System and Machine Learning Based Method for Smart E-Mirror Recommendation 2024-01-2464

Autonomous vehicles or self-driving cars and semi-autonomous cars provide numerous driving assistance to the driver and passengers. However, with the advancements in driving technology the driver’s experiences and preferences are also taken into consideration for achieving the advanced safety goals. The drivers’ comfort and experiences are considered with the design and development of modern-day vehicles makes the driver’s preferences crucial for providing the driving experience. During the process of driving, the experience received by the driver is associated with the functional safety of the automotive system and hence the experience should be smooth with respect to safety. In this regard, the personalization in the space of the driver assistance system is gaining importance in analysing the driver’s behaviors and providing personal driving experiences to the driver. From the driver’s perspective the driving environment, the driving patterns of the different drivers, road conditions, etc. changes constantly. And hence, instead of providing a static driving experience to the driver through the usage of advanced driver assistance systems, an addition of personalized driving experiences to the driver assistance system makes the driving experience smooth. One of the critical driver assistance systems is the E-mirror, which is a camera-based system to replace the rear view, and side view mirrors of the vehicle by providing a smart experience to the driver. The E-mirror personalization is achieved by recommending the driver his preferred E-mirror control settings over his previous usage history. So, here an E-mirror recommendation system is proposed, which provides user-consent based recommendations to the driver to enhance the driving experience. The personalized smart E-Mirror provides personalized experiences to the users based on their previous user settings and mirror preferences. This reduces the need for constant or periodic change in the E-mirror adjustments for every user or every usage of the user. The personalized system observes the user activity over a prolonged period of time in using the E-mirror across the different driving styles, driving environments, climate conditions for all the users and collects a driver behavioral data in using the e-mirror through his settings and preferences. The personal experience of the user is achieved by recommending to the user about the e-mirror settings over a pre-defined recommendation criterion. The proposed E-mirror recommendation system receives numerous vehicular, non-vehicular and driver’s activity parameters in modelling the users’ behaviors through a deep learning recommendation algorithm. The collected user activity data for the E-mirror preferences are observed and the model is trained on on-device for providing the E-mirror setting recommendations to the driver upon receiving the user-consent. Hence, the personalized e-mirror recommendation system automatically controls the E-mirror camera settings after receiving the recommendations and also after getting user consent for the same.