To do this I used one of the Mega 2650 boards that is in a small form factor configuration. This gave me 5 volts to run the actual rc receiver from and more than one UART to talk to the usb, input and output parts of the buses. I think you only need 2 UARTs but as I had a bunch I just used separate ones.
The 2560 also gives a lavish amount of ram so using ROS topics doesn't really matter. I have 9 subscribers and 1 publisher on the 2560. The 9 subscribers allows sending temp, voltage, rpm info back to the remote and flexibility in what is sent so that can be adjusted on the robot itself.
I used a servo extension cable to carry the base 5v, ground, and rx signals from the ibus out on the rc receiver unit. Handy as the servo plug ends can be taped together for the more bumpy environment that the hound likes to tackle. I wound up putting the diode floating between two extension wires on the (to tx) side of the bus.
The 1 publisher just sends an array with the raw RC values in it. With minimal delays I can get a reasonably steady 120hz publication of rc values. So now the houndbot can tell me when it is getting hungry for more fresh electrons from a great distance!
I had had some problems with the nano and the rc unit and locking up. I think perhaps this was due to crystals as the uno worked ok. The 2560 board has been bench tested for 30 minutes which was enough time to expose the issues on the nano.
