The Droid Racing Challenge 2017 took place on the 4th of July with teams from Australia and New Zealand competing. The event was a huge success with teams improving vastly over last year, the number of competing teams doubling in size and even receiving coverage from ABC News
The teams initially faced off against one another in a series of round robin matches over three rounds. With points scored for each section of the track completed and winning the race and penalties for leaving the track or hitting an obstacle. The top 8 teams then progressed to the finals where the winner was declared by a selection of judges, who weighed the droids speed and progress along the track against the amount of restarts and obstacle collisions.
The day started off slow with teams experiencing serious issues detecting the lines during round 1. However from there progress was quick, with Macquarie University soon completing the first lap of the day, followed by University of Queensland team 1 and University of New South Wales team 2. Obstacle avoidance also improved rapidly throughout the day with teams adjusting and calibrating for the unpredictable outdoor conditions.
After a demonstration provided by sponsors MathWorks on ways of using MATLAB to control the droids (click here for more). And lunch provided by sponsors Guzman y Gomez the following teams progressed to the finals
- Griffith University
- Macquarie University
- Queensland University of Technology Team 3
- University of New South Wales Team 1 & 2
- University of Queensland Team 1 & 2
- University of Wollongong
In the end University of New South Wales team 2 won the day completing the track quickly and also demonstrating fantastic obstacle avoidance. University of Queensland team 1 came in second also showing off great obstacle avoidance followed by the team from the University of Macquarie in 3rd place.
Teams were also judged by there creativity and the aesthetics of the droids. For this the Best Design award went to Queensland University of Technology team 1 for their droid design. Honourable mentions went to the team from University of Wollongong which used 3 raspberry pi cameras to detect the lines and objects (and some sunnies to get rid of glare). As well as to QUT team 2 for its use of machine learning via a neural network opposed to traditional computer vision programming methods.
The competition this year was a massive success with several teams completing entire laps of the course and integrating obstacle avoidance into the code. This was a huge improvement on last year from where only one team completed a full lap. DRC 2017 also saw 4 new universities present teams doubling the size of the competition, including our first international team from the University of Auckland.
This all would not have been possible without the many amazing people and organisations who generously gave their time and sponsorship to this event. The competing teams, QUT staff and students, volunteers on the day, and our sponsors Boeing, MathWorks, QUT, the Australian Centre for Robotic Vision, and Guzman y Gomez all deserve a giant thank you.
Also a huge thank you to the Club President Filippo Capurso who lead the executive team and pulled the event together.
QUT Robotics Club Secretary