Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite Behaviors (original) (raw)
Related papers
Game Theoretic Synthetic Data Generation for Machine Learning Based Satellite Behavior Detection
Space situational awareness (SSA) is needed to control satellite movement and space pervasiveness, which relies on quick and precise space object behavioral classification and discovery. Perhaps the biggest obstacle in adopting machine learning (ML) techniques and evaluating their performance in SSA applications is the lack of large, labeled datasets for training and validation. In this paper, we present a game theory enabled, data augmentation method, to produce datasets used by ML techniques for satellite behavior detection. Realistic sensor data (azimuth angle, elevation angle, range, range rate) are propagated using SGP4/SDP4. Then various maneuver strategies are produced by our space game model, played by on-ground radar and space objects. We use the two-player pursuit-evasion game to generate evasive maneuvering strategies for space objects. The game approach provides a method to solve SSA behavior detection problems, where the Resident Space Objects (RSO) exploits the sensing and tracking model to confuse the SSA observer by corrupting their tracking estimates, while the SSA observer improves tracking performance. The different cost functions generate different maneuver strategies. In addition, to simulating this rich training data we exploited generative adversarial networks (GANs) to further augment the simulated data. With simulated data plus GAN augmentation, various satellite behaviors are generated to produce synthetic datasets with labels for use by ML methods. Then a 3D convolutional neural network (3D-CNN) is provided the generated synthetic and labeled data with 143 possible satellite behaviors. The trained machine learning model efficiently and correctly classifies the satellite behaviors with a 97% rate. The model-based, game theoretic synthetic data, improves the training and validation performance of machine learning algorithms for satellite behavior classifications.
Convolutional Neural Network for Satellite Image Classification
2019
Multimedia applications and processing is an exciting topic, and it is a key of many applications of artificial intelligent like video summarization, image retrieval or image classification. A convolutional neural networks have been successfully applied on multimedia approaches and used to create a system able to handle the classification without any human's interactions. In this paper, we produce effective methods for satellite image classification that are based on deep learning and using the convolutional neural network for features extraction by using AlexNet, VGG19, GoogLeNet and Resnet50 pretraining models. The Resnet50 model achieves a promising result than other models on three different dataset SAT4, SAT6 and UC Merced Land. The accuracy of classification of this model for UC Merced Land dataset is 98%, for SAT4 is 95.8%, and the result for SAT6 is 94.1%.
Deep Learning for Understanding Satellite Imagery: An Experimental Survey
Frontiers in Artificial Intelligence, 2020
Translating satellite imagery into maps requires intensive effort and time, especially leading to inaccurate maps of the affected regions during disaster and conflict. The combination of availability of recent datasets and advances in computer vision made through deep learning paved the way toward automated satellite image translation. To facilitate research in this direction, we introduce the Satellite Imagery Competition using a modified SpaceNet dataset. Participants had to come up with different segmentation models to detect positions of buildings on satellite images. In this work, we present five approaches based on improvements of U-Net and Mask R-Convolutional Neuronal Networks models, coupled with unique training adaptations using boosting algorithms, morphological filter, Conditional Random Fields and custom losses. The good results—as high as AP=0.937 and AR=0.959—from these models demonstrate the feasibility of Deep Learning in automated satellite image annotation.
SmartSat Constellation - A Deep Reinforcement Learning Approach for Decentralized Coordination
2021
With rapid advancements in satellite technology, the amount of low earth orbit satellites has grown significantly which are primarily deployed for weather monitoring, earth observation or military purposes. Due to this reason, there has been an increased interest in enhancing the level of autonomy and cognition, onboard satellites to achieve optimal data collection. Optimal data is said to be collected when the satellites in a small sat constellation work together to collect information. This means that even if one of the satellites has missed out on some important information, the others can still collect them. A satellite constellation can be considered as a multi-agent reinforcement learning system. Having these agents coordinate with one another, can reduce the amount of time required to perform a task. The state-of-the-art satellite constellations follow a centralized coordination mechanism in which one primary satellite controls the rest of the satellites. This process is computationally more expensive and requires substantial communication between the satellites. It has a single point of failure and communication might be affected if the primary satellite fails. On the other hand, decentralized coordination allows agents to control their behavior themselves without the command of a supervised master. In this case, there is less inter-satellite communication which reduces the requirement for specialized onboard computational hardware. The proposal constitutes leveraging the Multi-Agent Deep Deterministic Policy Gradient [2] (MADDPG) algorithm to train the agents (satellites) to achieve optimal data collection. There are multiple use cases for the proposed solution such as illegal maritime activity tracking, natural disaster detection and assessing building damage after a natural disaster. The proposed solution focuses on tracking of ships in an extensively simulated environment for which a custom ship environment was created by leveraging OpenAI Gym [12]. By providing on-board autonomy, we aim to reduce frequent Earth Station (ES) communication significantly and enhance data collection capability.
Deep Learning-based System for Change Detection On-Board Earth Observation Small Satellites
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
In recent years, the important evolution in the number, potentiality, and diversity of Earth Observation (EO) satellites, has resulted in dramatic increases in the payload data volume and rate. However, these exponential increases in the generated data volume are creating a significant bottleneck onboard EO satellites due to transmission bandwidth limits and communications delays. Onboard imaging payload data processing can provide an appropriate solution to alleviate the induced data bottleneck. It can also facilitate rapid response for decision-making operations. Change detection is one of the most significant functions in onboard payload data processing systems that enables a real-time reaction to natural disasters such as flooding, earthquakes, and volcanic eruptions. In this paper, we address the problem of automatic change detection onboard EO satellites. This research work aims to design an automatic onboard change detection system (OCDS) that can run on existing flight-proven hardware by taking advantage of the attractive features of a leading model in deep learning called Convolutional Neural Network. The contribution of this work is twofold. An efficient algorithmic solution for change detection based on Deep Learning that fulfills space environment-induced constraints is first proposed. Second, a preliminary hardware architecture of the proposed OCDS is designed based on payload data processing flight-proven hardware. The experimental results demonstrate the efficiency of the proposed deep learningbased change detection approach and the suitability of the designed OCDS for onboarding on EO small satellites.
2021
The ability to perform near real-time data association and automatic detection and classification of resident space object (RSO) maneuvers is highly desirable. The potential benefits include anomaly resolution and change detection that may improve the accuracy of the orbital state of Earth orbiting satellites and debris. A challenging aspect of Space Situational Awareness (SSA) and Space Traffic Management (STM) is detecting an RSO maneuver based solely on observations and current knowledge of the orbital states. Modern techniques have been developed by STM researchers using statistical filtering techniques with various degrees of success. Artificial Intelligence/Machine Learning (AI/ML) techniques have seen significant growth in recent years and pose viable approaches within the space domain’s solution space to augment, supplement, and potentially replace traditional methods with varying degrees of performance. The contribution of the present work is to demonstrate the feasibility ...
Satellite Imagery Classification with Deep Learning : A Survey
International Journal of Scientific Research in Computer Science, Engineering and Information Technology, 2020
Object detection from satellite images has been a challenging problem for many years. With the development of effective deep learning algorithms and advancement in hardware systems, higher accuracies have been achieved in the detection of various objects from very high-resolution satellite images. In the past decades satellite imagery has been used successfully for weather forecasting, geographical and geological applications. Low resolution satellite images are sufficient for these sorts of applications. But the technological developments in the field of satellite imaging provide high resolution sensors which expands its field of application. Thus, the High-Resolution Satellite Imagery (HRSI) proved to be a suitable alternative to aerial photogrammetric data to provide a new data source for object detection. Since the traffic rates in developing countries are enormously increasing, vehicle detection from satellite data will be a better choice for automating such systems. In this research, a different technique for vehicle detection from the images obtained from high resolution sensors is reviewed. This review presents the recent progress in the field of object detection from satellite imagery using deep learning.
Towards Automated Satellite Conjunction Management with Bayesian Deep Learning
ArXiv, 2020
After decades of space travel, low Earth orbit is a junkyard of discarded rocket bodies, dead satellites, and millions of pieces of debris from collisions and explosions.Objects in high enough altitudes do not re-enter and burn up in the atmosphere, but stay in orbit around Earth for a long time. With a speed of 28,000 km/h, collisions in these orbits can generate fragments and potentially trigger a cascade of more collisions known as the Kessler syndrome. This could pose a planetary challenge, because the phenomenon could escalate to the point of hindering future space operations and damaging satellite infrastructure critical for space and Earth science applications. As commercial entities place mega-constellations of satellites in orbit, the burden on operators conducting collision avoidance manoeuvres will increase.For this reason, development of automated tools that predict potential collision events (conjunctions) is critical. We introduce a Bayesian deep learning approach to t...
Satellite Image Classification with Deep Learning: Survey
2020
Satellite imagery is important for many applications including disaster response, law enforcement and environmental monitoring etc. These applications require the manual identification of objects in the imagery. Because the geographic area to be covered is very large and the analysts available to conduct the searches are few, thus an automation is required. Yet traditional object detection and classification algorithms are too inaccurate and unreliable to solve the problem. Deep learning is a part of broader family of machine learning methods that have shown promise for the automation of such tasks. It has achieved success in image understanding by means that of convolutional neural networks. The problem of object and facility recognition in satellite imagery is considered. The system consists of an ensemble of convolutional neural networks and additional neural networks that integrate satellite metadata with image features.
DeepEmSat: Deep Emulation for Satellite Data Mining
Frontiers in Big Data
The growing volume of Earth science data available from climate simulations and satellite remote sensing offers unprecedented opportunity for scientific insight, while also presenting computational challenges. One potential area of impact is atmospheric correction, where physics-based numerical models retrieve surface reflectance information from top of atmosphere observations, and are too computationally intensive to be run in real time. Machine learning methods have demonstrated potential as fast statistical models for expensive simulations and for extracting credible insights from complex datasets. Here, we develop DeepEmSat: a deep learning emulator approach for atmospheric correction, and offer comparison against physics-based models to support the hypothesis that deep learning can make a contribution to the efficient processing of satellite images.