A5078367630- AI-based Problem Solving and Planning
- Distributed systems and fault tolerance
- Distributed and Parallel Computing Systems
- Constraint Satisfaction and Optimization
- Planetary Science and Exploration
- Spacecraft Design and Technology
- Scientific Computing and Data Management
- Optimization and Search Problems
- Astro and Planetary Science
- Reservoir Engineering and Simulation Methods
- Space Satellite Systems and Control
- Robotic Path Planning Algorithms
- Underwater Vehicles and Communication Systems
- Modular Robots and Swarm Intelligence
- Space Exploration and Technology
- Satellite Communication Systems
- Service-Oriented Architecture and Web Services
- Logic, Reasoning, and Knowledge
- Real-Time Systems Scheduling
- Model-Driven Software Engineering Techniques
- Atmospheric and Environmental Gas Dynamics
- Privacy-Preserving Technologies in Data
- Advanced Computational Techniques and Applications
- Machine Learning and Algorithms
- Scheduling and Optimization Algorithms
Jet Propulsion Laboratory
2016-2025
California Institute of Technology
2011-2024
Google (United States)
2018-2024
UCLouvain
2021
Harvey Mudd College
2021
Johns Hopkins University
2018
Washington State University
2009
Engineering Arts (United States)
2003
University of North Texas
1996
Space Information Laboratories (United States)
1993
A membership inference attack allows an adversary to query a trained machine learning model predict whether or not particular example was contained in the model's training dataset. These attacks are currently evaluated using average-case "accuracy" metrics that fail characterize can confidently identify any members of set. We argue should instead be by computing their true-positive rate at low (e.g., ≤ 0.1%) false-positive rates, and find most prior perform poorly when this way. To address...
Abstract Europa is a premier target for advancing both planetary science and astrobiology, as well opening new window into the burgeoning field of comparative oceanography. The potentially habitable subsurface ocean may harbor life, globally young comparatively thin ice shell contain biosignatures that are readily accessible to surface lander. Europa’s icy also offers opportunity study tectonics geologic cycles across range mechanisms compositions. Here we detail goals mission architecture...
ML models are ubiquitous in real world applications and a constant focus of research. At the same time, community has started to realize importance protecting privacy training data. Differential Privacy (DP) become gold standard for making formal statements about data anonymization. However, while some adoption DP happened industry, attempts apply complex still few far between. The is hindered by limited practical guidance what protection entails, guarantees aim for, difficulty achieving...
Because learning sometimes involves sensitive data, machine algorithms have been extended to offer differential privacy for training data. In practice, this has mostly an afterthought, with privacy-preserving models obtained by re-running a different optimizer, but using the model architectures that already performed well in non-privacy-preserving setting. This approach leads less than ideal privacy/utility tradeoffs, as we show here. To improve these prior work introduces variants of weaken...
NASA's Perseverance rover uses robotic autonomy to achieve its mission goals on Mars. Its self-driving autonomous navigation system (AutoNav) has been used evaluate 88% of the 17.7-kilometer distance traveled during first Mars year operation. Previously, maximum total evaluated was 2.4 kilometers by Opportunity 14-year lifetime. AutoNav set multiple planetary records, including greatest driven without human review (699.9 meters) and single-day drive (347.7 meters). The Autonomous Exploration...
Deploying multi-satellite constellations for Earth observation requires coordinating potentially hundreds of spacecraft. With increasing onboard capability autonomy, we can view the constellation as a multi-agent system (MAS) and employ decentralized scheduling solutions. We analyze problem (COSP) formulate it distributed constraint optimization (DCOP). COSP scalable inter-agent communication computation consists millions variables which, coupled with assumptions structure, make existing...
We discuss the various features of Earth-observing sensorWeb developed by Jet Propulsion Laboratory and Goddard Space Flight Center, which provide key science data about eruptions within hours for volcanologists around world. Onboard AI software evaluates request, orients spacecraft, operates instruments to acquire high-resolution images with hyperspectral analysis.
Numerous automated and semi-automated planning & scheduling systems have been developed for space applications. Most of these are model-based in that they encode domain knowledge necessary to predict spacecraft state resources based on initial conditions a proposed activity plan. The as often modeled series timelines, with timeline or set timelines represent resource key the operations spacecraft. In this paper, we first describe basic representation can state, resource, timing, transition...
Deep space missions can benefit from onboard image analysis. We demonstrate deep learning inference to facilitate such analysis for future mission adoption. Traditional flight hardware provides modest compute when compared today's laptop and desktop computers. New generations of commercial off the shelf (COTS) processors designed embedded applications, as Qualcomm Snapdragon Movidius Myriad X, deliver significant in small Size Weight Power (SWaP) packaging offer direct acceleration neural...
In this work we address the practical challenges of training machine learning models on privacy-sensitive datasets by introducing a modular approach that minimizes changes to algorithms, provides variety configuration strategies for privacy mechanism, and then isolates simplifies critical logic computes final guarantees. A key challenge is algorithms often require estimating many different quantities (vectors) from same set examples --- example, gradients layers in deep architecture, as well...
This article is a part of the Special Issue on Intelligent Systems for Space Exploration. The Payload Experiment (IPEX) CubeSat that flew from December 2013 through January 2015 and validated autonomous operations onboard instrument processing product generation Module Hyperspectral Infrared Imager (HyspIRI) mission concept. IPEX used several artificial intelligence technologies. First, machine learning computer vision in its processing. machine-learned random decision forests to classify...
Article An autonomous spacecraft agent prototype Share on Authors: Barney Pell Caelum Research, NASA Ames Research Center, MS 269/2, Moffett Field, CA CAView Profile , Douglas E. Bernard Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, Steve A. Chien Erann Gat Nicola Muscettola Recom Technologies, Centre, Mofett P. Pandurang Nayak Michael D. Wagner Fourth Planet, 155A Park Suite 104, Sunnyvale, Brian C. Williams Authors Info & Claims AGENTS '97:...
A number of successful applications automated planning and scheduling to spacecraft operations have recently been reported in the literature. However, these one-of-a-kind that required a substantial amount development effort. In this paper, we describe ASPEN (Automated Planning/Scheduling Environment), modular, reconfigurable application framework which is capable supporting wide variety applications. We architecture ASPEN, as well current control/operations progress.
This work addresses mission planning for autonomous underwater gliders based on predictions of an uncertain, time-varying current field. Glider submersibles are highly sensitive to prevailing currents so planners must account ocean tides and eddies. Previous in variable-current path assumes that perfect, but practice these forecasts may be inaccurate. Here we evaluate plan fragility using empirical tests historical which followup data is available. We present methods glider control a A case...
We describe a timeline-based scheduling algorithm developed for mission operations of the EO-1 earth observing satellite. first range operational constraints focusing on maneuver and thermal that cannot be modeled in typical planner/schedulers. then greedy heuristic compare its performance to both prior - documenting an over 50% increase scenes scheduled with estimated value millions dollars US. also relaxed optimal scheduler showing produces schedules scene count within 15% upper bound schedules.
A central problem in environmental sensing and monitoring is to classify/label the hotspots a large-scale field. This paper presents novel decentralized active robotic exploration (DARE) strategy for probabilistic classification/labeling of Gaussian process (GP)-based In contrast existing state-of-the-art strategies learning field maps, time needed solve DARE independent map resolution number robots, thus making it practical situ, real-time sampling. Its behavior exhibits an interesting...