SMART CAR Natural User Interface Technology CAR CUSTOMIZATION

SMART CAR Natural User Interface Technology CAR CUSTOMIZATION BY USER RECOGNITION

About the project…

Smart Car Definition: An automobile with an advanced system which can provide an enhanced driving experience and also provides active safety features

State of the Art Crash Avoidance Systems Energy Consumption Lane Tracking Night-vision Obstacle Detection Smart Cruise Control Driver-Car Vocal interaction

Car Technology in the works… Self-driven cars. Augmented Reality dashboards – Windshields and touchscreen windows (BMW & Toyota). Adaptive Highbeam Assist (Lexus) – gradually adjusting light distribution based on distance of approaching cars.

Future Car Technologies Vehicle-to-vehicle communication (V 2 V) – using wireless signals to send information between cars. Vehicle-to-infrastructure communication (V 2 I) – allow vehicles communication with traffic signals. Provide information to vehicle about safety issues Wi. Fi Internet Access in moving vehicles.

What are driver’s needs? COMFORT ASSISTANCE SECURITY

What do we provide? A user-friendly system, based on a NUIT: Recognition of the driver who takes a seat Loading of a user-based profile with a given customization The needs of the specific user are fulfilled

NUIT applied to Smart Car Which technologies are we implementing to develop a Natural User Interface? FACE RECOGNITION SPEECH RECOGNITION

Face Recognition Cross-correlation? Face-tracking to look for certain patterns to detect mood or fatigue/tiredness? …

Problems of Face Recognition Creating a unique face signature for every person Sun glasses Painted eyebrows Long hair Beards Viewing angle of face Poor lighting Big smiles Privacy

Speech Recognition Cross correlation? Dynamic Time Warping? Hidden Markov Models …

Problems of Speech Recognition Error rates increase as the vocabulary size grows The 10 digits "zero" to "nine" can be recognized essentially perfectly, but vocabulary sizes of 200, 5000 or 100000 may have error rates of 3%, 7% or 45% respectively. Spontaneous Speech like “uh” and “um” Environmental noise (e. g. Noise in a car or a factory) Acoustical distortions (e. g. echoes, room acoustics) Homonyms Hidden Markov Models

The aim: to satisfy driver needs COMFORT: Face recognition can enable automatic customization in order to fit the driver’s preferences. Automatic rear view mirrors set Automatic seat set Automatic suspension set Radio station presets Driving wheel’s height …

The aim: to satisfy driver needs ASSISTANCE: Speech recognition may be useful in order to provide certain capabilities to the car. Update task list (in the mobile phone) Give customization orders (turn on/off the radio, make a call, . . . ) without taking eyes off road. Select a destination to update GPS system …

The aim: to satisfy driver needs SECURITY: security may be improved by using face recognition. Engine won’t start if you are not an authorized driver Different drivers may have diferent permissions or speed limitations. …

Kiitos!