RIT's Very Own Autonomous People Mover
by Ryan Black | published May. 18th, 2016
With the likes of Google, Tesla and a myriad of other tech giants now seemingly more devoted than ever to the prospect of self driving cars, a group of engineering students right here at RIT have been hard at work on an autonomous vehicle of their own. Over the last two semesters, a group led by project manager Joe Hudden, a fifth year Industrial Engineering, and overseen by Computer Engineering professor Dr. Ray Ptucha, have been hard at work on the "RIT Autonomous People Mover" — a golf cart capable of driving itself. If all goes according to plan, their intentions are to to demonstrate their work at this year's Imagine RIT festival.
They aim to show off the vehicle's capabilities in a pre-programmed closed course,complete with obstacles,within the circle at the end of Lomb Memorial Drive, right next to The Sentinel.
"The cart will be able to drive people without any human intervention from point A to point B," according to Cody Smith, a fifth year Computer Engineering student working on the project. Ptucha emphasized how prudently the team will be handling the presentation."We will have a person on board the golf cart just for safety measures," he said."If everything goes wrong or something [like that], they could always manually hit the brake or hit the E-stop to kill the car at any point in time."
The capstone project brings together a diverse collection of students — including the likes of Electrical, Mechanical, Industrial and Computer Engineering majors — to collaborate and utilize each others' respective knowledge and skills.
"It's been a really big learning experience being able to work on a team of people where they have expertise you don't necessarily have and you have expertise they don't necessarily have," noted Smith. For him, experiencing how a lengthy and complex venture like the "People Mover" is split up into "lots of small parts, that individual people can contribute to," has been a valuable learning process.
Hudden, who handles the aforementioned organization, expressed a similar sense of appreciation for theteam's dynamic."The makeup of [this] team and how different it is really helped [us] produce the cart that it is today," Hudden said. He also explained how everyone's respective background helped with a different system or element of the cart, as well as bringing it all together.
Through such collaboration, the team has been able to design both hardware and software that works synchronously. Specialized computer algorithms interpret what the cameras on the cart see with corresponding 360-degree photos of campus from Google Streetview.In doing so, the "People Mover" can determine what area of the campus it is actually in. After utilizing particle filters, the cart then knows its exact position.This "Computer Vision Method" is used in conjunction with both high- and low-precision GPS,the latter of which is comparable to what's on your phone.
All of this is made possible by advances in machine learning, known as "deep learning." Almost every recent improvement inspeech recognition software in smart phones as well as other devices and applications is a direct result of this. As The Economist pointed out in May of last year, this sort of technology typically used to be able to only interpret large amounts of data as one layer of analysis. Yet within the past few years, it is has become capable of several "distinct, hierarchical [and interlinked] layers." Imagine jumping from only "dozens or hundreds of neurons" to a level more equivalent with the human brain.
"It's pretty awesome," admitted Hudden when asked about working with cutting edge technology like deep learning. "I think that's what RIT is all about: innovation."
With an endeavor like the "People Mover" implementing such advancements, Ptucha sees the project as, "a springboardto doadvanced research," as well as a showcase of the technological prowess and accomplishments of RIT and its students.
With research and testing for automated vehicles being so sought after at the moment, Ptucha also noted how RIT could be well positioned to attract investment from private companies. Just last year the University of Michigan opened a 32 acre test facility for autonomous vehicles named "Mcity," and in November Ford announced plans to use it to test their own self-driven vehicles. They are not alone; Toyota has funded $25 million worth of research for both Stanford University and the Massachusetts Institute of Technology respectively in 2015.
Looking past this year's Imagine RIT, the plan is for a whole new team of engineering students to pick up the torch come fall 2016. Hudden definitely wants what they've accomplished to be a "platform" for that group, something which they can build upon. At next year's Imagine RIT, Ptucha hopes to demonstrate a golf cart capable of maneuvering "a fixed" route along one of the driving paths, but for now he just wants to see how this year goes.
One day though, when the technology has proliferated and been perfected, he does hope to see a fleet of autonomous golf carts taxiing people around campus. Both students echo that belief, suggesting how in the future such a golf cart could be used to ferry visitors from parking lots, give automated tours of campus or even take people to and from class. Until then, one can see the prototype in action at Imagine RIT this year, as well as on our website – reporter.rit.edu.