Making self-driving cars a reality: Kettering University becomes driving force in AutoDrive

My graduate student, Shobit Sharma was featured in this article http://www.flintside.com/features/engineering-the-future-of-the-road.aspx. The original article is archived here for a future reference.

Making self-driving cars a reality: Kettering University becomes driving force in AutoDrive

FLINT, Michigan—They are the driving force behind cars that, well, drive themselves.

Students and faculty from Kettering University in Flint are one of just eight universities in North America selected to test the limits of the autonomous (self-driving) Chevrolet Bolt in the deserts of Yuma, Arizona with a chance to win the exclusive Society of Automotive Engineering-General Motors AutoDrive Competition.

“It’s stuff that we’ve always heard about, but now we’re actually working on it and able to do it,” graduate student Alex Rath, 22, of Fenton says. Rath, one of two student coordinators, has been working on the project since the beginning, helping create the initial proposal.

In October of 2016, Kettering, one of approximately 50 schools, was contacted about the competition to gauge if there was any interest in competing. By February 1, 2017, Kettering submitted their 25-page proposal document outlining the team, who would be on the team, the facilities of the university, how they would approach automating the vehicle ,and how they would do it safely.

At the end of September, Kettering’s team, and teams at seven other universities, received their brand new Chevrolet Bolt along with a LIDAR (Light Detection and Ranging) sensor, other various sensors, an Intel Crystal Rugged Server and a fixed amount of money to spend on the project, including funding for travel for certain team members, and transit of the vehicle. All decks are equally stacked, so the ingenuity and application of the students will be what sets the competitors apart.

This isn’t Rath’s first rodeo. During his undergraduate studies, he participated in the SAE Clean Snowmobile Challenge. Like his previous challenge, this challenge is very much about project and time management. In the first year alone, there are four major challenges.

The challenges, both dynamic and static, address everyday driving occurrences. Within the first year, teams must prepare for a stop sign challenge, where the vehicle needs to recognize a stop sign and safely stop, a lane following challenge on a technical driving course containing curves, waves and grades, an object avoidance challenge, avoiding static items in the vehicle’s path and a mapping challenge where the vehicle must have a graphical user interface that the user can select a destination. In addition to those four dynamic challenges are the safety and social responsibility presentations. These presentations will highlight the safety of the vehicle and address ethical questions surrounding autonomous vehicles.

“There’s a lot of things that we do (when driving) that we don’t even think about,” says Diane Peters, 46, an assistant professor and the primary faculty advisor for the project. How quickly is the intersection approaching and do we need to stop? Is that a stop sign with slush on it, or is it a different sign? There’s an object in the road, how should I avoid it?

“If there’s a stop sign ahead now, we have to decide that we’re going to stop. How much do we apply the brakes?” asks Peters of Flint. The questions, in our minds, have simple answers. However, translating that input from a sensor, to a computer and then to the mechanical components of a vehicle is a different beast. For every question, there are more behind it. For every answer to that question, there lie even more questions. This is the task of the team at Kettering. To answer these potential questions in a language an automated vehicle will understand.

To address all of these challenges and questions, the team at Kettering has just more than 60 students on the multi-disciplinary team as well as faculty for advice and guidance. Comprised of 50 undergrad students and 12 graduate students, the team is divided into three sub-teams: the sensors team, the controls and dynamics group and the safety team. Students and faculty come from backgrounds and areas of study such as mechanical and electrical engineering, computer science, and even industrial and manufacturing engineering.

The sensors team will deal with exactly that – sensors. They will decide which sensors to use, where to use them, how to efficiently mount them to the vehicle, and how to make them communicate with the vehicle. The controls and dynamics group is responsible for taking all of the data from the sensors, translating it and sending it back to the vehicle to make “decisions” and translating it into mechanical actions. The safety team focuses on “how not to kill your teammates,” according to Peters, as well as functional safety of the vehicle, such as stopping with critical failures.

Shobit Sharma, 26, came to Kettering from India after college and working in IT. There, he dealt with streams of data, but his heart was set on automation in the automotive industry—which brought him all the way to Kettering.

Another student coordinator, Sharma is also one of five pilots for the competition. The pilots were required to go to the GM Proving Grounds in Milford to demonstrate their ability to operate the vehicle safely in the event of a failure. Faults were triggered on the course and the pilots were required to safely bring the vehicle to a stop.

Failures are an inevitable aspect of the competition. Failures can mean various things including the vehicle going too fast, outside of specified parameters, moving outside of a designated lane or stopping too quickly. The “trick” is to program the vehicle to deal with the parameter failures with appropriate counteractions. Reactions that we often unconsciously perform. Going too fast? Let off of the accelerator. Drifting out of the lane? Adjust your steering.

“The ideal driver is actually pretty good,” laughs Peters, “but we aren’t always ideal drivers.” Creating an ideal driver through sensors and programming is a rather large task—and one that will be constantly evolving. Technology is advancing faster and faster as time progresses, and the team is uncertain where the competition will head after the first year.

“Plotting out what is going to happen three years from now is a guessing game,” says Peters, seemingly undaunted by the profound tasks that she, other faculty advisors and students face. “If we come in first, which I would love to see happen, our university would be absolutely thrilled.”

“You get a first place trophy and a lot of bragging rights,” smirks Rath, also seemingly undaunted. “Once you know where you’re going, it’s easier to figure out how to implement it,” he says.

And, rest assured, the Kettering team knows exactly where they’re going. To the Yuma GM Proving Grounds

Leave a Reply

Your email address will not be published. Required fields are marked *