Difference between revisions of "Sequoia"
| (2 intermediate revisions by the same user not shown) | |||
| Line 2: | Line 2: | ||
Sequoia is an open-source, 3U CubeSat that will demonstrate on-board image classification and processing with updateable machine learning models. The goal of the project is to obtain a high volume of scientifically important imagery for ecological and climatology research. Researchers have no need for images saturated with clouds or uninteresting areas—so why not filter them out with convolutional neural networks? We will retrain Sequoia’s deep learning with images taken by the satellite, uplinking improvements. The Stanford Student Space Initiative is developing deep learning models for forest fire risk assessment and detection and a number of other applications. The mission architecture is user-definable with the operator specifying desirable image locations or types and resolutions, and the satellite maximizing delivery of fully open-source images. | Sequoia is an open-source, 3U CubeSat that will demonstrate on-board image classification and processing with updateable machine learning models. The goal of the project is to obtain a high volume of scientifically important imagery for ecological and climatology research. Researchers have no need for images saturated with clouds or uninteresting areas—so why not filter them out with convolutional neural networks? We will retrain Sequoia’s deep learning with images taken by the satellite, uplinking improvements. The Stanford Student Space Initiative is developing deep learning models for forest fire risk assessment and detection and a number of other applications. The mission architecture is user-definable with the operator specifying desirable image locations or types and resolutions, and the satellite maximizing delivery of fully open-source images. | ||
==Mission Goals== | ==Mission Goals== | ||
| + | Official mission goals to be released soon... | ||
==Timeline== | ==Timeline== | ||
| Line 16: | Line 17: | ||
===Software=== | ===Software=== | ||
| − | + | Sequoia's software systems are logically split into two areas. There is a flight computer, the PyCubed, which controls all the low-level flight control functionality, including receiving commands from the ground, downlinking telemetry, running most GNC algorithms, controlling the power systems, and supervising our payload. The payload is a Raspberry Pi SoC that is equipped with three cameras and an S-Band radio. It collects images, processes them using ML models specific to a given research objective, and downlinks the results and images over its high-bandwidth radio. | |
===Guidence Navigation and Control=== | ===Guidence Navigation and Control=== | ||
Official summary to be released soon... | Official summary to be released soon... | ||
Latest revision as of 21:20, 13 September 2020
🌲Mission Overview🌲
Sequoia is an open-source, 3U CubeSat that will demonstrate on-board image classification and processing with updateable machine learning models. The goal of the project is to obtain a high volume of scientifically important imagery for ecological and climatology research. Researchers have no need for images saturated with clouds or uninteresting areas—so why not filter them out with convolutional neural networks? We will retrain Sequoia’s deep learning with images taken by the satellite, uplinking improvements. The Stanford Student Space Initiative is developing deep learning models for forest fire risk assessment and detection and a number of other applications. The mission architecture is user-definable with the operator specifying desirable image locations or types and resolutions, and the satellite maximizing delivery of fully open-source images.
Mission Goals
Official mission goals to be released soon...
Timeline
Full Post-Covid 19 Pandemic Timeline to be released soon...
Subsystems
Structures
The Sequoia CubeSat bus is being developed with COTS components and simple manufacturing techniques for cheap scalability. Our primary frame consists of four 6061 aluminum angle iron pieces joined by accurate, cheap laser-cut sheet components. Protolabs generously supplied the manufacturing for this project. The secondary structures incorporate 3D printed nylon, giving us lightweight, highly customizable mounting options. The frame was simulated to a 5x safety factor given the vibration, shock, and static loads placed on it to ensure adequate strength.
Avionics
Sequoia utilizes the PyCubed flight computer, developed and flight proven by the Robotic Exploration Lab at Stanford University. PyCubed and its dual watchdog timers will command a Raspberry Pi 4b running Linux for image classification and processing. PyCubed was chosen because of its open-source nature, careful COTS design paradigm, and popularity amongst CubeSat developers. PyCubed's use of CircuitPython, while at times challenging to work with, drives reliability over top-level spacecraft command and increases the probability of mission success. The power system relies on triple junction space grade solar panels for energy capture. Sequoia will use a 2S4P battery configuration, with COTS Lithium Ion cells. Sequoia will have two communications systems—a UHF LoRa radio with an omnidirectional dipole antenna for command and control, and a custom S-band with a directed antenna for imagery and large data downlink.
Software
Sequoia's software systems are logically split into two areas. There is a flight computer, the PyCubed, which controls all the low-level flight control functionality, including receiving commands from the ground, downlinking telemetry, running most GNC algorithms, controlling the power systems, and supervising our payload. The payload is a Raspberry Pi SoC that is equipped with three cameras and an S-Band radio. It collects images, processes them using ML models specific to a given research objective, and downlinks the results and images over its high-bandwidth radio.
Official summary to be released soon...