News & Events
ECE capstone team earns first in PEO/IET Paper Competition
Fourth-year Electrical and Computer Engineering students Nick Van Nynatten, Ryan Dowling, Jordan Nanos and Ryan Dick earned first place in the Student Paper Competition, sponsored by PEO (Kingston) and IET, March 22. The honour came as the team also earned top spot in the ELEC490 capstone design project competition.
“We felt pretty good after our performance at the capstone open-house demo day, so we thought we’d enter the PEO paper competition as well,” says Van Nynatten. “We also thought the PEO presentation would be good practice going into our final project demonstration.”
They won for their foundational design for a bin picking system that could, with some more refinement and development, be suitable for competition in the Amazon Picking Challenge.
“The bin picking challenge was part of our focus from the start, but we realized it’s not really feasible for us to enter it competitively after only six months of work,” says Dowling. “Still, we wanted to create a base for future capstone teams to be able to enter that competition with a robot that’s potentially world class.”
The challenge is to overcome a set of problems robotics engineers have been wrestling with for decades, one on the very frontier of computer vision, robotics and machine learning. How can you make a system that can best identify, sort, store and retrieve random objects in a warehouse with as little human input as possible? Students from all over the globe make large investments in time, money, and effort to compete for bragging rights and the chance to earn thousands in prize money.
“The team worked very independently” says faculty supervisor Michael Greenspan. “I described the problem to them, and pointed them in the direction of some solution alternatives. After that, all design decisions and implementation were theirs.”
Though bin picking problems are mostly about software now, one of the first hurdles for Van Nynatten, Dowling, Nanos and Dick was pulling the mechanical components together. They built a vacuum-lifter to manipulate their test objects, added a camera for vision, and repurposed a gantry rig from Greenspan’s lab to give their robot reach. After that, the hard part, the part for which they really earned their honours, was the software.
“This time last year when we were picking our project, the gantry was in pieces,” says Nanos. “Since then, with the help of the faculty, it’s now set up and we have this modular code base that other people can build on to continue to move the project forward, rather than having to go through the initial setup that we did .”
“There were six finalists in the competition, four from RMC and two from Queen’s,” says Queen’s engineering professor and one of three judges, Brian Surgenor. “They were evaluated on the quality of their presentation as well as the scope and novelty of the project itself. All gave excellent presentations but the team of Dowling, Van Nynantten, Nanos and Dick were given the edge with their demonstration of the capabilities of their bin picking robot.”