Drivers in today’s society are often distracted by cellphones, texting, drinking coffee, changing the radio station or using a GPS—leading to an increase in accidents. MABE assistant professor Subhadeep Chakraborty and his research team have joined forces with Civil Engineering professors Asad Khattak and Sashi Nambisan to integrate automated controls in vehicles that can predict a driver’s reactions to different scenarios and respond faster than the driver, which could possibly prevent accidents.
Chakraborty and his team are focusing on a vehicle that can monitor and get to know its driver. If the car knows the driver is prone to a certain action, it can alert the driver and potentially warn other vehicles around it to “watch out.” Providing drivers with this information efficiently may make the difference between avoiding an accident or contributing to the annual 871 billion in losses incurred by traffic crashes. The idea is to record the driver’s reactions to different scenarios using sensors equipped in a vehicle and then update a sophisticated software-based cognitive model of the driver. This model would essentially be able to determine the orientation and position of vehicles surrounding it, predict the human’s reaction and compute the optimal maneuver much faster than the driver.
To test this concept, the team is utilizing a Hummer that was donated to the MABE department by Mark Dean, a professor in the Electrical Engineering and Computer Science department. The Hummer has been fully instrumented with onboard GPS, a laser scanner on each corner, a 360o camera, and inertial measurement units. The team is using this platform to collect driver data and begin to assess the accuracy of this driver prediction model.
By Kathy Williams