|Subject:||Re: [Paparazzi-devel] simultaneously using two autopilot systems for reliability|
|Date:||Mon, 25 Nov 2013 12:18:32 +0200|
> How will you (the FPGA or the second autopilot) know if the primary autopilot freezes?
We are planning to take a signal from the autopilot which regularly kicks the FPGA (like regular watchdog timer of microprocessor). Sometimes, the system timer led of autopilot works (blinks at a different rate then 1Hz), but the imu does not work. Therefore, we may also get the AHRS align led.
Another option is, the FPGA can listen the Modem RX link. And from GCS, we can send a command to the FPGA showing the autopilot that we want to switch. When the FPGA detects this command, it switches the autopilot.
>Why is your autopilot freezing? Have you ruled out brownout? Can you reproduce it on the bench?
When the modem TX power is 25mw or higher, then the IMU does not work properly and also does the autopilot. It still sends messages, but the behaviour is not normal. The PFD screen does not work properly etc. We can see it on the bench. When the TX power is 1mW, then it works properly.
Sometimes, it completely freezes and the leds do not blink. I think that we have also problems in upload. We haven't yet tried the verification procedure of uploaded stream. The autopilot starts working, but when it jumps to a problematic part of the code, then it stops.
How will you (the FPGA or the second autopilot) know if the primary autopilot freezes?
We did this with 74LS157 multiplexer chip & AVR for the Outback Challenge 2013 competition. It was a proto-board/wire-wrap implementation (no PCB to share sorry) but another team has made a cleaner version here:
That device is for complying with competition rules, not solving a reliability problem. The extra complexity is dubious safety in my opinion.
Why is your autopilot freezing? Have you ruled out brownout? Can you reproduce it on the bench?
|[Prev in Thread]||Current Thread||[Next in Thread]|