The problem of tracking a moving target with a nonholonomic mobile robot, by using sensor-based control techniques, is addressed. Two control design methods, relying on the transverse function approach, are proposed. For the first method, sensory signals are used to calculate an estimate of the relative pose of the robot with respect to the target. This estimate is then used for the calculation of control laws expressed in Cartesian coordinates. An analysis of stability and robustness w.r.t. pose estimation errors is presented. The second method consists in designing the control law directly in the space of sensor signals. Both methods are simulated, with various choices of the control parameters, for a unicycle-type mobile robot equipped with a camera. Finally, experimental results are also reported.