March 23 2022 – This post was written in a different time when DJI didn’t dominate the hobby drone industry. I was involved with 3D Robotics at the time and was experimenting with home built platforms.  This paper is here only for its historical perspective and really doesn’t apply anymore in the world of commercial quadcopters that fly for 30 minutes on a single charge and provide amazing HD and 4K footage.

From November 2011:

I wrote a reply to a college student on my research for my current flying quadcopter. I copied the body of the email here for others:

Heino

I be glad to give you some info based on my project and what I’ve researched on various sites.
My quad is based on the JDrones kit released a year ago:
Jdrones Kit

I went the route of ordering a kit with motors/props and electronic speed controllers included since the big effort with a quad is working out the gains for the autopilot
and the kit was the default platform for software development by the open source developers. That being said, there are literally hundreds of quadcopter frames available as kits, plans, or fully assembled.
Many people use hardware store aluminum and other materials to make their own quad frames as you’ve probably seen on diydrones.com.

There are also multiple configurations that can be built (tricopter, quadcopter, hexacopter, octocopters, and y6 frames) based on complexity, physical size (footprint) and how much is to be lifted.

The big thing on multicopters is that battery life is everything and is always shorter than you really want for any configuration. That drives the choice of configuration, batteries, electronic speed controllers, and motors.
Before I bought my quad kit and started down the path, I was inspired by the engineering of an individual on Rcgroups – Old Man Mike who has gotten really long range and amazing battery life with his choice of hardware. This thread on RCgroups was my inspiration for my project – although Mike chose Mikrocopter hardware which is the benchmark in reliable multirotor copters and priced accordingly. (> 2k $) I’ve used several of his ideas on my quad – particularly with my video transmission system. Check out some of his high definition video he shot:
http://www.rcgroups.com/forums/showthread.php?t=1147430
OMM uses a system with lower quality live video and an on board hi definition camera for site seeing. I’m going down this path on my quad – but for now I have an SD video system integrated right now

To answer your questions:
There are many small cameras available. Your choice is analog standard definition video, digital video cameras (SD and HD) or point and shoot cameras.
Most fliers are flying small security cameras if using analog video, or the GoPro which is a light, awesome HD camera.

Here is the analog camera I’m using:
http://www.123securityproducts.com/kpc-vsn500nhb.html

This has worked well for me and has survived at least 7 major crashes and keeps on going.
There are many analog video cameras available and the major advantages – no lag – good for real time view, digital video cameras all have some lag – some substantial. Lag is a problem for remote flying of an aircraft so digital is usually out for live feeds.

Low power (30 ma at 12 volts). Disadvantages – GPS interference – can cause major problems on a quad since the components are located at the center of the aircraft in close proximity. I was able to minimize this issue on my quad
with adding a small capacitor to the camera.
Here is a supplier with several different analog cameras – some with more GPS interference than others:
http://www.readymaderc.com/store/index.php?main_page=index&cPath=11_62
I bought the RMRC-520 and it did not work well and caused fairly severe GPS interference.

For good photography from the air, I plan to use a second digital camera.
Here is the camera I plan to add early next year – Go Pro
http://www.readymaderc.com/store/index.php?main_page=product_info&cPath=74&products_id=234
It is used in many fixed wing and multicopters and is very light.

some sample video of the GoPro mounted on an Arducopter with servo stabilization:

I plan to implement this camera mount and add 2 servos for stabilization from the autopilot. The Go Pro can transmit live video – but with lag and at a less than HD resolution.

For transmitting the video to the ground there are several different options, but the key again is avoiding interference with other transmitter/receivers on the quad. Here is a list of the transmitter/receivers on my quad:
1. Futaba 14 channel receiver – 2.4 Ghz
2. Xbee digital data link transmitter – for connection to the ground station to monitor the flight – 900 Mhz
3. Fox 800 mw video transmitter – 1.280 Ghz
4. GPS receiver – 1.4 Ghz.

The quad is quite a challenge for EMI interference and it is important to not choose components that don’t transmit/receive on the same frequencies. You can buy legal video transmitters in the following video ranges: 400 Mhz, 900 Mhz, 1.2 Ghz, 2.4 Ghz, and 5.8 Ghz. I also got a Technicians class Ham license so that I could legally transmit video on the transmitter. This is a general requirement in the US, but is not as big an issue in the rural areas. Since I fly near the city I wanted to be able to fly at local R/C clubs without any issues.

I chose 1.280 Ghz since lower frequencies offer better range and less multipath interference (signals bouncing around objects such as houses). If you did choose a tranmitter frequency such as 2.4 ghz or 900 Mhz you would get interference in the video picture. Another advantage of 1.280 Ghz is that the antenna is small. 5.8 Ghz is only for fairly short distances and has high multipath interference. 400 Mhz has fairly large antennas for good performance and is not good for a small quad.

I bought a Chinese video transmitter and receiver at this site:
http://www.bevrc.com/bev-13g-800mw-p-52.html

I’ve attached a photo of the installation of the video transmitter and the dipole antenna on my quad. The copper board in the line is a video filter to suppress interference from the harmonics of the 1.280 Ghz signal. It can interfere with GPS and the R/C receiver, so it is in line to help with the issue. I also mounted the dipole antenna away from the core electronics to help with interference. The second photo is the video camera and tilt pan mount.

There are other transmitters available but the one I linked to works very well and has a range of 2 miles or more with a good high gain ground antenna.
I’m using a high gain patch antenna with antenna tracker driven by the GPS downlink from the quad. Here is a link to my diydrones post covering this project. The antenna is highly directional and can track the quad as it flies. There is also a video I did covering the software mods I made to support this function:

http://diydrones.com/profiles/blogs/modified-ardustation-software

For flying the quad I added an OSD to the video camera output to draw status info on the video downlink. Here is a link to the software project I posted to diydrones for the OSD data interpreter. There is a live video sample showing the osd and live video. The camera is having some contrast issues which explans the weird darkening of the video – I plan to figure out this issue, but it hasn’t been a problem on other videos I’ve recorded.
http://diydrones.com/profiles/blogs/using-a-remzibi-osd-with

As to the autopilot, I chose the diydrones APM and oilpan IMU shield board. There are other multicopter IMU solutions, but the APM uses the Arduino form factor and development software so it is easy to work with. There are a number of people using the APM for fixed wing, multicopters, and vehicles, so it is versatile and the software is getting quite good. Since a quad is almost 100% a software problem, it is important to get rock solid software and the current version has been in development for 11 months and is finally approaching a release candidate. Navigation and mission planner integration is first rate and it supports ground changes to the stored flight plan.
http://code.google.com/p/arducopter/wiki/AC2Assembly

As to the physical parameters and cost of my setup here is a list of my setup:

Quadcopter frame $320 (with motors and electronic speed controls for each motor, props, power distribution board)
APM and oilpan IMU – autopilot $250
GPS for navigation. $40
Magnetometer – for heading measurement $50
Sonar sensor – for low altitude landing, altitude hold $50
Xbee 900 MHZ transmitter/receiver – for datalink to the ground $150 http://store.diydrones.com/Xbee_Telemetry_kit_p/kt-telemetry-xbee.htm
video camera – $80
Video transmitter/receiver $90
Microstrip video transmitter RF filter – $15
Transmitter antenna – home built dipole – $10
Low loss video transmitter cable and connectors $25

1.280 Ghz patch video antenna $60

Tilt/pan antenna tracker assembly $90
Antenna tracker electronics and extra xbee for receiving GPS position $100
USB to serial cable for Arduino programming $20

The following R/C receiver/transmitter is really overkill but it is what I use:
Radio control receiver $200 (Futaba 14 channel)
Radio control transmitter $800 (Futaba 12FGH) –

The following R/C unit is used by many and works well (Update 10/22/2012 – the 9x is having issues with the way failsafe works – causing fly aways for some. May not be the best choice):
Turnigy 9 Channel Radio
$70

For OSD display functionality:
OSD $110
Current sensor – measuring battery current draw for display $45
Arduino datalink processor for OSD symbology $20

For high definition video camera option – with camera mount, and go pro – $300

For powering the quad – I tend to use 2200 mah 3S Lipos for 7 to 9 minute flight times. I also use heavier 3300 mah 3S Lipos for 10 to 12 minute flight times. Higher capacity batteries give marginally more flight time but the
limit will be about 15 minutes with 5000 mah batteries.

I buy all my batteries from www.hobbyking.com (China) and the cost per battery is about
2200 mah $10
3300 mah $22
5000 mah $40
http://www.hobbyking.com/hobbyking/store/uh_viewitem.asp?idproduct=9183

You’ll need 1 or more Lipo chargers plus a 12 volt charger source for charging the batteries. You can figure about $40 a charger and $50 for a computer power supply to power the charger. The chargers from hobbyking are pretty good.

To get longer than 15 minute flight times will require going to larger motors, larger frames, and perhaps 6 motor configurations (hexa copter) It would be possible to get 15 to 20 or 30 minutes as in the OMM video at the beginning of this email. OMM spent several thousand dollars for his setup, but it does work well. The APM will handle hexa and octocopter configurations and the cost of motors and electronic speed controls needs to be factored in. It is also possible to go to 4S batteries (4 cells) to get longer flight times.
The motors and speed controls are all sourced from Chinese sources and are around $20 per motor and $20 per speed controller. Larger configurations will go up and US sources of motors will be 3 to 5 times as expensive. I’ve had very good success with the Chinese sources and buy them from Jdrones.com and www.hobbyking.com.

It is critical to have a balance between the frame, motor size, prop size, speed controller capacity and battery capacity when designing your own frames or mixing and matching. That is one reason I bought a kit to avoid the extra engineering time required. Also using third party frames requires tuning of the autopilot PID settings for stability and navigation. This is a non trivial process and requires many flights and a bunch of tweaking. I’m using almost default pid values, but I have been doing some final tuning of my navigation stability and that has taken 3 weeks of work waiting for good weather and flying at least 20 flights. I’m finally getting it tweaked in, but the number of flights will go up substantially if you go with an unproven design.

Another frame option is here: aluminum http://store.diydrones.com/Arducopter_3DR_partial_kit_p/kt-ac3dr-01.htm
My current frame has a mix of aluminum and polycarbonate plastic and actually is quite resistant to crashes. I’ve had several crashes and usually only bend one of the 4 arms and break some of the nylon bolts. All metal designs will be more rigid and will suffer more damage in a crash. There is a tradeoff between how rigid the frame is and how crash resistant the design is.

My quad’s current weight with 2200mah is about 3 lbs. A 10 lbs quad is significantly larger and would be very expensive – just for your planning purposes. Weight will directly affect the size of the batteries required and there have been papers written on flight time versus weight and battery/motors required.

Here is a link to a paper giving the math behind battery capacity for fixed wing and helicopters written by one of the diydrones members.
Battery Life Paper

Anyway, I hope this narrows your requirements definition and gives you an idea if this would be feasible. Quadcopters are extremely stable, easy to fly, and easy to automate at the disadvantage of flight time. Weight is the key factor for flight time and can be an engineering challenge. I have over 100 flights on my quad in the last 8 months and find it a real challenge but quite rewarding.

By the way – the ground support software and logging capability of the APM is amazing: I’ve attached an example of the log review function showing sonar and barometric altitude of one of my flights. I’m using this capability to analyze my flights.

Hope this helps and I’ll be glad to answer any other questions.