The problem of controlling the magnetic path tracking guidance system of lightweight flexible Automated Guided Vehicles (AGVs) is presented. Light AGVs are usually designed to move small payloads. They have limited load weight capacity, since load weight values large with respect to the vehicle weight can significantly reduce the motion performances. So, guidance controllers are usually conservatively designed. Also, if the payload weight ranges in a large span, light AGVs path tracking controllers are tuned on the basis of the largest expected payload, with strong performance limitations at smaller weights. In this paper, a state space dynamical model for the differential drive path tracking system of the AGV has been developed. In order to guarantee desirable control specifications for a wide range of possible masses (50 < M < 1000kg), the payload mass is considered as uncertainty in the model. Then we analyze robust stability domain in the controller parameter space via randomization technique. Namely, we describe the regions where the controllers satisfy control action constraints (current limitation), desired speed defined by transient time values and regions such that ||T(s)||∞ ≤γ . This approach allows us to design controllers satisfying simultaneously several control objectives expressed both as common engineering specification and as mathematical constraints. Finally, two set of controller parameters have been chosen and experimental tests have been performed on a vehicle used in a commercial vehicle engines assembly line.