Raw Notes From Meetings

2019-05-15 - IHOP Aliso Viejo

Present: smallsat don, johnathan, colin, harsh, jo3 and dan via video link

teledutainment.com web site courseware jo3 will take notes (scribe), colin will be backup in jo3's absence don: we have a brain trust like sri (stanford research institute): we can do anything we can imagine or find others to join motion on table: everybody take classes on line courses mit, purdue, princeton, u of alabama, texas tech colin: not just astrodynamics (kepler's law, etc.), but also spacecraft attitude mechanics (subset of problems you need to know and solving techniques). satellites are not rigid bodies, rapidly gets combinatorial with integrals mit la grange and hamiltonian astrodynaics - get everything in a momentum space, picking the right set of coordinates dyn - moments of inertia and ... ast - small and large forces in different scale one thing that unifies is control theory - achieve a desired result in a perfect world takes two schools to build a gnc application attitude dynamics the mission always changes - moving from leo to lunar is a mission - details must be solved every time gnc for a family of spacecraft fundamental product - roll out customer to customer the peter c hughes book "spacecraft attitude dynamics" two major problems with space travel - space is huge, and you are completely blind - it is complex space ops are very "blind" don: gnc - have knowledge of earth, moon, gravity, thousands of variables - complexity is in question algorithms - solve on the ground problem spacecraft does not know where it is. apollo - kalman filter (series of measurements to reduce noise - joint probability) - allowed it to navigate without communicating constantly https://en.wikipedia.org/wiki/Kalman_filter dan: queueing system in gpu - every so often ask: are we on track, is it "good enough" process for finding out what spacecraft thinks - over time, you have to pont where you think craft will be and you get like 20 minutes link margin on smallsat 5cm horn antenna 22kbps provide "easy com" to other missions nasa has a dsn on earth - includes mars - delayed tcp - we need a relay station at 450,000 km - maybe l1 or l3 and l4, also l2 (above the moon) - TDRAS do issue is the math to get into colin new astrodynamics using topology - stack nodes - composition problem to connect nodes from two separate bodies - direct transcription - stack nodes and try to solve - foliate the geodesics johnathan in topology, everythng is composable but you may not be able to preserve the same structure colin used to do kepler conics - large number of naturally stable gravitational structures - like elliptical orbital, topological data analysis https://en.wikipedia.org/wiki/Topological_data_analysis purdue - kathleen howell - direct transcription methods https://www.researchgate.net/profile/Kathleen_Howell don geodesic is the path an object takes when falling - all geodesics form a manifold - direct transcription is an optimization to make sure you get there and you can afford to get there colin gps on satellite in leo gps did not converge until 24min 37sec don create a geodesic which includes the thrust - fundamental problem is that spacecraft is dead reckoning dift bias and noise - iru is only good for 7 or 8 hours - a hemispherical gyro is too heavy wide band augmentation system - reflect it in ares where gps does not have good coverage - a bouy in space that transponds - may need like 7 of them we need 250 transponders but gps needs 3, or 4 for comparison what courses? colin's list - dan's book - probability and random statistical methods - papoulis third science - estimation theory, optimization theory first course digital control systems - global optimiztion - robust control systems dan bought: mehtematical methods - will send a list of all books purchased estimate on completion of 6dof software in one year after that, we will have a base line gnc runs a 137 ksloc (kilo source lnes of code) written in ada - number of man hours to write an loc is the same regardless of language - shuttle ada included - specs from NASA - 3-4 hours per loc through qualification test - 400,000 man hours for a gnc with a 6dof in there action item: find code bases to reuse - gmat, 6dof - take architecture as listed and estimate each part 25-50 kslocs - simplify fewer actuators, and other choices - or basebsall stand around doing nothing for hours then you fight study: dynamics and controls - videos course number will be in title on youtube octave is an open source matlab https://www.gnu.org/software/octave/ another level of specification: select algorithm definition - solve equations with numerical methods - at a hand-waving level - special place on the wiki "algorithm definition" discussion level dan: book "modern control engineering" GMAT general mission analysis tool NASA https://sourceforge.net/projects/gmat/files/GMAT/GMAT-R2016a/ http://gmat.sourceforge.net/docs/nightly/base/index.html https://en.wikipedia.org/wiki/Electrically_powered_spacecraft_propulsion http://www.doxygen.nl/index.html http://mail.softcafe.net/