Advanced search
Start date
Betweenand

Project CARINA: localization and control

Grant number: 13/24542-7
Support type:Regular Research Grants
Duration: June 01, 2014 - May 31, 2016
Field of knowledge:Physical Sciences and Mathematics - Computer Science
Principal Investigator:Denis Fernando Wolf
Grantee:Denis Fernando Wolf
Home Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

The idea to develop fully autonomous vehicles has been studied since the 80s and has several practical applications such as reducing the number of accidents on roads and highways, increase the mobility of the elderly and people with special needs and increase the efficiency of transit in general. Despite its importance and the great attention that this issue receives by researchers in robotics, this issue still presents many challenges and open questions due to the high complexity of it. The CARINA project proposes the development of a system of autonomous driving based on sensor fusion algorithms and artificial intelligence. This research plan proposes the development of algorithms for the location and control of autonomous vehicles, within the context of the CARINA project. The location is a key information for autonomous vehicles by allowing the execution of operations as overtaking, conversion and driving on urban roads. Mechanisms that provide the location information should be sufficient to ensure that the vehicle does not encroach on the track in the opposite direction or in an inappropriate place. GPS sensors of low and medium cost have relatively high uncertainty, especially when it comes to urban environments. Tall buildings and trees can cause significant errors in position and can also make the sensor out of stock for a while . Even GPS costly sensors are subject to this kind of problem. In this project we propose the use of continuous metric maps, based on Gaussian process for the localization task of autonomous vehicles on urban roads. This type of chart represents the environment through a covariance matrix of a multivariate Gaussian function, and has several advantages over the methods used in the job location in the literature. Among them is the continuous representation of space without the need for discretization (inaccuracy) , and high power of inference areas not detected by the sensors .We intend to adapt the continuous metric maps for different types of environmental information to be stored and used for the location as: guides, traffic lanes and other markings. After the step of mapping the environment, we intend to use the Monte Carlo method to estimate the vehicle's position in real time, fusing sensor information available map of the environment. Besides the location, this research plan also addresses the control of autonomous vehicles. The steering and speed control is critical to the operation of autonomous vehicles in urban environments. This project proposes the use of adaptive control theory for the solution of this problem accurately and robust. (AU)

Articles published in Agência FAPESP Newsletter about the research grant
Brazilian university tests self-driving taxi service 
Articles published in Pesquisa FAPESP Magazine about the research grant:
The driverless future 

Scientific publications
(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)
MASSERA FILHO, CARLOS; TERRA, MARCO H.; WOLF, DENIS F. Safe Optimization of Highway Traffic With Robust Model Predictive Control-Based Cooperative Adaptive Cruise Control. IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, v. 18, n. 11, p. 3193-3203, NOV 2017. Web of Science Citations: 15.

Please report errors in scientific publications list by writing to: cdi@fapesp.br.