I ran into a machine that for some unexplained reason would hit a kernel bug and then get a kernel panic.
This usually left the machine in an usable state, requiring a reboot.
After some googling, I found an answer.
Below are the important bits:
# vi /etc/sysctl.conf
The arguments below, instructs the kernel to reboot 10 seconds after a kernel panic occurs
kernel.panic = 10
It is of course possible to make the 10 seconds as long as you need.
Save the file and apply the setting.
# sysctl -p