CSIR to enter robot race

Johannesburg, 08 Jul 2008

A CSIR-University of KwaZulu-Natal team is developing an autonomous automobile for the next Darpa Grand Challenge.

Darpa, the US Defense Advanced Research Projects Agency, is best known for creating the Internet. It is sponsors a global competition to develop a driverless vehicle that can navigate complex terrain without human intervention.

The first Darpa Grand Challenge was held in March 2004 and featured a 228km desert course. Fifteen robot cars attempted the course but none finished. The next year, four vehicles completed a 212km route in less than the required 10-hour limit and Darpa awarded a $2 million prize to the winning team from Stanford University.

The latest CSIR newsletter says the South African team plans to automate two mini Baja Bugs for the 2010 challenge.

"The team will equip the vehicle with various sensors and positioning systems to determine all the characteristics of its environment to enable it to carry out the task it has been assigned."

The newsletter says the aim of SA's participation is to generate visibility for both the CSIR and the University of KwaZulu-Natal "to attract top students and researchers-engineers in this scarce skills environment".

"This project, dubbed 'Renoster', is ideal to expose younger engineers to collaboration with peers in different disciplines. At the same time, it creates the opportunity for natural and productive mentoring," says Riaan Coetzee, the CSIR's mechatronics and micro-manufacturing research manager.

"We do not even know at this stage what format the Darpa challenge will have next time," says Coetzee, "but it is an exciting opportunity and challenge for the team. It proves that research and development can be fun."

The university's main focus will be to optimise GPS strategies for accurate calculation of the Baja Bugs' location. The university has an established programme for final-year students to semi-automate a buggy. This will be expanded through involvement in the project.

The university can use this practical project for final-year and postgraduate students, while the mechatronics and micro-manufacturing research group further develops skills in autonomous platforms.

The first focus of the group will be to get the vehicle operational. From there, the project will focus on analysis of the data before various sensors and actuators are installed. The design and implementation of software to control the buggy autonomously will then be added. The latter is the most complex area and poses a major challenge, Coetzee says.