Vision-Based Steering Control, Speed Assistance and Localization for Inner-City Vehicles

Area of Science:

• Robotics and Computer Vision
• Automotive Engineering
• Artificial Intelligence

Background:

• Autonomous vehicle technology is advancing, but sophisticated sensors increase costs.
• Developing cost-efficient solutions for automated driver assistance systems is crucial.
• Existing systems often rely on expensive sensor suites, limiting accessibility.

Purpose of the Study:

• To propose a cost-efficient autonomous vehicle system using a single monocular camera.
• To enable automatic steering control, speed assistance, and vehicle localization.
• To validate the system's performance for vehicles following predefined paths, like public transport.

Main Methods:

• Utilized a single monocular camera for perception and control.
• Implemented a computer vision approach to detect road lane lines for path following.
• Employed specially designed visual markers for vehicle localization and speed control assistance.
• Developed a vision-based control system for lane keeping under urban speed limits.

Main Results:

• The system successfully controlled steering, speed, and localization using only a monocular camera.
• Real-world driving tests on a closed circuit validated the approach.
• The vehicle traveled 7 km autonomously at speeds up to 48 km/h without interruption.

Conclusions:

• A single monocular camera system is feasible for autonomous route following in specific scenarios.
• This approach offers a cost-effective alternative to complex sensor systems for automated driving.
• The demonstrated system shows potential for applications like public transport automation.