(L to R) Ohad Rusnak and Omri Elmalech with their tracking shopping cart.
(photo credit: SHARON TZUR, TECHNION)
Future trips to the supermarket could be a lot easier if a robotic cart developed by students at Technion-Israel Institute of Technology takes off.
Omri Elmalech and Ohad Rusnak have developed an algorithm that makes it possible for you to go through the store with your shopping cart trailing you through the aisles.
Cousins and childhood friends who were fascinated with building things and understanding how they work, Elmalech and Rusnak played a lot with Lego and toy robots when they were young.
“We always knew that when we grew up we’d study engineering at the Technion,” Elmalech said this week.
“When we got to the stage where we were supposed to choose a final project, it was obvious to us that we’d choose something that had to do with robotics, and this is what led us to the faculty’s control robotics and machine learning lab,” added Rusnak.
The idea for this project came from a shopping experience Rusnak had with his mother.
“One of my mom’s favorite pastimes is grocery shopping. She loves to load up the shopping cart. When I go with her, I get tired of following her, pushing the cart up and down the aisles,” he said. “So we had an idea of developing a robotic shopping cart – one that is autonomous and can replace me in getting the job done [of following her].”
“Ohad’s mother loved the idea, and we started working on it seriously,” added Elmalech. “We started to search the Internet for similar projects and were surprised not to find autonomous carts.
We saw some shopping carts capable of tracking a shopper using remote sensor, but none that were capable of tracking a shopper autonomously.”
Although several faculty members felt the idea was a bit too ambitious for an undergraduate project, since it involves real-time image processing, the two were determined.
On the top of the cart, the two installed a 3D Kinect camera capable of in-depth images that can recognize people in the picture. The camera then sends the information to a processor that runs the identification algorithm and ensures that of all the people in the picture, it will track only the one individual it’s supposed to follow.
“It’s a system that must continually learn,” Elmalech explained, “and the processor stores new information and improves its performance continuously.
Developing the algorithm was the major and most significant challenge in this project.”
They built the cart.
“We found an old, dismantled shopping cart in a garbage dump.... We rebuilt it, welding its structure anew so it would be effective for our purposes. We installed the Kinect camera, a powerful battery and PC card, while adhering to the product design,” Rusnak said.
Kobi Kohai, the lab’s chief engineer, mentored the students throughout the project.
“The project received a grade of 100,” Kohai said. “From the faculty’s point of view, the process that students undertake in order to build the device by themselves is the most significant part of their professional experience, and not necessarily the final product.
During the course, students are required to investigate new technologies and are exposed to entrepreneurship and product development.”
The two Technion students agreed that their project was successful, but said it remained a work in progress.