Robot is being more intelligent with the visual information and analysis. Specially, the functionality of the object or scene recognition with a high accuracy is essential for the autonomous control and navigation. However, they are too complicated to be implemented in an embedded platform for mobile robots driven by a battery. In this paper, for the high processing speed with low power consumption and small form factor, we propose the architecture of System-on-Chip including the Very Large Scale Integration (VLSI) based on Speeded Up Robust Features (SURF). Finally, the dedicated chip using a 55nm low power process is implemented and integrated with a commercial robot successfully.