Road traffic crashes are the leading cause of death among young people between 10 and 24 years old. In recent years, both academia and industry have been devoted towards the development of Driver Assistance Systems (DAS) and Autonomous Vehicles (AV) to decrease the number of road accidents. Detection of the road surface is a key capability for both path planning and object detection on Autonomous Vehicles. Current road datasets and benchmarks only depict European and North American scenarios, while emerging countries have higher projected consumer acceptance of AV and DAS technologies. This paper presents a selected Brazilian urban scenario dataset and road detection benchmark consisting of annotated RADAR, LIDAR and camera data. It also proposes a novel evaluation metric based on the intersection of polygons. The main goal of this manuscript is to provide challenging scenarios for road detection algorithm evaluation and the resulting dataset is publicly available at www.lrm.icmc.usp.br/dataset.