Odometry is one of the basic features for any blind assistance device to be able to navigate its location in indoor and outdoor environments. Hence, many papers try to cover this topic in many ways. This paper aims to provide three different ways to find odometry with optimum accuracy and precision. The first two methods that depend on tracking anybody by x-shape method and special color pixels by optical flow method, can interact with each other due to different conditions. The third method can work at any condition because it depends on the difference between frames and fitting between them by the machine learning system. Also, one of the needed features is to define the relation between obstacles and floor level to give the ability to avoid them without any problem from detecting the floor as an obstacle. This paper also deduced from this feature a crossroad algorithm to aid blind and impaired people to cross car roads with more safety.