Prototype driving “coach” improves mileage, car safety
Sure, cars are useful… but they’re also energy-consuming, environmentally-unfriendly and even life-threatening. Recent research led by researchers from the University of Oulu, “Personalised Assistance for Fuel-efficient Driving,” develops a prototype for a driver assistance system that would — after gathering what we might call context clues — improve safety and fuel efficiency and reduce environmental impact by sharing key information with drivers.
The “context-aware” system “collects, fuses and analyses diverse information, like digital map, weather, traffic situation, as well as vehicle information to provide drivers in-depth information regarding their previous trip along with personalized hints to improve their fuel-efficient driving in the future,” explain the researchers. The system also analyzes its own performance and driver feedback “to correct itself to serve the driver more appropriately.”
Two of the major problems associated with cars are their impact on air pollution and the inherent danger of car crashes. Factors including car model, traffic, weather conditions and the driver’s own experience impact both of these issues. “Often, a driver cannot understand or distinguish all these factors in order to optimize his driving,” explain the researchers. The Driving Coach system outlined by the researchers provides the driver with “information on his driving behavior and instructions for safer and more environmentally-friendly driving.”
While driving assistant tools are plentiful on the market, in development and even in academia, the researchers in this case studied the problem from the “computing point of view,” proposing a reference architecture as well as a Coach prototype. The prototype “capture[s] and understand[s] the situation a driver is in,” uses on-board sensors and external services to obtain relevant information, and adapts its decision-making based on driver progress and feedback.
The researchers drew up an extensive visualization of the Coach, showing how the driver’s actions, conditions outside the car and the Coach’s own analysis work together to yield recommendations about the trip, which the driver can then use on her next trip. They also created reference architecture, which computer science buffs can check out here, for such context-aware driving systems like their own Coach.
The Driving Coach solution they propose doesn’t offer real-time support, as some car systems do, but instead analyzes data post-trip to provide drivers with information that may inform upcoming trips. “Driving Coach uses diverse real context information to perform the analysis, namely weather, road geometry, driving behavior, and traffic situation information. With such a rich set of context sources, it is unique,” point out the researchers.
The researchers also suggested potential improvements and future evolutions for the Driving Coach: a personalized cost function that compares costs of different routes; improvements to the fuel consumption models based on more detailed data, and extra features, such as the driver’s ability to annotate a trip. (Annotating a trip might be useful in a case where, for example, a driver gives a friend’s kids a lift to school; the driver could “inform” the Coach that the extra trips are necessary and not time-wasters.) The system might also be eventually evolved to include real-time interaction with the driver.
Related Posts
Category: Transportation