The use of agricultural robots is emerging in a complex scenario where it is necessary to produce more food to feed a crescent population, decrease production costs, fight plagues and diseases, and preserve nature. Around the world, there are many research institutes and companies trying to apply mobile robotics techniques in agricultural fields. Mostly, large prototypes are being used and their shapes and dimensions are very similar to tractors and trucks. In the present study, a small-scale prototype was designed, aiming to facilitate the controller development phase and the execution of experiments in the university using a farm-like scenario (before validating the controllers in a real scenario). It is important to highlight that all control parameters were parameterized to allow the control portability to other prototypes. Helvis is an electric small-scale car-like platform whose traction and steering systems are powered by Maxon motors and driven by EPOS2 boards. Its navigation system uses 2 LiDAR sensors (UTM-30LX) to scan the environment (one in the front and the other in the back) and an Inertial Navigation System (IG500N) to estimate its orientation. In this paper we present experiments carried out in rows of a corn crop field. As previously mentioned, before making the experiments in a real farm, a farm-like scenario was constructed in the lab to calibrate the controller parameters. Since each cornrow constitutes itself a discontinuous wall, a filter based on the LiDAR data was developed in order to create virtual continuous walls. So, these virtual walls were used as references for a wall-follower control system. When it is possible to create walls in both sides of the robot, the navigation problem can be simplified to moving the robot in a virtual aisle. The filter calculates the distances between robot and virtual walls, which are used as input data for the fuzzy controller responsible to keep the robot in the path between corn rows. Its output signal acts in Helvis’ steering system. Real environment experiments allowed adjusting fuzzy rule set and improving robot performance.