Funded by 863 National High Tech Project “Research on Guide Service Mobile Robots with Dynamically Changed Environments”, this thesis focuses on differential wheeled mobile robots with dynamically changed indoor environment. The two research topics of this thesis are motor control and visual tracking navigation of differential wheeled mobile robots. The novel work and contribution of this thesis can be summarized as follows: A kind of brushless DC motor drives whose maximal power reaches up to 1500 walts are developed based on AVR Mega48 MCU. A novel analog mosfet bridge drive circuit used in the motor drives is presented. The kinematic model and dynamic model of mobile robots whose mass centers are not necessarily on the common axis of two driving wheels nor on the middle of two wheels are proposed. Based on the dynamic model, the differential equation between speeds and torques of two wheels are derived. The differential equations of two DC motors are adopted for the derivation of the state equation of mobile robots. A novel control law which is called DMMC (Dynamic Modeling based Motor Control of mobile robots) is proposed for controlling the two driving motors of differential mobile robots. Simulation shows that the DMMC controller responds to the inputs without static error. A novel adaptive bandwidth mean shift algorithm or ABMS is presented for solving the bandwidth selection problem and bandwidth updating problem in the classical mean shift algorithm. The adaptive bandwidth mean shift algorithm in 2D case or 2D-ABMS for short is thoroughly analysed. Based on the 2D adaptive bandwidth mean shift algorithm, a novel object tracking algorithm, which is called adaptive bandwidth mean shift object tracking algorithm or ABMSOT for short, is proposed. Experiments show that the ABMSOT is capable of simultaneously tracking the position, scale and rotation in real-time. Based on the 2D adaptive bandwidth mean shift algorithm, a novel adaptive bandwidth object detection algorithm, or ABMSOD for short, is proposed for object detection and localization. Experiments verify that the ABMSOD is not only able to detect the existence of certain objects, but also able to find the best position, scale and rotation of the objects in a given image. A novel open modular mobile robot software architecture with operating system compatibility, communication abstraction, device abstraction and interface specification is presented for long time navigation software developments. Based on the open modular software architecture, a navigation software NAVOP(NAVigation OPeration system) is developed with the help of ARIA. Tracking navigation experiments is fulfilled using ABMSOT and ABMSOD algorithm.