In-Vehicle Speech Command Operated Driver Assist System for Vehicle Actuators Control using Deep Learning Techniques
Prasanna Gawade, Suresh Kumar P · 2021
Considering road accidents due to distracted driving, according to NHTSA around 10% to 17% injuries and fatalities are reported every year. Last year NHTSA logged 3142 fatalities due to distraction. To minimize this, considering driver's safety and convenience, a speech command operated in-vehicle actuators control driver assist system using Deep Neural Network is proposed. The proposed system assists to keep eye on the road, and hands on steering wheel ensuring safety. Existing actuation are manually performed by driver causing distraction thus developing an in-vehicle customized driver assist system for accepting speech command inputs, processing, recognizing, classifying using deep neural networks and actuating the desired vehicle actuator.Customized speech commands for Automotive Actuators Control (AAC) and generic benchmark Google Speech Commands (GSC v1) are validated using CNN and ConvLSTM model.Classification accuracies obtained are 94% and 98.30% for AAC and 94.47% , 92.20% for (GSC v1).