We don’t need to wait for fully autonomous vehicles to benefit from autonomous driving technologies. Life-saving technologies like this one will be preventing accidents in the future even in situations that require extreme maneuvering skills.
You’re driving down the highway when suddenly a box falls from the truck in front of you. You know that you need to react immediately. There is no room for hesitation, but also no room for mistakes. Every millisecond counts. Your pulse quickens. Your heart races. Can you handle this situation safely?
Sometimes we can, sometimes we can’t. Many factors such as speed, distance to the obstacle, as well as our attention and skill level determine whether we have enough reaction time to make this maneuver safely.
For those situations when we don’t, Tulga Ersal, along with his co-principal investigator Jeffrey Stein, are trying to help.
A research scientist in mechanical engineering, Ersal partnered with Toyota Research Institute (TRI) on a project that helps your car automatically steer clear of a collision even in high speed situations.
This project is a companion to the Toyota Guardian(TM) automated driving development project that is ongoing at TRI. UM’s research helps to inform and validate Toyota Guardian advancements, which TRI has demonstrated on closed test courses.
Ersal and his colleagues at U-M and TRI are now pushing vehicles to their limits in their simulations to identify how to make this automated collision avoidance idea work at the dynamic limits of the vehicle. Therein lies our highway scenario.
“This technology comes into play when you can’t avoid an obstacle by braking alone. It continuously assesses whether you are at the point of no return, and if you are, it kicks in to perform an extreme lane change for you to keep you safe,” Ersal explains.
“It knows what the vehicle can and can’t do, and on the fly calculates based on the given situation what the safest emergency maneuver is and how to execute it.”
Ersal’s foundation in Mechanical Engineering (his Masters and Ph.D. are both from U-M) and expertise in system dynamics and control gave him the necessary foundation to take on this challenge together with TRI researchers.
They work collaboratively on this project. The U-M team has developed an initial version of the algorithm and are now transitioning it to TRI for enhanced testing in simulation first, and on a real vehicle next. These evaluation results will help the U-M team better understand the limitations of the algorithm under more realistic scenarios and increase its robustness.
“We’ve established a strong collaboration with Dr. Ersal and Dr. Stein, and their team,” said Vishnu Desaraju, researcher at TRI and project liaison. “I think it’s a great example of how industry and academia can work together on state-of-the-art safety technology that can save lives.”
U-M and TRI have a deep partnership, with over 45 collaborative research projects. Ersal’s is one of these, and this research project represents one of many research aspects involving the collision technology alone.
Ersal notes that the problem space U-M is in with this project fits nicely within the framework TRI is planning, noting that “there are a lot of research questions to solve, such as formulating an algorithm that can solve the problem, and solve it efficiently and reliably, while working with small reaction times.”
On a holistic level, these technological advancements benefit from collaborative partnerships like this one: “This is always an iterative process. We learn from the actual world to identify the research problems. We then create an abstraction of the problems in computer and try to figure out how our algorithm can deal with them. When we think the algorithm is mature enough, we go back to the real world again and the process repeats,” Ersal notes, glancing at his computer, and you can imagine the simulations he’s envisioning for tomorrow.