2007 Urban DARPA Grand Challenge

Sydney-Berkeley Driving Team logo

Short Description

The Sydney-Berkeley Driving Team is made up of researchers from:

  1. University of Sydney’s Australian Centre for Field Robotics (ACFR)
  2. University of Technology Sydney's (UTS) Mechatronics and Intelligent Systems Group
  3. National ICT Australia (NICTA), and
  4. University of California Berkeley - Electrical Engineering and Computer Sciences (EECS).

The team represents a unique collaboration in the area of field robotics and control. The ACFR, UTS, and NICTA team members have world-renowned experience with sensor fusion for autonomous vehicles at the theoretical and practical levels. UC Berkeley members have real-time avionics experience from the Berkeley Aerial Robots group, and members with experience from the two previous DARPA Grand Challenges.


Rio Tinto
PMD Technologies



ACFR Urbran Grand Challenge Team

Vehicle lead: Michael Moser
Electronics: Ashod Donikian
Mechanical: Iain Brown

Software lead: Ben Upcroft
System Architecture: Alex Makarenko
Navigation: Michael Moser
Sensors and Perception: Alen Alempijevic, Ashod Donikian, Jose Guivant, Roman Katz, Anindha Parthy, David Johnson, Bertrand Douillard, Todd Templeton
Control: Michael Moser
Path planning: Alex Brooks, Ben Upcroft, Robert Fitch, Jonathan Sprinkle, Humberto Gonzales, Esten Gotli
High level reasoning: Robert Fitch, Joel Veness, Will Uther

Eduardo Nebot, Hugh Durrant-Whyte, Shankar Shastry

Communication and Corporate Relations
Olga Sawtell

No. 5 - Toyota Rav4

No. 5 - Toyota Rav4

Technical Approach

Building reliable robotic hardware is a difficult task in itself, but the major challenge of this competition is in software. The main difficulties arise from the task’s complexity and its dynamic real-time nature. Other contributing factors include the distributed and cross-platform computing environment, the large number of software contributors, and the need to use existing code.

Our software is built using a Component-Based Software Engineering (CBSE) approach. This offers modularity, software reuse, and flexibility in deployment, all of which are necessary to address the problems listed above. Applied to a robotic application, CBSE means that algorithms are mapped to a set of components. Components run asynchronously and exchange information through communication.

The on-board computer system currently has four hosts (not counting diagnostic laptops that are often connected to the system), and this number is likely to grow. The on-board computers use two operating systems: Ubuntu Linux and QNX. The software is written by about a dozen people from four organizations (this number is higher if we count the authors of the existing components used directly in our system).

The computing hardware uses off-the-shelf rack-mounted PCs with Intel dual-core processors. The hosts are connected with a standard 1-Gigabit Ethernet hub.

Some but not all parts of our system require real-time features. For example, there is a strong need for accurate time-stamping of sensor data. A vehicle in the competition moves at speeds of up to 30mph (48km/h) and small sporadic delays in the standard Linux kernel can have significant negative impact, particularly in navigation. We use a dedicated host running the real-time QNX Neutrino operating system for all interactions with sensor and actuator hardware.

Among many challenges presented by the DARPA competition, software complexity is one of the most difficult. (This statement may also be true for the field of mobile robotics as a whole.) The component-based approach helps manage this complexity by breaking up a monolithic implementation into manageable parts.


  • Group Leader - Sydney
    Ben Upcroft
    Phone: +612 9036 7057
  • Group Leader - Berkely
    Jonathan Sprinkle
    Phone: (510) 642-0348
  • Business Consultant:
    Olga Sawtell (Sydney)
    Phone: +61 2 9351 7690
  • Public Relations:
    Barbara Mueller Blackford (Berkeley)
    Director of Corporate and Foundation Relations
    College of Engineering, UC Berkeley
    209 McLaughlin Hall, MC 1722
    Berkeley, CA 94720-1722
    Phone: (510) 643-9289


2007 Urban DARPA Grand Challenge team


June 2007 - Engineers Sydney - Robot car contest (2MB pdf)


Toyota RAV4 driven by 4 PCs


8 May 2007 - The Australian newspaper - Bot contest for ruler of roads


April, 2007 An Autonomous Vehicle Using Ice and Orca an article in
In ZeroC's Connections
newsletter, no. 22, 2007, pp. 2-6

More information

  • Feb 2006 - DARPA announce the next Grand Challenge VIEW
  • Visit the Wiki.