New Sensors!! :-D Colision Avoidance

WooHoo so I picked up my new ultrasonic sensors from my old workplace today!  that means it's time for me to start work on collision avoidance!! :-D  something I desperately need hahahaha...

first stage for me is to get the arduino uno reading the sensors, I might then set it up to talk to the ardupilot board on i2c and integrate the system from there...

Please forgive me for starting so many projects at once, but I want to make the most of my unemployment while I can!! great time to innovate and get all the ground work done before it gets slowed down by real work and earning a living ;-)

will keep everyone posted :-)

Views: 2483

Comment by marc.kulhavy on October 4, 2012 at 4:44am
This is great! A thing a friend of mine and me are thinking about all the time. We are using the apm to create high resolution pictures of areas. For better resolution, we have to choose lower flight levels and sometimes trees are higher than the planned route. If this works, you have your first customer...
Comment by sebiss07 on October 4, 2012 at 10:05am

Very Good !

Comment by John Githens on October 4, 2012 at 11:16am

@Jesse:  I am very, very interested in your progress, and how the experience can be applied to a quadrotor. Keep on sharing (and avoiding collisions)!

Comment by Jesse on October 4, 2012 at 4:03pm

@Marc: I'm not sure how it'd go avoiding leafy trees, I will do some tests as I write the code an let you know... but I figure some level of collision avoidance would be better than none at all!  @John: I'm planning on sending as many sensors as is connected to the APM in a packet, be it 8, 6, or 4, my original theory was one sensor per arm but I was thinking about it last night, and I'd consider using 8 sensors on a quad 2 per arm at about 100 to 110 degrees separation so that there's some level of overlap and a bit of angel coverage... 

to start with I'm going to go with a "go the other way" style of avoidance and publish that, from there we can build a more comprehensive set of decisions and actions that can be followed...

Comment by John Githens on October 4, 2012 at 4:21pm

@Jesse: Sounds like a plan. 'Angel coverage' is good. LOL. In my ruminations, it seems like the typical quadrotor, if not flown expressly for aerobatic maneuvers, would most frequently need to avoid obstacles in horizontal planes from about 45 degrees below - and above - the horizontal centerline of the four props. Anyone flying in enclosed spaces would need more sensor coverage to avoid obstacles in the upper 90 degree zone (for example, to avoid a physical ceiling), or lower 90 degree zone (for example, to avoid a large rock or a person's head...). Just some thoughts. I like the "go the other way" style for now. That might be sufficient in many cases.

Comment by John Githens on October 4, 2012 at 4:32pm

@Jesse: Tossing in another thought. At Home Depot I picked up a Ryobi Sonic Tape sonar device for $12 USD. Does a remarkable job of measuring wall-to-wall distances (but not for more demanding contexts). There might be reason to explore adapting those to UAVs. And hey, it comes with a red LED laser to make sure you are pointing to the intended location... suggests a surreal scene at night...

Comment by Jesse on October 4, 2012 at 5:19pm

@John: these sensors are basically the same tech; I got all 10 of them off ebay for something like $20, between that and the arduino uno it's not too expensive... hacking a sonic tape measure would be significantly more expensive, especially when you'd consider using 6 to 8 of them on a quad/hexa... size and weight would be a consideration as well...

and to be honest I hadn't actually considered vertical CAV yet, although with my implementation plans it wouldn't be a big step... 

Comment by marc.kulhavy on October 4, 2012 at 11:55pm
Jesse,

my thoughts were the following. We have to fly low altitude missions with a copter to make aerial orthophotos.
The flightpath is several parallel flights with an offset of some dozen meters in between. The copter always flies facing in the direction of the flight. It would need two sensors to work properly, one facing down, the other facing in the direction of flight. The logic would be like this:
- measure distance to obstacle in front of flightpath
- if distance to obstacle is less than a previously set value, lets say 5 meters, stop forward movement and memorise this gps point
- begin to move upwards until the obstacle is out of the sensor limit, memorise this point
- move upwards from that point for the set value, so another 5 meters, keep this point in memory
- begin to move forward again and watch the bottom sensor, he will begin to see the obstacle
- once the bottom sensor does not see the obstacle any more, memorise this point
- move another preset distance, in this case 5 meters forward and stop
- begin descent to previous flight altitude
- continue flightplan

There is a little more logic needed, but i hope you understand what I mean. There has to be another refinement done if behind the obstacle is another and lowere one which would prevent going back to planned flight altitude, but I think this could basically work with only two sensors.
I am not really good in programming, i did some projects with the atmega, but if there is a way I can help you with testing and developing, just drop me a line. Cav will become more and more important over the time, so this is a pretty good project to get involved.
Regards
Marc

Developer
Comment by Jani Hirvinen on October 5, 2012 at 2:09am

Jesse how many you plan to have on your first sensor array, as we could look to connect them to jD-IOBoard and that way free some computing power from main controller. Later on we can look to create full dedicated PCB with MCU on it for this use. 

One thing that I am really concerned is to have those sonars pointing horizontally as propellers etc are making massive amount of noise. Long time a go on original avoidance system written by Jose we used to use Infrared sensors but as we know, those are not sensitive enough and their beam is really narrow. Good thing on IR is that they are not affected by noise coming from motors/props.

Comment by Aaron Curtis on October 5, 2012 at 6:16am

FYI, I've been working on an obstacle avoidance system for a while now for my upcoming work inside caves. I have 4 of the maxbotix long range sensors and have set them up in chained mode to avoid interference between sensors. I also have 4 infrared sensors and am hoping to work out a system involving all 8.

Comment

You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service