Address translation is the process of converting between logical addresses used within a computer system to physical addresses used by memory chips and I/O devices. It involves breaking the logical address into page number and offset, using the page number to index a translation table to find the corresponding physical frame number, and concatenating the frame number and offset to form the physical address. The main types of address translation are network address translation (NAT), port address translation (PAT), dynamic address translation, and static address translation. Address translation provides advantages like conservation of public IP addresses and protection of private resources, but also disadvantages like added delay and difficulty in troubleshooting.