Abstract-Path estimation is a big challenge for autonomous vehicle navigation, especially in unknown, dynamic environments, when road characteristics change often. 3D terrain information (e.g. stereo cameras) can provide useful hints about the traversability cost of certain regions. However, when the terrain tends to be flat and uniform, it is difficult to identify a better path using 3D map solely. In this scenario the use of a priori knowledge on the expected road's visual characteristics can support detection, but it has the drawback of being not robust to environmental changes. This paper presents a path detection method that mixes together 3D mapping and visual classification, trying to learn, in real time, the actual road characteristics. An on-line learning of visual characteristics is implemented to feedback a terrain classifier, so that the road characteristics are updated as the vehicle moves. The feedback data are taken from a 3D traversability cost map, which provides some hints on traversable and non-traversable regions. After several re-training cycles the algorithm converges on a better separation of the path and non-path regions. The fusion of both 3D traversability cost and visual characteristics of the terrain yields a better estimation when compared with either of these methods solely.

No abstract

This paper presents an exploration of the ceiling analysis of machine learning systems. It also provides an approach to the development of pedestrian recognition systems using this analysis. A pedestrian detection pipeline is simulated in order to evaluate this method. The advantage of this method is that it allows determining the most promising pipeline's elements to be modified as a way of more efficiently improving the recognition system. The pedestrian recognition is based on computer vision and is intended for an autonomous car application. A Linear SVM used as classifier enables the recognition, so this development is also addressed as a machine learning problem. This analysis concludes that for this application the more worthy path to be followed is the improvement of the pre-processing method instead of the classifier.

According to WHO, traffic safety is one of the major concerns of this decade. Due to this, researchers worldwide aim to reduce the large number of fatalities in traffic accidents. This paper presents GISA as a contribution to this scenario. It consists of a platform for testing autonomous car algorithms. Firstly, some requisites to build an autonomous vehicle are presented, followed by sensors, their placement and ROS middleware. Some tests are presented to check the platform performance, including localization issues and obstacle detection. Results show that GISA is a consistent platform for implementation of autonomous car algorithms.

No abstract