Training Deep Networks with Synthetic Data: Bridging the Reality Gap by Domain Randomization | Read Paper on Bytez