Fully autonomous indoor quad with video transmission

Hi, I'm responsible for an initial design (and then probably implementation) of a fully autonomous indoor quad with video transmission. I'm a beginner in the field of RC with some background in control theory and image-processing, pretty solid programming skills and a soldering iron :). 

The functional requirements are:
  1. Stable indoor flight while piloted manually (to be used in the beginning phases of the project)
  2. Real-time flight data transmission
  3. Able to be piloted via PC (with joypad or so)
  4. Camera onboard, real-time video transmission
  5. Remote camera tilt control (only via PC)
  6. Obstacle avoidance
  7. Autonomous landing and take-off

...and that's it for the beginning. After a couple of days of feasibility study I've decided to go with: 
  1. ArduCopter Quad KIT (I already feel lucky to begin with the project once the kit is already available, or at least will hopefully soon again be)
  2. HMC5843 - Triple Axis Magnetometer
  3. XBee Pro 60mW 2,4GHz
  4. Camera - don't know which yet
  5. Infrared Proximity Sensor Long Range - Sharp GP2Y0A02YK0F for obstacle detection
  6. LiPO Battery
  7. Hobby King Battery Monitor 3S
  8. Ultrasonic Range Finder - Maxbotix LV-EZ4 for autonomous landing
  9. Futaba 7C transmitter

And then the questions:
  1. Can you spot any inconsistencies between the requirements and parts choice already? Am I missing something?
  2. I just need a simple black&white camera, high resolution not required, I'd like to real-time process the data in my PC and send the results back to quad for navigation, so it just needs to suit the airframe.
  3. How to transmit video to the ground? I've read Xbees 900MHz are used to do that, do I need a separate Tx/Rx?
  4. Assuming my Futaba 7C (2,4GHz) has frequency hopping, can I use 2,4GHz Xbee for flight data transmission?

Now, I know it's a lot. So please answer even if you just know one answer or have any comments on my choices. Especially questions 3 and 4 are important to me, as I know very little of radio transmission.

Thanks in advance and regards.

Views: 734

Reply to This

Replies to This Discussion

this is exactly the Parrot ArDrone you want to do!!!
Parrot indeed sounds a little similar. I need the drone as a testbed for diverse control strategies, mainly vision based, though, so I'd like it open source. So I'd like to be able to change things as I need them over time.
I'm trying to do almost the same thing as you... without the cam side. Not working hard those time, but.

I manage to compile the code to control the ARDrone from my laptop yesterday, and, I'm working on a seccond version of my GauiQuad that I want it indoor able to avoid object with sonar sensor.

http://sites.google.com/site/gauiquad/

Code baseline is from the AeroQuad, but I have modify it a lot. Still far to get something stable, and it's the first step! But, I want to play a bit with the drone this time!
Nice quad, nice flying. I wouldn't like to get too off topic here, though, as I'm really interested in answers to my questions. Especially questions 3 and 4 are very important, as I know very little about radio transmission.
3) you will need actual video tx and rx, xbee's are for data - not video.
4) in theory they should be ok together, given they will be inside and range is los at those frequencies. you will need to keep the two antenna's as far apart as you can and hope for the best.
I am also working on indoor navigation with my drone. Take a look here:
http://www.diydrones.com/profiles/blogs/indoor-uav-3d-position

A piece of advice:
I suggest to use ultra sonic range finder also for "obstacle avoidance". Their beam are wider, thus detecting obstacles is more reliable.

For the radio link: you should consider using just 1 radio link, because you plan to control it via a PC. That's what I am doing on my drone, with just 1 bluetooth link for all usages: sending high level commands, manual control, telemetry, etc...
Thanks Leon. I initially planned to use IR for obstacle avoidance suggested by Jose Julio's airframe: http://diydrones.com/profiles/blogs/arduimu-quadcopter-part-iii . I've seen, this does not work flawlessly on all angles relative to obstacles, though.

Does an ultra sonic sensor like you used solve this problem? What other problems does this introduce? Theoretically at 30 degrees beam-width one should use 12 sensors to fully cover surrounding space. This, I think, would lead to false readings as sensors would interfere. Don't you have the same problem while using 4 sensors?

About flight-data and video transmission: I've heard from another source, using Xbee 2,4GHz together with Futaba 2,4GHz, even with frequency hopping, musn't work well (as Futaba 7C is also based on Xbee), so I won't use this. It is said, such solution could introduce control lag, for instance. Though I'd like to control my quad via PC I'd like to preserve manual control possibility, so 2,4GHz is for Futaba. Bluetooth is also 2,4GHz, so I'd rather use Xbee 900MHz - any other suggestions?

I don't yet know what about video transmission. Do you have any suggestions on that, preferrably a transmitter / receiver set I could then use, e.g with EasyCap, to view and analyse image on my PC?
Thanks Paul, I've asked some more questions below.
I tried several sensors on another project: a wheeled autonomous robot
http://ze.bot.free.fr/index_en.html
Both sonars and IR sensors have troubles when obstacle is not well oriented.
IR sensors are less reliable: they see more artefacts, and don't detect all obstacles.

I only use 4 sonars. My aim is not to detect all obstacles, but only to detect walls and floor. Il just want to locate my drone in 3D. Yes, these sensor can interfer with each other. I only measure with 1 sensor at a time, and I set a 60ms delay between 2 measurement.

In my drone, I use a single bluetooth radio link even for controling it via manual control. Take a look at the first pictures of my drone below. I changed all the electronics of my radio-control, and I put a bluetooth module inside.
http://heli.bot.free.fr/

You can use a precise (+-2cm) Indoor "GPS": http://marvelmind.com/.

Here a couple of demos:

Reply to Discussion

RSS

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service