MIT and car makers developing AI ‘driving buddy’

MIT researchers are collaborating with Volkswagen and Audi to develop an AI-based ‘driving buddy’ that could help motorists avoid traffic jams or remind them to fill a nearly empty petrol tank. By observing a driver's habits, an artificial intelligence tool called AIDA (Affective Intelligent Driving Agent) can learn routines and how best to assist with navigation and maintenance. AIDA is being developed by Audi and the Electronics Research Lab of the Volkswagen Group of America  in collaboration with MIT's Personal Robots Group and its Media Lab and SENSEable City Lab.

“The key lies in all the real-time feeds that AIDA can process, both from inside the car and real-time information about the city outside the car,” according to professor Carlo Ratti, director of the SENSEable City Lab. “By combining and analyzing all these feeds, AIDA can start to understand your mood [and] the goals you would like to achieve.”

The project was inspired by the amount of sensor data streaming in from current automotive systems used to monitor everything from weather to traffic conditions. By fusing data streams and personalizing the relevant elements, AIDA could perform the same kind of tasks as a driving buddy.

Unlike a nagging backseat driver, the mood icons of AIDA's ‘face’ include a smile. Fitted in the centre of the dashboard, AIDA has an array of sensors designed to interpret the driver’s mood based on facial expression and other cues. In its learning process, AIDA would amass a personal database of a driver's most frequent destinations and routes, then compare them with its knowledge of environmental conditions, including local weather or special events.

AIDA would be capable of understanding typical human activities such as going to a shop or filling station. This ability is integrated into its database of locations and its observations of the driver's habits, such as driving to work at the same time every weekday. When the software senses that help is needed, for example to avoid a traffic jam, the system suggests an alternative route.