|
|
Thread OP
|
Question
Is there a reliable Obstacle Avoidance system for DIY multi rotor systems?
Anyone? Products like the Phantom 4 seem like they will add a lot more use case scenarios for multi rotors but I cannot see myself investing in a multi rotor I cannot customize or upgraded, is likely unrepairable and uses proprietary batteries.
|
|
|
|
|
|
|
Be patient, OA is still at infant stages for Mr's. Be patient it is on its way. personal opinion is that over the years I have never had the need for it and do not see the need for it now. Guidelines say fly LOS, flying LOS should never put you on a collision course with anything unless operator error is involved like a to low RTH height
|
|
|
|
||
Thread OP
|
Quote:
|
|
|
||
|
|
|
Yes, the Matrice has a full 360 OA system available, and it's completely customizable, including a vision processing platform called the Manifold. Check it out. Ain't cheap.
http://www.dji.com/mobile/product/guidance https://www.dji.com/mobile/newsroom/...orm-developers |
|
|
|
|
|
Full scale OAs are still in the lab. For now, I don't think the market is ready nor welling to pay for the cost. The realsense (YUNeec/Intel) has the distance issue so does the P4 (optical) system. It has resolution issue. The Neato LIDAR can overcome such shortage and can look 360. The cost of LIDAR is coming down but still cost more than realsense or kinectic sensor. Also, I can solve the payload and computation demand issue. In addition, the chip makers (intel/qualcomm/Nvidia/etc) have their own agenda. On paper, qualcomm dragonfly is ideal for the application. it has the gps, gyro and image processing features. I have been waiting for them to add SFM or optical flow algorithm to it. It has been 2 years and wait countinue. Ganted, both algorithm is computation dmanding and may overtax the chip.Meanwhile Intel Quadchip with realsense is released last year. but they have to use hex to solve the payload issue.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
|
Quote:
OA goes back many years on DIYDrones. The current offerings IMO are novelties. However, a practical affordable solution may be in the works. http://diydrones.com/profiles/blogs/...age=2#comments http://scanse.io/#firstPage |
|
|
||
|
|
|
OA is not a marketing gimmick. I believe it is a critical component for unmanned missions. you can't expect no new construction or tree growth on your route. When OA is mature enough to detect thin branch and high voltage line, people will demand OA for their drones.
Right now, the MIT OA seems to hold the promise, but there are still too many moving pieces. please keep watching out for OA |
|
|
|
||
|
Quote:
|
|
|
||
|
||
|
Quote:
Sheesh. http://spectrum.ieee.org/automaton/r...ots-and-drones ![]() It will be ready for Pixhawk (Arducopter) flight controllers and beat the socks off anything currently available at the consumer level. MIT? For ~$250? Really? |
|
|
Last edited by slimething; Jun 01, 2016 at 10:43 PM.
|
|
|
|
|
Yeah, that's kind of like the marketing gimmick in WWI when they came up with synchronized machineguns that could shoot between prop blades! They didn't complain about shooting above the blade or hitting bullet deflectors on the props. But once they came out with it it sure made for a way more accurate gun and all the bullets counted!
|
|
|