An Intra-Vehicular Wireless Multimedia Sensor Network for Smartphone-Based Low-Cost Advanced Driver-Assistance...

Area of Science:

• Automotive Engineering
• Computer Vision
• Wireless Sensor Networks

Background:

• Advanced Driver-Assistance Systems (ADAS) are predominantly found in high-end vehicles due to the cost of existing wired vision sensor solutions.
• Current ADAS implementations rely on extensive wired harnessing, increasing complexity and cost, limiting adoption in lower-end vehicles.

Purpose of the Study:

• To design and test a practical, low-cost wireless ADAS alternative.
• To leverage smartphone processing power and existing wireless technologies for enhanced vehicle safety features.
• To make ADAS technology more accessible for economical vehicle segments.

Main Methods:

• Developed a camera-based wireless sensor network utilizing a Wi-Fi Direct topology for intra-vehicle communication.
• Employed a smartphone as the central processing unit, running an ADAS application.
• Utilized a convolution neural network on the smartphone to process video streams from the wireless sensor network for object detection.

Main Results:

• Successfully implemented lane detection and collision detection features using wireless sensor data.
• Demonstrated real-time detection capabilities, confirming the network's performance.
• Validated the feasibility of a smartphone-based, intra-vehicle wireless network for ADAS.

Conclusions:

• A low-cost wireless ADAS solution is achievable by integrating smartphones and wireless sensor networks.
• The proposed system effectively extends perception capabilities to lower-cost vehicles.
• Wi-Fi Direct offers a viable topology for intra-vehicle wireless communication in ADAS applications.