Enhanced CNN based approach for IoT edge enabled smart car driving system for improving real time control and...

Area of Science:

• Intelligent Transportation Systems
• Edge Computing
• Autonomous Driving

Background:

• Human-driven vehicles differ significantly from IoT edge-enabled smart driving systems.
• Real-time control of steering, throttle, and brakes is crucial for autonomous navigation.
• Edge computing offers potential for real-time data processing in autonomous systems.

Purpose of the Study:

• To investigate critical control factors differentiating human-driven vehicles from IoT edge-enabled smart driving systems.
• To improve autonomous vehicle decision-making using edge computing and high-precision sensors.
• To develop and evaluate a Convolutional Neural Network (CNN)-based system for enhanced vehicle navigation.

Main Methods:

• Gathered real-time time-series data from LiDAR, radar, GPS, IMU, and ultrasonic sensors.
• Utilized edge nodes for preprocessing sensor data before cloud server transmission.
• Employed a CNN architecture with convolutional, pooling, and fully linked layers for control vector prediction, enhanced by image data.

Main Results:

• The CNN-based edge-enabled driving system demonstrated superior navigation accuracy compared to conventional methods.
• Achieved a Root Mean Square Error (RMSE) of 15.123 and a loss value of 2.114.
• Edge computing effectively reduced computational delay, improving vehicle autonomy.

Conclusions:

• Edge computing significantly enhances vehicle autonomy and reduces processing delays in smart driving systems.
• The proposed CNN-based approach offers a more effective solution for autonomous navigation.
• Future work will involve real-world validation to assess system performance in dynamic environments.