Видео

This isn't just a quick fix. I'd have to see screenshots of what you're doing and see if you're typing in the commands properly. If you'd like to email me with some screenshots and photos of your board I can try and help

If you have a data set that's a function of time I would recommend using a runge-kutta or Euler integration scheme. I have a few videos on that if you do a search.

matplotlib, it's a python library

@@Guglhupfscout24 it's called the psuedo inverse.

Well torque is equal to inertia times angular acceleration. So I made an assumption that current was proportional to torque since that's energy put into the system and this proportional to the angular acceleration.

Alright, thanks for your response. Also, the attitude control demonstrated in this seminar series is only for stationary desired attitude. If we want to perform non-stationary attitude control such as nadir pointing, do we have to use a different control law/equation.

I'm pretty sure the equations are the same but you'd just have to change the commanded angle.

So compute desired quaternion everytime and convert it to Euler Angle as ptpcommand in the control block? By using angles in the control block, do we face risk of singularity issues.

Well yes. Converting from quaternions to Euler angles introduces the risk of singularity. Better to do a control scheme that creates an error signal with quaternions

Have you read my aerospace mechanics book? github.com/cmontalvo251/LaTeX/blob/master/Aerospace_Mechanics/aerospace_mechanics.pdf Chapter 7.5 goes through the math behind getting the position of Earth w.r.t the Sun. I also have a few videos on orbital elements already that help in that area.

@@CarlosMontalvo251 Hi, thanks I implemented it successfully. Also, the density model which you have used is fairly simple considering the fact that you've used LEO sats. Do you know any model / paper i can refer to for densities at medium and high earth orbit altitudes?

Unfortunately I'm not aware of any papers about density off the top of my head. I'm sure Google scholar will turn something up. Alternatively you could get a textbook on astrodynamics

I probably can. Can you be more specific about rocket stability? Are you talking about modeling the stability margin and the pitch of the rocket?

@@CarlosMontalvo251 Yes, I am trying to confirm stability margin of a model rocket. Its not too hard to find the stability margin in ideal conditions. If there is wind it can change the angle of attack, which would mean a new normal force and COP/stability margin. I haven't found good sources on how to calculate this new COP since most suggest basic string tests or complicated CFD models. A walk through of how perturbations can affect AOA and stability would be awesome. Modeling the rocket with disturbances would be cool but I suppose calculating the stability margin and damping coefficient is enough to ensure stable flight.

@@kinghenry238 Most of the models I use assume a center of pressure that is constant with angle of attack. Modeling the change in center of pressure is pretty difficult and I've only done with wind tunnel tests or CFD tests. You can also do live fire tests but again those are tough. For small angle of attack the center of pressure is supposed to be constant anyway. Things like Mach number and angle of attack can affect it but I don't have any theoretical equations or data on that only empirical data.

@@CarlosMontalvo251 Ok, that's good to know. I might have overthought the disturbances then. I'll try CFD out, but that is a beast in and of itself. Thank you for your response and video series, they have been helpful!

Well let's see. The x-axis goes through the front face which means you need to use the Ly and Lz side, the y-axis goes through the size face which means you need to use the Lx and Lz side and then finally the z axis goes through the top and bottom faces so that needs to be Lx and Ly so Is = (ms/12) * [Ly^2 + Lz^2 ; Lx^2 + Lz^2 ; Lx^2 + Ly^2] which means you are right! The code has been updated. Thanks for your addition.

@@CarlosMontalvo251 Thanks for considering, much appreciated!

It depends on your coordinate reference frame. Typically the ECI (Earth Centered Inertial) frame is with the z-axis going out of the north pole. I think you're referrring to the NED (North East Down) reference frame where the z-axis always points towards the Earth's center. I talk about a lot of these reference frames and transformations in my book. github.com/cmontalvo251/LaTeX/blob/master/Aerospace_Mechanics/aerospace_mechanics.pdf

@@CarlosMontalvo251 Thank you for your reply, sorry for the misunderstanding. So what I meant is that when you compared the NED frame with the ECI, you have added a negative sign to the NED z-axis value. However, when I went through your code on GitHub, I didn't find that same negative sign that was added in the video there. I would like to know that now when I use the NED should I add a negative sign to the z-axis or keep it as it is? PS: Thank you for writing this book, I have learnt a lot since I started reading it. Unfortunately I did not have the chance to complete it, but I will complete it once I get a chance!

@@3RDC4RD Are you saying there's an inconsistency with the video posted here and the code posted online?

@CarlosMontalvo251 yes, in the video you added a negative sign to the z-axis (NED), and in the Github code that negative sign is not present

@@3RDC4RD Hey thanks. Not sure how I missed that. I even readded the comment in there too.

Did you lower the timestep? Sometimes the simulation won't converge if the timestep is too large

@@CarlosMontalvo251 But ode45 uses adaptive timestep right ? Do I have the control over timestep ?

Actually the error said that the timestep got so reduced, it went below permissible values and caused the algorithm to stop.

@@MrEngineer_ This simulation uses RK4 so you have to set the timestep in the main.m

My apologies. The latest version of the software uses ode45. The simulation with ode45 probably threw an error when you went too close to the center of the planet and the gravity model became too nonlinear.

You can actually go either way. You can use the form. x(i+1) = x(i) * ( 1- s ) + s * y(i) and have s go from 0 to 1 or x(i+1) = x(i) * s + ( 1 - s ) * y(i) where s goes from 1 to 0 with the same effect. Apologies for the confusion but as long as you bound s from 0 to 1 and understand that the edge cases swap when the equations change you can still use the filter in either case.

@@CarlosMontalvo251 Alright understood, Thanks for your response.

github.com/cmontalvo251/aerospace/tree/main/adcs_seminar_series

@@CarlosMontalvo251 Thank you! Do you have any code that involves spacecraft detumblization?

@@joshuathompson6275 The code above detumbles with magnetorquers and reaction wheels.

Unfortunately I dont. This student just did this experiment has an independent experiment.

I just checked the current source code and I think it is up to date but double check and if it's still wrong send me the line number you're referring to. github.com/cmontalvo251/aerospace/tree/main/adcs_seminar_series

@@CarlosMontalvo251 Yes, you are right. It is up-to-date. Thanks for your response.

The model outputs the x and y coordinates in the J2000 frame including the position of Mars. So if you want to do a reference frame transfer you would just need to substract the J2000 frame positions from the Mars X/Y coordinates.

Unfortunately I don't know. The JPL website maybe? ssd.jpl.nasa.gov/orbits.html

Sounds like you just need to use "sudo"

@@CarlosMontalvo251 It was folder permissions actually but thanks!

www.mathworks.com/matlabcentral/fileexchange/34388-international-geomagnetic-reference-field-igrf-model

@@CarlosMontalvo251 🔥🔥

Yes as long as you input the correct Julian Day and time

@@CarlosMontalvo251 thanks..I got them

Can you tell me how to calculate lunar nodes position in sky

@@narayanpatil4085 The moon has it's own ephemeris data. I would recommend you look that up on the JPL website ssd.jpl.nasa.gov/orbits.html

That's the distance between discretizations in the x-axis

homemade rocket. to see the curvature of the earth you'd need to be at the same level as a commercial airline. Same reason why you cant see the curve while on a mountain.

Unfortunately we moved away from the arduino and to the Raspberry Pi

seems to be over compensation from PID loop, try tuning it

@@headSoup We don't have this drone anymore. This is from 2018.

LOL

My student made these so I'd have to ask him for the plans if he's willing to share. I'll get back to you soon.

Rk4 is a fixed step integrator so the timestep is constant. What do you mean by fractional?

@@CarlosMontalvo251 usually i see the runge kutta written as evaluating k2, for example, at t0+dt/2 and at y+k1*dt/2. maybe it is because k2 is already an approximation for that fractional time step?

I'm still not following. RK4 is called RK4 because it's a 4th order taylor series approximation. The 4 function calls are used to create k1,k2,k3, and k4 with "fractional" timesteps. These are on lines 65-68. If you skip to 3:10 into the video you'll see.

I feel like the G's pulled during hovering to forward flight would make everyone puke. Definitely would be fit for the military but even then when you scale things up everything gets harder. Just look at the V-22 program.

But as with anything. I'd say try it and see what happens.

@@CarlosMontalvo251 haha you're probably right about g forces, it would need a more gradual transition. I am personally aware of two attempts to make a civilian tailsitter. Nasa Puffin and Zeva Aero. There is also blackfly, but it's not really a tailsitter, more of a hybrid. In my opinion, the power of a tail sitter is that it's a conventional plane essentially, it can have a normal sized wing suitable for gliding. But i haven't seen this approach used by anyone so far. There are a few concepts though, like Bell apt.

@@CarlosMontalvo251 oh, almost forgot the jetman from dubai, he had a winged jetpack. Took off vertically from his feet, and then transitioned. Would be really cool if he had a longer wing and electric motors.

You actually don't need any extra add ons. Just the standard MATLAB version will work.

Thank you for the quick reply and helpful content Sir. Does this apply to the other ADCS in MatLab tutorials as well?

​@@de_michael1222 There is only 1 ADCS software on Github. There are multiple videos because every major edit that I've made has been done live on youtube. So if you downloaded the latest iteration of the software you wouldn't need any other packages.

@@CarlosMontalvo251 alright, thank you very much!

I assume its because of adding the disturbing forces and moments to the simulation? is It possible to fully stablize the satellite even having all the disturbing forces?

Great observation. That is a timestep issue. In the video I changed the timestep = 1.0 so the code would run faster. In order for the RK4 integrator to converge properly for attitude control you really need the timestep = 0.1 or lower. That will increase simulation time but increase accuracy and make the curves smoother as you say. Hope that helps!

@@CarlosMontalvo251 Thank you for the response Sir. I tried extremely low values but the spacecraft is still not completely detumbling.

@@SA-ys8egDo you still have sensor errors and biases turned on? Because it will never go to zero if you have errors turned on.

@@CarlosMontalvo251Sir, I will send an email to you showing plots with the same parameters but different codes. The previous code is able to detumble the satellite completely while the new code (after adding the total current calculation for reaction wheels) has chattering even if the simulation is run for 20 orbits with timestep of 0.01. And yes I turned of sensor errors biases. I also removed the affect of disturbance forces which I suspect were causing the satellite to keep tumbling but that was not case. Thank you for the response.

The best place is probably here github.com/cmontalvo251/LaTeX/blob/master/Aerospace_Mechanics/aerospace_mechanics.pdf