Assam Alzookery may have graduated from the University of Michigan in Ann Arbor, but the “eureka!” moment behind his startup Intvo happened during a trip to New York City. After spending a few years working at DENSO and researching camera vision systems for autonomous vehicles, the software engineer spent time traveling to gain real world insights of the myriad environments self-driving cars will have to navigate. Gotham is notoriously pedestrian-first, with the highest population density of any city in the country and almost 70,000 people per square mile. Watching New Yorkers go about their day-to-day lives perplexed him.
“From an engineering perspective, I began thinking about how an autonomous vehicle could drive in this kind of environment,” Alzookery said. “No two pedestrians act the same, and understanding their behavior is a huge hurdle to overcome. Predicting it is harder, but that’s Intvo’s sole mission,” he said.
“If a person is standing on the sidewalk looking for a vehicle, his behavior will be different the moment he sees a car,” Alzookery said. Behavior data could help eliminate pedestrian-vehicle fatalities. From 2009-2017, there were 494 in Ann Arbor, according to The Ann Arbor Observer. “We want to increase safety to meet the vision for Ann Arbor, which is zero pedestrian deaths,” he said.
Alzookery founded the company last May to provide smart software solutions that use artificial intelligence to observe, classify, and predict pedestrian behavior, solidifying confidence in autonomous cars with the help of artificial intelligence. Intvo has since become a recipient of a $60,000 grant from the Michigan Economic Development Corporation’s (MEDC) PlanetM early this August. Currently, Intvo is testing its behavior prediction software in Ann Arbor, tapping into the city’s infrastructure and network of 2,500 connected vehicles. By next year, the plan is to deploy its software to the city’s network and have it running in real time, notifying drivers when a distracted pedestrian is nearby. For now, Intvo is validating data and behavior models until they hit an internal threshold of accuracy.
“Are these models we’re building 80-90 percent accurate?” he posited. “If not, we have to go back and collect more data and refine our models.”
Operating within Ann Arbor and in Michigan itself provides a few benefits. Of course, there’s the city’s infrastructure and vehicle network, but the state also offers a full four seasons of weather types. The last bit is just as important to Intvo’s success as the former.
“Having this kind of weather gives us a big differentiator from other companies in other states,” he said. “The most important thing is testing software in a real environment, because this is where our software will live.”
By this time next year, Alzookery hopes it will move from its current pilot program and begin deploying Intvo’s software for real-time testing on the city’s streets, delivering pedestrian behavior data to drivers as it’s gathered. Intvo will also be demonstrating its tech at the North American International Auto Show next June, with the intent to gain attention from automotive suppliers and then pushing it out to commercial applications.
Intvo has also joined the PlanetM Landing Zone in downtown Detroit – a coworking space for mobility startup companies started by MEDC and the Detroit Regional Chamber’s MICHauto.
Article by Timothy J. Seppala – detroitdriven.us