Maybe someone could provide some comments on the following (maybe crazy) idea.

Could a low-cost UAV be created with the following parts:
1. an airframe that was 'inherently stable', i.e. enough dihedral that you didn't need to worry too much about roll.
2. A low-cost ($60) GPS module
3. An accurate (quick response) barometric altimeter (like the intersema module)
4. Some kind of processing engine (PIC or similar) which sits between the Rx and the servos/speed controller.

I see a lot of work on various forums in the direction of Gyros, and Accelerometers, but are they really necessary?. I can understand if the airframe wasn't very stable (low, or even negative dihedral), but if it was stable would that simplify the problem?

The GPS module would provide accurate Longitude/Latitude information for Rudder control, and the barometric altimeter would provide accurate altitude information for elevator control.

Looking forward to your comments...

Try http://tam.plannet21.com/ for a trans atlantic model that maybe will give you some ideas.

Regards Peter

Anthony,
Yeah, it is possible to build a UAV for $200 over the cost of the plane. This guy ep92 sells an autopilot that controls the servos for around $130, handheld gps for $60 and the altitude is already done through the gps satellite position. The only thing left is airframe stability, which if it is already a high wing trainer and the servo gain is set low enough is do able. If you really need the stability you can get an FMA co-pilot for $100 and mix it with the ailerons for more stability.

Many thanks for the replies Guys. The TAM link is very interesting, but these guys used a gyro-based system, which I was trying to avoid.
I'm worried about using the CoPilot where I live, which is right at the foot of a mountain. Has anyone tried this I wonder?

Try:
http://gpsinformation.net/main/autonomous.htm
http://www-2.cs.cmu.edu/afs/cs/project/chopper/www/capability.html
http://www-robotics.usc.edu/~avatar/hardware.htm
http://www.aber.ac.uk/compsci/Research/robots/airborne/
http://www.micropilot.com/
http://sun-valley.stanford.edu/~heli/info.html#GPS
http://www.rotomotion.com/prd_REV2.4.6DOFK.html
http://www.u-nav.com/

One day I will discover how to do proper links in Quick Replies!!

regards Peter

Aha, hidden in the u-nav link is, I believe, exactly what I was looking for, thanks Peter:

From: http://www.u-nav.com/FAQs.html
Attitude Control Most UAV autopilots perform many tasks, including navigation, altitude hold, flight path guidance and attitude control. Attitude (roll and pitch) control is the most basic function any autopilot must perform because without proper attitude control, all of the other functions are pointless. The specific attitude control requirements for a particular air-vehicle are usually driven by its particular aerodynamic characteristics. The type of attitude control must be matched to the type of air-vehicle you have chosen. From a system perspective, your choice of airframe will have a significant effect on the overall complexity of the autopilot system and the resulting costs (and weight). The aerodynamic stability of the air-vehicle will determine the response and displacement range of the attitude control. In some cases, an attitude control may not be required at all. For example, a rudder steered airplane (no ailerons) wouldn't require any roll attitude control at all, since the vehicle has positive aerodynamic roll stability. Similarly, a powered-parachute or Blimp wouldn't require any attitude control and a boat or ground vehicle wouldn't require any pitch control.

There are several solutions for maintaining air-vehicle attitude, some more complex (and expensive) than others. PICOPILOT-NAV-A uses a rate based roll attitude control. Rate-based wing levelers have been used on GA (light manned) aircraft since WWII. They are relatively simple and cost effective, however they do have bank angle limitations (45deg. max.) and have a slower response than an IMU based attitude controller. A more precise (and expensive) solution, is a 6DOF IMU (inertial reference) or vertical gyro. An IMU is a good choice for an airframe that responds quickly or is relatively aerodynamically unstable. Horizon sensing attitude controllers such as the FMA CO-PILOT or the FUTABA PA-2 can be used as external attitude controllers. Horizon sensors are relatively inexpensive and are fast enough to handle airframes that would normally require an IMU. Horizon sensors are not as reliable as inertial sensors because they do not work in all weather or terrain conditions. To use PICOPILOT with an external attitude controller, select a PICOPILOT-NAV-R (rudder output) and a PICOPILOT-ALT-T (throttle output) since these versions do not contain an internal attitude controller. For an external attitude controller setup, PICOPILOT is connected between the RC receiver and the manual inputs of the Horizon controller (FMA CO-PILOT) where it provides turn and climb/descend commands. The response time required for an attitude control is determined by the aerodynamic response of the airframe, which involves the moments of inertia and control surface authority (which increases with airspeed). In simple terms, a small, fast plane with short wings and short tail will require the most robust attitude sensor, typically an IMU. The performance of the autopilot itself is not the only element however, other airframe component characteristics, such as the servo speed and control surface authority contributes to the overall response of the attitude control loop. While 1 Hz (once per second) may be adequate for a motor-glider or trainer air-vehicle, a jet or flying-wing may require a control loop as fast as 50Hz. When using RC equipment, servo transit time is usually the limiting element in an attitude control loop. A fast RC servo typically has a (full range) travel time of 0.1sec. By itself, that would limit the control loop to 10Hz (at full travel) or 20Hz at 50% travel, 40Hz at 25% travel.

Good idea, you may want to look at the UAV forum, and cross post this post there... there's some activity there.... It's all the way at the bottom.... in the trial forums

The simplest.. for cost purposes.. plane would be a converted free-flight.
These are purposely designed to be self-stabilizing around the 3 axes.
And fly at a single speed.
Controls then act more like trimmers than manuvering controls..
Moderate changes to pitch, roll and yaw, with the plane's geometry handling the stability.
A standard rudder-elevator-motor control system is adequate.
An FMS Co-pilot will assist in keeping the nose up and the wings level.
Power management affects climbing-diving.
Finding a "sweet spot" in the throttle travel might be sufficient, but a speed governor wouldn't be out of place, if the plane goes beyond visual range.
For BVR, something more might be needed, to assure the attitudes, altitudes and airspeeds aren't varying beyond limits.