Generally, an intelligent humanoid robot system should have dual arms, mobile ability and stereo vision for establishing the interactive capability in unknown environment. Stereo vision system detects unknown objects or obstacles from their visual range and estimates 3D coordinates in a working environment. Dual arms and mobile platform are employed to approach the detected object and execute interaction based on visual servo information. Here a wheel-based 10 DOF dual arms mobile robot with FPGA hardware control structure and a Digital signal processor (DSP) based CMOS stereo vision system was designed and built to construct an distributed embedded low cost visual guided intelligent robotic system. The intelligent fuzzy sliding mode control strategy was employed to design the robot arm and platform motion control system. The experimental results show that the proposed stereo vision algorithm has good recognition ability and accuracy. Each joint angular tracking error of humanoid robot can keep within 0.05° by using distributed FSMC controller.