| Grant number: | 14/09096-3 |
| Support Opportunities: | Scholarships abroad - Research Internship - Doctorate |
| Start date: | October 07, 2014 |
| End date: | June 06, 2015 |
| Field of knowledge: | Physical Sciences and Mathematics - Computer Science - Computer Systems |
| Principal Investigator: | Denis Fernando Wolf |
| Grantee: | Alberto Yukinobu Hata |
| Supervisor: | Fabio Tozeto Ramos |
| Host Institution: | Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil |
| Institution abroad: | University of Sydney, Australia |
| Associated to the scholarship: | 12/02354-1 - Localization for autonomous vehicles in urban environment using continuous occupancy maps, BP.DR |
Abstract Localization is a fundamental information for autonomous cars by making possible operation such as overtaking, turning and driving in the streets. Commercially available GPS devices have a relatively high imprecision, mainly in urban streets due to the presence of constructions that block the GPS signal reception. A widely adopted solution for localization estimation is matching the sensor information with a pre-built map of the environment. Metric maps have been used for vehicle localization, but they are sensible to occlusions and associated to high memory demand. In urban streets, road features (road shape, horizontal and vertical signalization) are usually used to vehicle localization. Thus, pedestrians and other cars may difficult the road feature detection and consequently prejudice in the vehicle localization.Recently, Simon O'Callaghan and Fábio Ramos proposed a novel mapping method named Gaussian process occupancy map (GPOM) that is robust to occlusions and can represent the environment through a continuous space, so making unnecessary the grid representation. Using GPOM representation it is possible to estimate the occupancy of areas blocked by obstacles and sensors couldn't observe. Considering the GPOM properties, the adoption of this representation for vehicle localization problem can be advantageous due to its robustness to obstacle occlusions.This project proposes the development of a localization method that use GPOM structures to obtain the vehicle position in urban areas. The environment perception will be performed only by a multilayer LIDAR sensor. The main contribution of this work will be a localization method that doesn't require additional processing to deal with obstacles. Moreover, the representation of large environments can be optimized by not requiring the map discretization, as classical metric maps. The localization will be validated in urban streets using an autonomous car prototype. (AU) | |
| News published in Agência FAPESP Newsletter about the scholarship: | |
| More itemsLess items | |
| TITULO | |
| Articles published in other media outlets ( ): | |
| More itemsLess items | |
| VEICULO: TITULO (DATA) | |
| VEICULO: TITULO (DATA) | |