Hum, how about if I ask one robot what the other one think and not directly what they think ?
1. If I ask Robot A what door Robot B will say is the freedom door, he will lie and point to the opposite, therefore the Death door.
2. If I ask Robot B what door Robot A will say is the freedom door, he will tell the truth and point to the Death door because he knows Robot A is a liar.
The trick is Robot A will always lie, so either he will lie directly (about B's answer) or indirectly (when A will truthfully say he'll lie).
So I have to ask one Robot which door the other Robot will point out as the good door, then they will always point the wrong one out because of that liar Robot A. And I will have to take the other door.
Don't know if I'm clear