Poster ID
17

Abstract: (Not available.)