EV
Making self-driving cars safer, less accident prone
New AI model could enhance self-driving car safety
University of Georgia
Self-driving cars rely on artificial intelligence to predict where nearby cars will go. But when those predictions don’t match reality, that discrepancy can potentially lead to crashes and less safe roadways.
That’s why a recent study from the University of Georgia developed a new AI model to make self-driving cars safer.
This study introduces an AI model for self-driving cars, designed to predict the movement of nearby traffic and incorporate innovative features for planning safe vehicle movements.
"The planned trajectory of the self-driving car may turn out to collide with the actual trajectory of another vehicle.” —Qianwen Li, College of Engineering
The study used data from the I-75 freeway in Florida to predict other cars’ paths and determine the motion of the self-driving car when following another vehicle.
Previous research mostly predicts surrounding traffic movements and then plans a self-driving car’s motion. This separate approach, however, makes crashes and near-misses more likely.
“That’s why we wanted to consolidate those two steps — to make the autonomous vehicle operation safer,” said Qianwen Li, lead author of the study and an assistant professor in UGA’s College of Engineering. “And as illustrated by our experiments, that approach does help with safety performance.”
AI needs to do more than predict traffic
To keep drivers safe, self-driving cars have to be able to accurately anticipate the movements of surrounding traffic. However, it’s difficult to know what other drivers will do on the road.
“There are always differences between your prediction and the reality,” said Li. “The planned trajectory of the self-driving car may turn out to collide with the actual trajectory of another vehicle.”
The new model was designed to take prediction errors into account, as eliminating them isn’t possible.
Li’s group is also working on developing more complex AI models for self-driving car operations, such as large learning models like ChatGPT. Traffic scenarios could be fed to these models, and they would determine the best course of action.
However, large language models have limits. While they’re effective at making high-level decisions related to how to respond to different situations, planning the movements of a car isn’t what they’re built for.
“How do we make a perfect lane change that is safe and also efficient?” said Li. “How do we come to a smooth stop for pedestrians without inducing any riding discomfort? Basically, how do we design the specific trajectories? That part we do not ask ChatGPT or large language models to do because they do not have the capability to do so. Traditional trajectory optimization models can do a much better job based on our experiments so far.”
Balancing safety and mobility in vehicle artificial intelligence
Designing AI for self-driving cars is a balancing act. Maximizing safety often comes at the cost of mobility.
If a self-driving car is taught to drive as safely as possible, for example, it will stay far behind the car in front of it. While a safer option, that distance would likely reduce the number of cars that could fit on the roadway at a given time.
Similarly, focusing too much on mobility could result in cars driving too aggressively, increasing the risk of crashes.
“We’re still working on how we train the model in a way that can balance the safety and mobility performance,” said Li.
The study was published in Transportation Research Part E. Co-authors include Handong Yao of UGA’s College of Engineering, Xiaopeng Li of University of Wisconsin-Madison’s Department of Civil and Environmental Engineering, and Chenyang Yu of McGill University’s Mathematics and Computer Science Department.
Journal
Transportation Research
Article Title
Safety aware neural network for connected and automated vehicle operations
Article Publication Date
10-Dec-2024
Nationwide study looks at when and where EV owners use public charging stations
Initial results show that demand during working hours is high in California
AUSTIN, TX, Dec 10, 2024 – Electric Vehicles (EVs) represent a promising mode of transportation that can help the United States reduce its carbon emissions. But barriers such as the high cost of installing and using EV Charging Stations (EVCS), their limitations in supplying emerging demand, and their uneven distribution throughout the country limit access for many Americans.
Researchers at the University of Maryland are using supercomputers and machine learning methods to analyze a full year of real-time data collected from individual EV charging ports at more than 50,000 publicly available stations throughout the country. The primary focus of the study is to estimate demand and peak times at EV charging stations.
“Understanding EV users’ charging behaviors will provide valuable insights into their needs, enabling us to efficiently deploy EV charging infrastructure to meet the emerging demand,” says civil engineering Ph.D. candidate Safoura Safari. She will present the team’s preliminary findings in December at the annual meeting of the Society for Risk Analysis in Austin, Texas.
Led by scientists in UMD’s Center for Global Sustainability and Center for Disaster Resilience, the study utilizes individual charging port statuses recorded every 10 minutes at 54,000 charging stations located in three power grid zones: California, Texas, and the Northeast. It is the first study of its kind to use real-time charging port data to understand charging behaviors across different time intervals.
Machine learning is used to extract “clusters” or groups of behavior from the data. In their analysis, the researchers look for variations of charging patterns on daily, monthly, and seasonal scales — accounting for temperature impacts on battery range (which impacts charging decisions).
The first round of analysis based on data for the month of August 2023, revealed working-hour patterns in all three zones, with differences in charging behavior and power demand:
- In the California zone, about 45 percent of stations showed a working-hours charging pattern (6 am – 6 pm), with peak utilization reaching 80 percent of total stations’ charging capacity during those hours. This suggests that charging stations in this zone may be especially vulnerable in the event of a disaster or disruption of power — with the potential for lines of vehicles to form at public charging stations.
- In the Texas zone, about 30 percent of stations showed a similar workday pattern (between 9 am and 4 pm), with peak utilization averaging 60 percent of total capacity during those hours.
- In the Northeast zone (comprising six states), 28 percent of stations demonstrated a working-hour pattern (8 am – 6 pm), with peak utilization of 55 percent during those hours.
Safari points out that the variation in charging behavior variation between zones may be due to several factors: differences in the price of electricity, the availability of public charging stations to EV owners, and incentive programs that may encourage owners to charge their vehicles during off-peak hours.
As they continue to crunch a year’s worth of data on supercomputers, the team expects to find other “clustered” patterns in EV charging station usage — for example, during overnight hours, holidays, local and national events, and weather-related disasters. They hope that power grid operators can use the results of their analysis to efficiently price electricity, invest in developing new charging stations or expand existing ones, and balance supply and demand — ensuring charging port availability and reducing wait times.
A secondary focus of the research is to explore potential inequities in access and use of EV charging stations on a national scale. “Our findings can set the foundation for future research on the equitable accessibility, availability, and utilization of EV charging facilities across diverse communities with varied socio-economic status,” says UMD assistant professor Jiehong Lou, assistant research director of the Center for Global Sustainability.
In a previously published study, Lou and colleagues found that lower-income households face less accessibility to public EV infrastructure in both urban and rural geographies.
According to a Pew Research Center report, there were over 61,000 publicly accessible charging stations in the U.S. as of February 2024. Experts have found that many more are needed to accommodate future EV ownership. EV charging stations are mostly accessible to residents in urban areas, with only 17 percent located in rural areas, according to the report.
###
Safoura Safari, Deb Niemeier, Jiehong Lou are presenting this research on Tuesday, December 10, from 1:30 pm, at the JW Marriot Austin, Texas.
Spatial and Temporal Patterns of EV Charging Usage: A Nationwide Analysis Using Machine Learning Techniques – Tuesday, December 10, 1:30 p.m.
About SRA
The Society for Risk Analysis is a multidisciplinary, interdisciplinary, scholarly, international society that provides an open forum for all those interested in risk analysis. SRA was established in 1980. Since 1982, it has continuously published Risk Analysis: An International Journal, the leading scholarly journal in the field. For more information, visit www.sra.org.
SwRI showcases capabilities, evaluates novel fire-safety method with customized test
Customized lab test advances understanding of EV fire dynamics
Southwest Research Institute
image:
Southwest Research Institute successfully customized and conducted a full-scale evaluation of a novel enclosure for damaged electric vehicles at risk of catching fire. The custom test highlights SwRI’s expertise and ability to help advance emerging fire technology where no standardized testing exists.
view more
Credit: Southwest Research Institute
SAN ANTONIO — December 10, 2024 — Southwest Research Institute successfully customized and conducted a full-scale evaluation of a novel fire mitigation method designed to safely store damaged electric vehicles and batteries. SwRI engineers demonstrate customized research and development support for government and industry clients with novel projects and challenges related to emerging technologies where no standardized testing exists.
“SwRI established the nation’s first fire-focused research program 75 years ago, so our extensive capabilities allow us to develop novel tests to evaluate emerging technologies,” said Senior Research Engineer Kyle Fernandez, who led the experiment. “For this project, we created a customized test because no standardized test exists for EV containment enclosures yet.”
As the popularity of electric vehicles rises, the likelihood of accidents involving EVs also increases. Different makes and models of lithium-Ion batteries, which power electric and hybrid vehicles and can be hazardous when compromised, also proliferate.
“With a lot of emerging industries, the test standards haven't caught up with the new technologies,” said Karen Carpenter, director of Fire Technology at Southwest Research Institute. “SwRI specializes in creating customized test programs to evaluate scenarios that don't necessarily have a standardized method associated with them.”
Transporting or storing damaged EVs for repairs comes with inherent fire risks due to a phenomenon called thermal runaway, which can cause an electric battery to heat up so quickly it sparks a chain reaction that can lead to an out-of-control fire.
“Once an EV gets into an accident, the vehicle is potentially compromised and can catch fire at any point,” said Fernandez.
If a fire occurs in a repair facility, it can spread to nearby vehicles. Due to this risk, the National Highway Traffic Safety Administration recommends storing damaged EVs 50 feet away from other vehicles and structures, which can be challenging in urban settings with limited space.
“With our technical know-how and extensive lab space, we developed the capability to replicate thermal runaway to evaluate a vehicle enclosure to mitigate fire. Using thermocouples, we measured the temperature on the interior and exterior of the enclosure as fire fully engulfed the vehicle,” said Fernandez.
Using cameras, engineers monitored the fire from a safe location while collecting corresponding temperature and air quality data. While the interior wall reached nearly 2,000 degrees F at the height of the flames, the exterior wall remained cooler at just over 350 degrees F at the peak when the team flooded the container with an extinguishing agent to test the watertight seal. The customized experiment provided valuable data about the enclosure’s effectiveness.
SwRI’s Fire Technology Department is celebrating 75 years of fire technology research and development. Our engineers and researchers offer standard, nonstandard and custom fire testing services along with fire protection engineering, fire services technology and smoke toxicity evaluations. SwRI operates one of the largest environmentally friendly fire research facilities in the world, with more than 40,000 square feet of lab space, equipped with pollution abatement and wastewater collection systems.
For more information visit, https://www.swri.org/industries/fire-research-engineering.
Storing damaged EVs poses a fire hazard due to thermal runaway, a phenomenon that can cause a battery to ignite due to a chemical chain reaction. Southwest Research Institute replicated thermal runaway to test the novel fire mitigation method.
Credit
Southwest Research Institute
No comments:
Post a Comment