This paper presents a methodology of identification to dynamic modeling for designing an open-loop controller that linearizes the relationship between linear and angular velocities, in the stationary state, of a differential robot with respect to motor speed commands. The proposed approach is particularly valuable to make a nonlinearity compensation for low-cost robots that lack onboard sensing capabilities. By utilizing visual measurements obtained from cameras, i.e. third-person, the sensorless identification process enables the design of an open-loop controller without the need for embedded sensors. This cost-effective solution facilitates the application of linear dynamic controllers in robotic systems where precise control and linear dynamics are required, even in scenarios with limited resources. The proposed methodology offers a practical and accessible solution for enhancing the capabilities of low-cost robots and expanding their range of applications.