Every single RC-based autonomy project has had the same fundamental safety problem. Since prototypes (and sometimes full production systems) can suddenly stop working properly, how do you design a safe RC system with autonomy? Most systems use a heartbeat signal from an RC system, so that if the RC signal goes away, the vehicle falls out of the sky. Additionally, most systems have a switch that controls whether the RC radio signals are fed into the vehicle or the autonomy signals are fed into the vehicle. Since this challenge is the same for nearly every UAV, ground robot, etc, I figure it deserves to be standardized so that people are more inclined to have a safe and effective backup system in their amateur projects.
The basic challenge goes like this:
A solution looks like this and uses two sets of muxes and a microprocessor:
UPDATE: I have since found a partial solution to this project. Multiplexing a set of 4 2-input RC inputs with a 5th channel as the selector has been done via this board: http://www.pololu.com/catalog/product/721 for $20.
UPDATE: I have since found a partial solution to this project. The validation of the signal can be done with this board, but it outputs a single digital line rather than the PWM signal needed to multiplex with the previous board. http://www.pololu.com/catalog/product/752
One could design my ideal system by feeding the throttle line into the validation board, which would be fed into a microprocessor. The microprocessor would be looping on the validation input, and outputting a PWM control signal to a 2-layer set of the mux board mentioned above. The first layer would multiplex the autonomy and radio signals, using the 5th channel, into the “control” signals. The “control” signal is then multiplexed with a set of 0 PWM signals generated by the microprocessor, using the microprocessor generated PWM control signal.