Swarm robotics is an approach to multi-robot control based on social insects and other natural systems, which shows self-organization and emergent characteristics, with disruptive applications on robotics and possibilities in a variety of areas. But, being a relatively new field of research, there are few experimental platforms to its study, and most of them are crafted for very specific tasks and algorithms. A general study platform of swarm robotics, by itself, is a non-trivial technological deed and also a very valuable asset to a research center willing to run experiments on the topic. In this work, two important algorithms in multi-robot collaborative control strategies are studied: path finding and collective transport. A complete analysis of the biological mechanisms, models and computer abstractions that resulted in the development of those algorithms is shown. To perform the multi-robot experiments, several “iRobot Create” mobile robots are employed. Virtual sensors and virtual walls are implemented in real time in the experimental system through cameras and especially developed computer vision software. Virtual sensors allow the incorporation of ideal sensors in the experimental system, including complete models of real sensors, with the possibility of adding virtual noise to the measurements. This approach also allows the use of sensors to detect virtually created objects, such as virtual walls or virtual pheromones. Each physical robot has a customized embedded system, based on the ARM microprocessor, which receives the virtual sensors readings through a radio link in an also customized protocol. The behavior of each autonomous agent is locally calculated using the high-level programming language Lua. Even though the virtual sensor readings are transmitted from an external centralized computer system, all behaviors are locally and independently calculated by each agent. The adaptations of the studied algorithms to the platform with virtual sensors are analyzed, along with its limitations. It is shown that the experimental results using virtual sensors are coherent with results from the literature using very specialized and expensive robot/sensor setups. Therefore, the developed platform is able to experimentally study new control strategies and swarm algorithms with a low setup cost, including the possibility of virtually incorporating sensors that are still under development or not yet commercially available.