bayburt escort bilecik escort aydın escort osmaniye escort sivas escort erzurum escort ordu escort balıkesir escort gümüşhane escort ısparta escort giresun escort manisa escort aksaray escort çanakkale escort tokat escort
The future of testing self-driving cars

The future of testing self-driving cars

The future of testing self-driving cars

Photo courtesy of Evan Krape, Courtesy of IDS Lab

UD Professor Andreas Malikopoulos’ ‘Scaled Smart City’ bridges the gap between driving simulation and real-world testing

On the second floor of the University of Delaware’s Spencer Lab, a 20-by-20-foot model city — complete with tree-lined streets and brightly colored shops and homes — is being researched by researchers at the Information Decision Sciences (IDS) Institute. A place to test innovation. For self-driving cars within the safe limits of controlled settings.

All of the work in designing, developing, and testing new algorithms will take place in this Scaled Smart City run by Andreas Malikopoulos, Terri Connor Kelly, and John Kelly Associate Professors of Career Development in the Department of Mechanical Engineering. Now, with new publications and research grants in hand, the group is poised to continue its work to enable AI-piloted vehicles to safely share roads with human drivers in the future. I’m here.

After joining UD in 2017, Malikopoulos built scaled smart cities to bridge the gap between driving simulation and real-world testing. He notes that full virtual simulations do not accurately account for all errors that occur in the real world, and that his level of testing on the streets with full-scale vehicles can be dangerous without proper safety precautions. I explained that there is a possibility.

“The problem with simulation is that it is destined for success,” says Malikopoulos. “When I run the simulation, everything runs perfectly. For example, there are no misunderstandings between vehicles. The advantage of our scaled smart city is that we can see and address any shortcomings of the algorithm before it is actually tested on a vehicle.”

One of the biggest challenges facing researchers working in the field of Connected Self-Driving Vehicles (CAVs) is that while the CAVs integrate all the information they need to drive safely, they also require supercomputers. Malikopoulos said the goal is not to have complex algorithms. Run.

“How we use information from vehicles to coordinate with each other to avoid stop-and-go driving and learn what to do when there are points of contention such as intersections, roundabouts or construction areas. I’m working on it,” he said. “To achieve this, we use control theory to develop algorithms that can efficiently handle these challenges.”

One component of the IDS research program is led by PhD student Heeseung Bang, a project focused on developing a CAV control framework that combines routing, coordination and control. His approach involves combining advanced machine learning algorithms that help CAV predict traffic conditions and understand how to navigate the road, such as which road to choose at a given intersection and what to do. It involves solving decision-making problems at various levels that his CAV faces. Deal with complex human behavior.

“By solving small problems and effectively combining those results, we are solving huge optimization problems,” says Bang. “For example, CAVs can be coordinated to cross intersections without stop-and-go driving, and that information can be used to predict traffic congestion at intersections and assign the fastest route to each vehicle. ”

Infrastructure for testing new algorithms

The group recently published an article on how scaled smart cities can be used to test and validate new CAV control algorithms before taking them into the real world. Decorated the cover of the December 2022 issue IEEE control systemthe article also discusses specific case studies that demonstrate how that algorithm helps CAVs navigate certain traffic scenarios safely, such as roundabouts, intersections, and confluence roads.

“We have shown how testbeds can help prove new concepts for new mobility systems while understanding the gap between theory and real-world implementations (Honda Research Institute), AM Ishtique Mahbub, a controls engineer at Aptiv, and Logan Beaver, a postdoc at Boston University, are all graduates of the IDS Lab.

“At the forefront of education, we need to scale the training and education of graduate students by exposing them to a balanced combination of theory and practice, integrating research findings into existing courses, and engaging undergraduates in research. We reported in an article how Do Smart City is being used, creating interactive educational demos and reaching out to K-12 students,” says Malikopoulos.

The group recently won the IEEE 2nd Annual International Conference on Digital Twins and Parallel Intelligence for the ‘digital twin’ of Scaled Smart City, a fully virtualized version of the setup that can be accessed remotely by collaborators. Received a paper award. .

Created at the height of the COVID-19 pandemic, the digital twin will allow researchers to develop and test new algorithms in a fully virtualized setting before running them in a scaled smart city, said Malikopoulos. I can.

“Once the algorithm runs smoothly on the digital twin, it is easy to implement in physical testing,” he said.

Transport equity and human-CAV interaction

The IDS Lab will expand its research in this area thanks to two National Science Foundation grants. One examines the fairness of the transportation system and the other studies the interaction between humans and her CAV.

For the Transportation Equity project, the lab is working with collaborators at Boston University and MIT to combine data on user preferences, energy usage, and travel times to suggest travel incentives, such as train ticket discounts that lead to transportation. I am working on a $1.2 million project to Systems that deliver broader economic, environmental and social benefits. The goal is to “learn and train from travelers,” said Malikopoulos, who said the experiment was scaled up in a smart city, followed by a larger experiment in the city of Boston.

“The goal of this project is to promote incentives to ensure fairness in transportation and ensure that all travelers have an equal opportunity to accessibility,” he said. “But equities are very abstract, and the challenge we are working on is how to integrate equities into a mathematical framework. It’s really rewarding to see how it can help us deal with this.”

To study human-CAV interactions, the lab was awarded a grant of approximately $500,000 to combine data on human driving behavior with control theory and reinforcement learning approaches. This allows the CAV to “learn” how to respond appropriately to different possible scenarios. Driving patterns vary greatly from person to person, from overly cautious or uncertain drivers to more aggressive drivers.

“If there is a human in the loop, the problem becomes very difficult because of unexpected behavior. ,” Malikopoulos said. “We are very excited about this project because this is a significant problem in the field at the moment: a way to allow humans to interact with self-driving cars in an efficient and safe way.”

Visualize the future of smart cities

Fleets of self-driving cars won’t hit the road overnight, but Malikopoulos believes that by developing the algorithms these complex systems need to operate safely and effectively, researchers will be able to become “smart.” The time has come to start visualizing the future of the City.

“We have made a lot of progress and these technologies are mature enough that we will soon see them deployed in controlled environments,” said Malikopoulos. “That said, there’s still a long way to go. Everything is happening at a fast pace, so it’s important to follow a rigorous and sound investigative procedure. If we can do that, I think we’ll be on our way to the finish line.”

Interested in continuing CAV research after graduation, Bang enjoys working on something that provides a “sketch of the future transportation system” and a glimpse of what the transportation corridor might soon look like. I’m in.

“By working in this field, it’s exciting to see and predict how this technology will affect the future,” he said. It would be very rewarding if I could do something like that.”

Leave a Reply

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

%d bloggers like this: