Replies to This Discussion

Permalink Reply by Bot Thoughts on January 13, 2012 at 10:44pm

Wished I could advise. I've never built a flying autonomous vehicle. For that matter, I've yet to build a functioning autonomous ground vehicle. I do agree that an off-the-shelf IMU/AHRS solution is the best way to go vs. trying to roll your own with limited experience. You can always work up to that point, but sometimes it is easier to get there if you first experiment with someone else's solution.

• ▶ Reply

Permalink Reply by Just Robotics on January 14, 2012 at 6:20pm

Bot Thoughts,

 

Thanx a lot

Your "thought" right on the "spot" ;-)

I am from Computer Science background, so I wanted to work on those areas first where I can apply my acquired skills. I shall definitely learn control theory and advanced dynamics and all related sciences, but at a slightly later stage.

 

Let me give it a try, will keep you guys updated

• ▶ Reply

Permalink Reply by Tom on February 5, 2012 at 9:24am

I think, developing a UAV simulation in, e.g., Matlab / Simulink, from scratch might be a better place to start, if you really wan't to understand how things work. I see the learning process the following way:

1. First thing to understand when beginning work on developing an autopilot is the particular UAV's kinematics and dynamics. You might want to create a visualization for the tool, then, to ease test your mathematical model easier.
2. Then understand how on-board sensors work, i.e. simulate their readings adding some random noise.
3. Try to implement an AHRS using simulated on-board sensors' readings.
4. Follow with PID attitude control system's design.
5. You'll know what can be done next when you reach this point.

Regards,

Tom

• ▶ Reply

Permalink Reply by Kanishka on February 13, 2012 at 8:15pm

Hi.. 

I was following this discussion and this site in general with great interest. 

After reading some documents in the internet the conclusion I came about the general process of attitude estimation as follows. 

Please correct me if I am wrong here.

1. In first place acceleration and angular rates are taken in to the system

2. Because these sensor inputs are so noisy the inputs are added together to get more accurate measurement.

        For this approach fusion mechanism is needed and here is the place the complementary filter/ Kalman filter comes in to the role.

3. Attitude measurements found from above is related to the aircraft body itself.

         ie: we cant still see the attitude with respect to the earth ( If we rotate the aircraft in a axis 20 degrees irrespective to the initial possition the attitude we get is 20 )

4. Then this attitude estimation can be used to find the attitude of the aircraft with related to the earth.

        To this only the Eular method, DCM or quartarnian methods been used.

If there is something wrong in my understanding can anyone please point this out to me.

Thanks a lot in advance.

• ▶ Reply

Permalink Reply by Tom on February 14, 2012 at 2:04am

Ad 2. Yes and no. Yes, the sensor measurements obtained from cheap gyroscopes and accelerometers are quite noisy. Generally, in case of very slow rotating and non-accelerating objects, 3-axis accelerometers alone would suffice to obtain angular orientation - basing on gravity measurement, which always acts towards Earth's center. But in case of a UAV which is not slow rotating or non-accelerating also linear and centrifugal accelerations would be measured and added to gravity measurement which occludes raw gravity vector. In the long run, though, with some low-pass filtering, this still provides a noisy, but non drifting rotational orientation information. As in case of an autopilot we need precision and high frequency attitude estimate, this is where gyroscopes come into play. Because gyroscope measurements include a time-varying bias, this bias cannot be easily eliminated (e.g. via subtraction). As gyroscopes provide angular rate measurements, this signal needs to be integrated to obtain angular orientation. This is how gyroscope-drift arises. So information obtained from those two sensors, if fused, gives us an attitude estimate without drift (thanks to accelerometers) and fast enough (thanks to gyroscopes). The role of a complementary filter (called DCM by some) is to provide means to estimate gyroscope drift from accelerometer measurements and eliminate it (to put it simple)

Ad 3. Yes -  a strap-down gyroscope provides angular rate measurements expressed in the body frame. Yes and no - a strap-down accelerometer measures inertial frame accelerations expressed in the body frame.

I very much recommend the lecture of Direction Cosine Matrix IMU: Theory by William Premerlani and Paul Bizard, if you haven't read it yet. It's written in an understable fashion without too much mathematical language, so it's intended for a reader like most of us here.

• ▶ Reply

Permalink Reply by Kanishka on February 14, 2012 at 8:11pm

Hi Tom,

I really do appreciate your reply. Thanks a lot for that. 

Your comments confirmed what i thought before. actually my language skills are not good enough to ask exactly what i want. :)

I read that draft document and i following this site with great interest. 

thanks agian

• ▶ Reply

Permalink Reply by al on April 23, 2012 at 1:57am

hi TOM,

where does one find such files like the pdf you linked?

I am very interested in a deeper understanding of the functioning of the ardupilot mega, maybe a code explanation/manual....

thank you

• ▶ Reply

• ‹ Previous
• 1
• 2
• Next ›
• Page