by James Lam | Interviewers: James Lam, Vaibhav Bahadur, Donald Lam | Photo Credits: Donald Lam | Also Contributing: Tommy Hu | 10 January 2013
After securing the 9th place in the DARPA Virtual Robotics Challenge back in June last year, Team HKU has purchased her own Atlas robot and finished 12th in the Miami trials 20 – 21 December 2013.
The software engineering behind the robot is everything. Amongst the team of coders, five were actually undergraduate students.
We sat down with Chim Lee, a year 2 Electrical Engineering local student in the 3-year curriculum, and Rita Liu, a year 2 student from Guangzhou studying Computer Science in the 4-year curriculum, who together with Jacky Yu, Jackie Chan and Tommy Hu, are the undergraduates who helped program the HKU Atlas.
Getting On Board
They had no prior robot programming experiences in their secondary schools. They were invited to join the Atlas team back in March 2013 by Professor Robert C. Roberts, who they had previously consulted when they encountered a postgraduate level question.
“We are both in the M2 robotics team,” said Chim. M2 is the HKU student robotics team. It was a problem that they ran into while programming robots for the M2 team.
To solve the problem, they soon started their own research with their M2 teammates, focusing on the filtering algorithm: the algorithm that removes noise and enable more accurate system control for the robot to move.
And Professor Roberts, impressed with their work and their sound programming knowledge, put their names down on the recruitment shortlist.
Fun with Leap Motion
Here’s a video clip of Atlas being the sure-win robot in paper scissor stone. The secret lies with the camera sensors on the robot’s hand. After detecting your hand gesture, the robot will immediately pose its hand to the hand sign that beats yours. Technically, it is cheating because it reads your hand sign before posing its, but that’s in milliseconds, so you don’t even notice it.
Learn As They Go
The hand of Atlas, developed by Sandia National Labs, has 28 degrees of freedom, with many tactile sensors all over its fingers and palm, including two cameras.
This became their full time jobs in the latter half of the last summer vacation. They learnt about the complicated interface of the open source Robot Operating System (ROS), and co-wrote many parts in the controlling of the robot’s right hand.
“It was a bit of C++, a bit of python, a bit of everything. You can use a lot of languages in ROS.”
As always, programs freshly written for a robot, especially one as expensive as the Boston Dynamics Atlas, are first tested on a simulation program before they are run on the robot to avoid damage.
Their work was fun and there was little pressure. They still went back to the laboratory regularly to help out when school started. They both agree that it has been a precious and remarkable experience, working with postgraduate students and professors.