Teresa Schuster/Staff Writer
Students may be driven to school by semi-autonomous vehicles having full conversations with each other, thanks to ongoing research into vehicular communications at FIU.
Two Ph.D. students, Sandhiya Reddy Govindarajulu and Imtiaz Nasim, along with faculty, are working on technology designed to integrate machine learning with vehicle networks in order to make vehicles safer and more efficient.
Their research focuses on transmitting information between cars and is intended to pave the way for semi-autonomous vehicles, which Govindarajulu said would be safer than current vehicles.
“We might save a lot of people,” Govindarajulu said. “In 2017, there was a report of almost 40,000 road accidents per year. We are losing 100 lives per day.”
According to Govindarajulu, the current technologies used to make regular cars less dangerous are not always reliable.
“For example, in heavy rainfall or snow or fog, normal cameras don’t work properly,” she said. “Our aim is to connect these technologies from one vehicle to another vehicle. It’s almost like sharing information.”
Nasim said that the research is geared towards semi-autonomous vehicles, which still allow for some human intervention, instead of fully autonomous vehicles, which operate independently of humans.
He believes this will be safer.
“Cars only know what you program them to do,” he said. “They cannot react to a new situation. [But humans] can feel sleepy. You cannot depend on either one of them.”
To accomplish this, vehicles must be able to receive and transmit large quantities of information quickly. But this presents many challenges, particularly if they are traveling at high speeds.
“When you are a normal cellular user, one cellular tower is supporting you,” Nasim said. “When you’re using a vehicle, moving at a very high speed, you’re out of the range of a cellular tower very frequently. If you lose your connection or information for just a second or millisecond it may be very dangerous.”
To help overcome this, Nasim is researching ways to maximize the download capacity of the cars, which would let them receive more information in a given period of time. He compares this to downloading a movie on a smartphone.
“You want it to be downloaded at a very high speed. When you increase the download capacity, it may take two minutes instead of ten,” he said.
In the interest of speed, it is also necessary to communicate directly between vehicles, instead of always using a tower as an intermediary, which Nasim said would be easier.
“Vehicles are not designed to do this,” he said.
Aside from the mechanics of vehicular communications, which the team’s research focuses on, there are other challenges.
Govindarajulu said that privacy is also an important consideration since vehicles would transmit information about their speed and surrounding environment to other vehicles.
“Privacy is the first concern. A few people don’t want to share that information,” said Govindarajulu.
Govindarajulu also warned someone could hack the communications system, sending vehicles false information and misuse it.
Despite these challenges, both he and Govindarajulu believe that the technology will be overwhelmingly beneficial.
“It would make our driving safer and save a lot of time as well,” said Govindarajulu, who believes that it could also help victims of car accidents. “If there’s an accident, we can send video information to the paramedics, we can say what is the condition of the patients.”
And the technology wouldn’t only be for cars. Nasim said that it would have other possible applications and that it could be used in ships and airplanes in the future.
The team has already been recognized for their research, being the recipients of an NSF grant, among others, which Nasim considers significant.
“It’s a great achievement for FIU that their faculty and students have already gained big grants from the NSF and others,” said Nasim.
Both Nasim and Govindarajulu mentioned that Google, Tesla, and other companies have already developed cars that incorporate machine learning. Nasim said that if there is rapid progress in the field, by 2024 students can expect semi-autonomous vehicles to become widely available.
Last spring, Govindarajulu, whose research focuses on making components of antennas and radio frequency systems smaller, was awarded an award for the best student paper in a competition. She considers this a major turning point in her research.
“It was important to me to be socially responsible,” Govindarajulu said. “There are many technologies which are complicated. Normal people can’t use [them]. This will impact people.”