I'm doing a group school project and our team needs to buy a drone kit that's capable of adding on a full autopilot system along with a few extra sensors. I'm not very familiar with what's out there to buy and we don't want to commit to a kit then later on down the road learn that we are very limited with how we can modify it.
Any ideas on what type on controllers or full kits to buy? Or at least a good place to start?
Any help would be great!
Check kit-drone.com ;)
Yes, GPS is a problem, Pixhawk (ArduPilot) uses the GPS for autonomous navigation.
Generally indoors = no autonomous operation.
There are other sensors you can use, the Pixhawk supports the PX4 optical flow camera, but it is really only programmed currently for position maintenance, not navigation and generallu the drift from using it for navigation would be cumulative and thus suboptimal for navigation anyway, so not really suitable for autonomous, non-GPS navigation.
Interestingly the very first quadcopter that actually can do a reasonable job of non-GPS navigation has just come out, the DJI Mavic supports quite sophisticated camera flow and stereoscopic 3D camera vision sensing which can allow quite a bit of autonomous activity even with an occluded GPS.
Don't know if that helps you much, but duplicating or exceeding the Mavic's level of capability will be a very significant project if you elect to undertake it.
The 3DR Solo has some optical flow capability, but it is not programmed to be autonomous without GPS and the Solo companion computer is closed source, so you would not reasonably be able to write code to do this even if you were otherwise capable of doing so.
Lidar is just beginning to be a reasonable sensor consideration and I would encourage you to be looking into Lightware's offerings of both laser rangefinders and 2D scanning LIdar.
In particular they are coming out with a new very small but highly capable laser rangefinder with in the next few months that should also be favorably priced for this use,
Although this will be a great sensor for this use, make no mistake, to implement autonomous capability using this is a very - very significant software development project and I would suggest you look into something which uses ROS like the Erle Brain II flight controller, a Pixhawk alike, but with extra work put into producing a ROS compatibility stack.
I strongly suggest you contact these guys as they have been very proactive in helping the DIY community.
If this is where you are heading, be aware, this is a really big project.
The other method of environment relative navigation is generally 3D stereoscopic camera based.
It is very computationally intensive.
Thanks a lot for all the information!
Gary McCray, Yes we have been looking into all the regulations surrounding autonomous drones but fortunately for our project we are doing a small scale prototype that will only operate in an area the size of a room, most likely indoors for now. Pixhawk is actually the one I've been looking into! (mainly because it keeps coming up on my google searches). I'm assuming most of the autonomous flight feedback comes from the GPS, accelerometers, gyroscopes and barometers? If we're operating indoors I'm assuming the GPS will not work properly. What other sensor types do people use for autonomous flight? Maybe radar and infrared? Lidar?
Ray Anderson, actually the building of the drone is more of the preliminary step before we can actually start the project. It's more about what we're doing with the drone and the integration of other sensors and equipment that follows. So honestly, the least amount of work we have to do to get the drone up and flying and sourcing of parts the better as this is a time sensitive project (well we still have 6 months). I really appreciate all the help with your reply! I'll definitely be contacting you via email and picking your brain. Thank you so much for the offer to help!
You are in the right place to get some opinions, for sure. I will make the assumption that since this is a school project, you are expected to build this from the ground up in order to understand the complete system rather than buying a ready to fly system and then modify it. If that assumption is correct, here is my opinion: First, purchase an ecalc account to run your power option calculations (http://www.ecalc.ch/xcoptercalc.php) Then, if you plan to fly in an aquatic environment, use a frame like this (http://www.aquacopters.com/store/p2/Project_X-Frame_Only.html) or for more normal operation (http://www.getfpv.com/tarot-650-sport-quadcopter-with-retractable-l...) I will also assume that you are only interested in electric propulsion, so the motors and speed controllers will be your next most important decision. Keep in mind that this entire equation changes if you desire long duration flight times, which may cause you to consider nitro fuel powered propulsion with collective pitch flight control (http://www.curtisyoungblood.com/platforms/tdp-mr/). For now, I will stick with the assumption of full electric propulsion and shorter (and cheaper) flight duration. The next step is to use ecalc to help you choose motors and speed controllers and propellers. You will see that the frame manufacturer makes recommendations for this hardware, but I always run their recommendations through ecalc to decide what I may want to do to improve performance (use their recommendation as a baseline). All along, I have been assuming that you are interested in using Pixhawk as your flight controller since you are on this site and it is open source and fully expandable, as you requested. Buy the full startup kit (http://www.ebay.com/itm/like/152015100513?lpid=82&chn=ps&ul...) which includes GPS, PPM Encoder and Voltage Sensor. At this point you should decide on a radio, and I suggest the Taranis as what appears to be the most popular due to it's capability and price (http://www.getfpv.com/frsky-taranis-x9d-plus-2-4ghz-accst-radio-mod...) There are a few model options to choose between, but this is a great base radio. A few items that may not be obvious to purchase are a Power Distribution Board, 3S or 4S Lipo batteries and charger (3000 Mah minimum - pay attention to the weight vs. C rating but I go as high as I can afford in C rating), UBEC, Battery connectors, Sticky back velcro tape, Double-sided foam tape, Velcro battery straps, Indestructible propellers, 60/40 solder, practice soldering skills with a good iron, lost model alarm... From here, you have additional options to add to the airframe (this is listed in my personal priority order, yours may differ) 1. Camera Gimbal 2. Optical Flow Sensor (to assist the GPS in position hold while recording) 3. Telemetry Radio (to transmit real time data and enable the use of a Ground Station such as Mission Planner or QGround Control) I can be reached at firstname.lastname@example.org if you need clarification or would like to talk. I am working for a University on a drone design, so this discussion is very relevant to what I am doing. Please keep in mind, these are my opinions and recommendations based on personal experience. At least it gives you a place to start.
The Pixhawk actually offers several options which permit full autonomous flight.
However, you should be aware that government regulations (in the US and Canada) require a pilot in command who maintains continuous line of sight visual contact with any aircraft and who can retake "effective" manual control at any time.
In Canada there are also several other restrictions and you need to look at those first.
Although the Pixhawk is an open source project, undertaking inserting your own code while possible is not for the faint hearted or inexperienced.
It is quite possible that the existing built in modes will already support your desires, so you may likely not need to program it directly, in which case the supplied firmware and Mission Planner software will provide everything you need.
Still, it is probably the flight controller most in line with your goals and as I said, with the standard code there are several modes that can support complete takeoff to landing full autonomous flight.
But you still definitely do need to have a pilot in command who can instantly retake control and perform a manual return and landing (and who keeps the plane in continual visual contact).
Hope this helps,