Design and Implementation of an Embedded Vision System for Industrial Robots

Sammanfattning: Enhanced vision system attached to any industrial robot undoubtedly increases its accuracy and precision. Today, vision in industrial robots is facilitated almost exclusively via external or fixed cameras which may cause rather inconvenience due to obstacles in the line of sight. In this master thesis project, an Embedded Vision System is designed and developed to stream live video of a robot's focus point while being attached to its arm/actuator. Within the scope of this work, a working prototype has been achieved which is capable of producing live video stream on 640 X 480 resolution with a frame rate slightly above 9 FPS, having 256 colors for each pixel, displayable on a regular LCD display monitor. The system has been realized in a Spartan 6 platform and an Aptina image sensor has been used to acquire pixel information. I2C interfacing has been used to program the image sensor, data transfers have been facilitated by DMA cores, an off-chip DDR2 memory has been used for frame buffer and HDMI has been used as video out. Feasibility of adding Ethernet transmission capability has also been investigated.