Quadcopter Build Questions

[lapinozz]

Hi folks!

I'm intending to build a drone for various purpose, I always to build one.
For example I might mount a Logitech C920 that I have and try to do some video analysis to reconstruct the environment in 3D.
I'd like to try autonomous navigation in interior situation.
Since I'm probably going to do some of the stuff in "non-legit" ways I'd like it to be quite big and powerful so I can mount random thing like a grappling under it. I don't care much about speed or how maneuverable it is, I will focus more on stability and battery life time.

So my plan is to use a Raspberry Pi as a controller with a simple wifi adapter for communication and video processing (might offload some tasks to an external server) and an Arduino to control the physical devices.

Here's my current component list, I'm honestly quite lost.
Will they work well together?
Are the motors/batteries/propellers/power distribution suitable for my needs?
Will I be able to control the 4 in 1 ESC with my Arduino?
Is gyroscope + accelerometer enough or do I need another kind of sensor?

(Sorry I can't put the links because I don't meet the minimum requirements)
Super-H600 Frame
200g Vibration damping balls
Matek FCHUB-6S PDB w/Current Sensor 184A, BEC 5V & 10V, 3-6S Lipo
Turnigy Bolt 3450mAh 4S 15.2V 65~130C High Voltage Lipoly Pack w/XT60
Turnigy Reaktor 300W 20A 6S Balance Charger now with NiZN and LiHV
Quanum MT Series 4108 475KV Brushless Multirotor Motor Built by DYS
GWS Style Slowfly Propeller 12x3.8 Black CW
GWS Style Slowfly Propeller 12x3.8 Black CCW
Kingduino MPU6050 3-Axis Gyroscope 3-Axis Accelerometer

If you could help answer my questions or have any tips or warnings I'd be really grateful.
Thanks!

 

[wafflejock]

Regarding the DIY flight controller and programming I'd say drop that idea for the most part or rather look at using existing flight controller (FC) and open source software as the basis for your expansion/changes.

I'm interested in doing some of the same things so it will be good to see what others have to say here as well but will give you my 2 cents. Quick background this is coming from someone who has been flying for a few years, but mostly acrobatic quadcopters and a web developer/programmer professionally.

There are lots of good FCs available, I've used the omnibus F4 based ones (some quick searching seems to indicate these may not easily run iNav) on a couple of quads and have been happy with them overall (another alternative is PixHawk). The FCs typically run "BetaFlight" which is open source software https://github.com/betaflight/betaflight or can be loaded up with other software like iNav or I believe ArduPilot (though I'm not sure what quad support is like on this software, I've only used BetaFlight myself for acro human controlled flight).

Any FC and decent firmware/software will handle the regular balancing of the craft and filtering of noise which are the major problems to solve to get the thing flying stably. The iNav or ArduPilot are meant to support more automated control features like reading GPS data from a module and moving between waypoints etc. Any of these FCs is basically a STM32F4 (if an "F4" board) which is a MCU that uses ARM architecture, so you have a pretty common platform amongst all of them and most of them include at least as good of an IMU as the MPU6050 (just shop around a bit for "flight controllers").

The other things you mentioned might be difficult or impossible though I'm not one to stop someone from trying (or trying things I think might be impossible myself). That is the navigating inside (you'd need to do Simultaneous Localization And Mapping [SLAM], which is usually done with a lidar, check out rplidar but not sure how this would work on a vibrating platform), also towing things around means big, whereas navigating indoors means small so not sure you'll find the one size fits all solution.

Regarding your components https://www.ecalc.ch/ <-- is a good place to punch in some hypothetical numbers and pick some components that are similar to what you're expecting to get then get a good calculation of the in theory numbers for flight time and/or carrying capacity given some configuration of motors/propellors/batteries etc. I have an account I think so will try punching in what you listed so far and let you know if it screams at me about anything.

 

[wafflejock]

Don't think you included the ESC you're planning on in the parts list too would be good to add that here when you have a chance. From looking at the numbers I don't know that it will matter much what ESC you get really it will just effect the torque/punch-out or ability to change speeds quickly but really not sure that will really effect something that big much anyway and the amp numbers look low to me... let me know if you see anything I keyed in wrong to the input at the top.



My gut tells me a bigger battery can get you up to 15 minutes of flight time (6Ah, the battery... the calc confirms my guts suspicion)... Above 6Ah the weight of the battery and thrust to weight become a serious issue, even at 6Ah you'll be pushing the throttle pretty hard just to get it hovering but should still be able to sustain hover for a bit longer than 3.4-4Ah.


Not to drone on too long (see what I did there), but regarding my previous post about using an existing FC you still have the option of feeding control signals into the FC from something other than a conventional receiver which could be sending your custom controls in but I wouldn't exclude the FC entirely from doing the motor control unless you plan to spend a considerable amount of time solving the filtering/PID problems but they aren't anything too novel (IMO)... Okay I'm done editing I hope you enjoyed the book.

 

[lapinozz]

Thanks for the answers!
You're right I forgot to put the ESC (Turnigy MultiStar 30A BLHeli_32 4-in-1 Race Spec ESC w/ F4 FC, OSD & BEC (2-4S))

(Edit) Would a FC like this do the job?
MATEKSYS Betaflight F405-CTR Flight Controller w/ OSD, PDB, Blackbox & BEC Current Sensor

For the 3d reconstruction I've been looking at monocular RGB SLAM for some time, some of the online library seems quite promising!
I really wanted to make my own flight controller but maybe you're right and using a well tested one is wiser, I guess they are easy to interface with?

I think you but the wrong weight for the battery it's 380g, I don't know how much of a difference it makes ¯\_(ツ)_/¯
Speaking of which, the thrust ratio seems a bit low will getting better motor help with that? But I guess then the battery life diminish?
Can I go for more cells to fix the problem?

Thanks for taking your time for this!

 

[lapinozz]

If anyone has any suggestion on what motor I should use I'm all ears!

 

[wafflejock]

Thanks for the answers!
You're right I forgot to put the ESC (Turnigy MultiStar 30A BLHeli_32 4-in-1 Race Spec ESC w/ F4 FC, OSD & BEC (2-4S))

(Edit) Would a FC like this do the job?
MATEKSYS Betaflight F405-CTR Flight Controller w/ OSD, PDB, Blackbox & BEC Current Sensor

For the 3d reconstruction I've been looking at monocular RGB SLAM for some time, some of the online library seems quite promising!
I really wanted to make my own flight controller but maybe you're right and using a well tested one is wiser, I guess they are easy to interface with?

I think you but the wrong weight for the battery it's 380g, I don't know how much of a difference it makes ¯\_(ツ)_/¯
Speaking of which, the thrust ratio seems a bit low will getting better motor help with that? But I guess then the battery life diminish?
Can I go for more cells to fix the problem?

Thanks for taking your time for this!

Hi again, forgot to respond till I saw the alert again but just took a look at the FC and it looks good to me. ESCs sound reasonable too.

Regarding the thrust to weight I was just taking stab in the dark regarding the end weight of the craft so that would take more actual adding up of values or just weighing the real components to get an accurate figure but I was guessing roughly 2-3 times the weight of my 250mm quad, it might be more or less and there is some "payload capacity" it shows in the bottom right panel of the output in the calc that I believe just calculates the extra weight you could add before your lift would become 0.

The calc will account for some basics regarding the weight to make sure you don't give it a weight less than the combined weight of the drive motors but beyond that it's up to the user (me in this case) to key in the right values.

For controlling the FC I've not written my own code to do this but they all accept multiple protocols for input and I'm guessing there are libraries available for writing your data with the correct protocol to the UART you want to communicate with the FC over. Worst case scenario I'm positive you can use PWM signals from arduino or raspberry pi and most FCs will support PWM input (though it's considered one of the worse options of those available since one line per signal and I believe slower and more prone to interference than other protocols/signals).

Regarding RGB SLAM I've not actually heard of that at all yet so will look into it, the main concern to me with any sort of SLAM though is attempting to stabilize the sensor and isolate it from the vibrations from the motors/frame or otherwise attempt to filter out that "noise" from the sensor input without losing too much data/granularity. Will look into this a bit more though maybe less of an issue than with a LIDAR based setup.

 

[wafflejock]

Ah wonder if you mean this: http://wiki.ros.org/rgbdslam
RGB-D SLAM just skimming it over now but looks like it's based on using a kinect like depth sensor with the IR dot grid pattern for getting depth of the whole 2D view, yah interested in hearing more for sure though. What kind of sensor are you planning on?

 

[lapinozz]

I meant more something like this

 

[wafflejock]

Wow that is incredible! I'm vaguely familiar with some neural network techniques but have never seen it applied to SLAM thanks for the pointer will give me somewhere to look into it more, totally news to me.

---

Quick googling pops up this lib for the pi+sbus

https://github.com/1arthur1/PiSBUS, guess is you can use a pi-zero W (or Pi3) to capture whatever data to relay to a computer/ground station and to send the control signals to the copter, only issue there will be the wifi range.

 

[lapinozz]

Well even if I have a FC I'm still gonna put an Arduino+RaspberryPi on it so that seems perfect, I guess only thing missing before I order all the stuff is to figure out how to pack more lifting power!

 

[lapinozz]

I'm still looking into insight as to how to give it more lift!