Slashdot Mirror
Ask Slashdot: Is a Home Drone Feasible?
dargaud writes "I live in an alpine setting and I'd like to be able to remotely view various remote valleys to check for ice formations for winter climbing. I wonder if there are cheap drones that could do that. Requirements would be: GPS guided on a preset route (no remote control necessary, and anyway there's no line of sight), at least 20km autonomy, 1 or 2 cameras on the sides to record valley walls, easy launching and autonomous landing (parachute?) at predefined point, ground detection to avoid crashes (if preset route is wrong or GPS echoes on valley walls as is often the case). Is there anything commercially available cheap enough, or any DIY that doesn't require a year of assembly?"
Totally depends on your skill levels in meshing the brains together and fabrication. There is not enough information in your post to determine either, but based on the tone I am getting your best bet is money, and for the features you want with the durability just to survive the terrain while carrying a small load, it aint going to be cheap.
Off the top of my head maybe model aircraft with telepresence would be the best mix.
Have you heard of this? http://www.sparkfun.com/products/8785
http://code.google.com/p/arducopter/wiki/ArduCopter
or for other options
http://diydrones.com
http://www.faa.gov/about/initiatives/uas/uas_faq/index.cfm?print=go#Qn2
You can only fly them as if they were R/C aircraft, which means line-of-sight only and you must have a link to the ground. There are also ceilings and rules about keeping them away from buildings, people, and especially aircraft.
The quadrotors that you see people putting cameras on are not UAVs, they are just remote controlled and someone on the ground is flying them in real-time. The FAA is moving very slowly on approving any sort of UAV flights (public or civil) although they are being forced by Congress to finally issue rules about how they might go about approving civil UAVs. Otherwise, right now UAVs can only be flown by the government, government contractors, universities, or in military airspace.
There are no commercial options that a private citizen can buy, and the DIY options will require lots of work. This is an area of active research in the robotics community, and implementing any one of the features you mentioned would probably be sufficient to get you a Ph.D. right now.
He's not looking for a solution just yet. He wants a starting point. You making things too complex too fast. The question is, can you do those things for cheap? Is it possible? If you have a question, the answer is "Ideal conditions"
What are the typical and maximum wind speeds in the valleys you are looking at? - No wind
How high above sea level are you, and what is the highest point you want the drone to get to? Sea level to 10 feet
Are there constraints on noise (ie will a loud engine cause avalanches? - Doesn't matter
Now build a simple solution. Lets see, GPS, cameras, autonomy, collision detection, 20km range. The cheapest is about $9,000. But it might not meet your needs.
Now the poster sees that and think 1 of two things.
1. Oh damn, I was hoping for something between $1,000 - $2,000 so I'm not going to find something that will work in my price range. I'll give up searching for now. Or
2. Sweet, That is well below what I'm willing to spend. Lets do some more research and ask more questions about what I really need now I know that this is feasible.
I'm in the middle of writing my EE thesis is on embedded control systems for UAVs and this is as good a distraction as any, so here goes:
The kind of specs you're talking about you'd be lucky to get for high-end military and commercial (mini) drones. You'll either be spending tens(hundreds?) of thousands on an off the shelf model or a lot of time developing, testing, crashing and fixing your DIY solution. There are hundreds of DIY drones on the net but I doubt any of them have the kind of reliable autonomy you're talking about.
Autonomy is especially difficult, you'll need to learn a lot of control theory, kinematics, Navigation/AI and possibly computer vision. Then rememeber that you need to fuse sensor data from gyroscopes, acceleromters, GPS, compass, altitude and airspeed sensors, and that all of these sensors are unreliable/error prone. You need to be able to deal with loss of GPS link which means you need to have an alternate means of localisation(which is very difficult). Also, every commercial system I've seen requires an always on downlink and manned base-station for control, even if this isn't technically necessary, it's pretty much mandatory for safety.
Making an autonomous UAV only makes sense as a learning exercise or for R&D but it's not a good way to get any work done. If your goal is to get aerial photos, stick a camera on an RC plane, get some video goggles, a long range radio and some flying lessons.
He's not looking for a solution just yet. He wants a starting point.
Then DIY Drones would be a better starting point than Slashdot. http://diydrones.com/
"Convert any RC airplane into a fully-autonomous UAV! Just add the APM 2 autopilot to any RC aircraft and it becomes a fully-programmable flying robot with a powerful ground station and Mission Planner. APM 2 is an open source, Arduino-compatible, pro-quality autopilot. It is the most advanced IMU-based open source autopilot available today, and provides an entire UAV control system with scriptable missions with 3D waypoints, in-flight uploading of commands and powerful ground station software. "
Features include:
Return to Launch with a flick of your RC toggle switch or a mouse click in the graphical Ground Station
Unlimited 3D GPS waypoints
Built-in camera control
Fully-scriptable missions
One-click software load, and easy point-and-click configuration in the powerful Mission Planner. NO programming required!
Replay recorded missions and analyze all the data with a graphing interface
Supports two-way telemetry with Xbee wireless modules.
Point-and-click waypoint entry or real-time mission commands while the UAV is in the air
Fly with a joystick or gamepad via your PC--no need for RC control!
Built-in failsafe will bring your aircraft home in the case of radio loss
"I've got more toys than Teruhisa Kitahara."
DISCLAIMER: IANAL, so this is not legal advice.
Having said that, I am rather interested in DIY drones, and therefore, I have been following technical and legal aspects of amateur drones/UAVs/UAS' for a couple of years. I don't see any *technical* reason why what you want to do isn't possible. However, if you live in the USA, I don't believe what you want to do is legal. As I understand, the FAA requires amateur operated drones to be under line-of-site control at all times. Here are some links to help you figure out the legal restrictions for what you want to do:
DIY Drones Regulatory FAQ
FAA Advisory Circular 91-57
Electronic Code of Federal Regulations
HTH!
MCSE? No, sir...I don't do Windows. Yes, I am an idealist. What's your point?
I was rather encouraged to see other people following my own earlier train of thought. 20km puts a quadrotor out of the running, even if wind velocity did not.
Also, for mountainous terrain, you would not want a glider-style wing plan. A flatter, more slippery design such as used for "slope soarers" would seem to fit the bill most appropriately. For the kind of fuel capacity needed, I will make a rough guess and say you would probably want something in the 1.5 meter wingspan range. X-country flight competitions in the 2-meter range have been around for years.
The big obstacle here is the control of flight dynamics. There is plenty of open-source software out there for quadrotors, but I don't know about fixed-wings. I have seen some very expensive navigation implementations in the industry rags, but I suspect that if DIYers can do a quadrotor, they can do a fixed-wing as well. I suspect the main issue is that you need to make your craft steer toward the programmed flightline, more aggressively the larger the deviation, with buffering and without exceeding the design requirements of the craft. That last part is probably the trickiest.
If this is really mountainous country then you probably need very accurate, 3-axis GPS (for altitude), or some kind of ground-sensing scheme to keep from running into a hillside. I suspect the more sophisticated GPS is probably the way to go. But that also requires carefully planned routes. You will want to stay away from hillsides, preferably routing it straight along the valleys between mountains or mountain ridges.
The equipment is out there. Modern liquid fuel model engines are definitely available in any size you may want. They even make radial
And finally: autonomous navigation via GPS is already out in small drones that are on the market, but they probably aren't in your price range. Not that they could not be... I think the current commercial builders are taking advantage of the topic's fledgeling status to gouge people.
So what that all leads me to is, I would:
(1) consider a fixed-wing plane in the 1.5m to 2m wingspan range.
(2) Make sure it's a style and plan that can accommodate an engine large enough to make the range you are talking about. Keep in mind that you are working against the law of diminishing returns to a certain degree: the larger the engine you put on, the more fuel you will need for it, which requires a larger engine, which requires more fuel. In practice, there are solutions to such things, but I suggest always getting an engine that is at least a little more powerful than you think you'll need. Not a lot, but some. Otherwise you will end up with a sudden gust of wind slamming it against a hillside.
Also, what I call the "Land Cruiser" effect can occur: if it is underpowered, the engine will struggle too hard to perform the way you want and thus use more fuel than an engine designed with that torque and rpm in the center of its range. (I hereby officially christen this effect after the old-school Toyota Land Cruiser, which had a huge 6-cylinder engine that struggled to do things that were well within the design parameters of the rest of the vehicle. But with a 3rd-party kit, the stock 6-cylinder could be replaced with a V-8 of close to the same displacement, which was not only physically smaller than the original, but actually had quite a bit more horsepower and ALSO got significantly better gas mileage. It was a win-win. The original Toyota engine was designed such that common U.S. off-road scenarios were at the fringes of its performance curve, rather than the middle where it was most efficient.)
(3) You don't want realtime telemetry at a 20km range. That would be (a) way too heavy, and (b) too expensive. Instead, you will want your on-board cameras to record the video to storage for retrieval and viewing later. And there are ways to overlay not just the clock on the video, but on-board sensor readouts, as well.
So y