Integrating Visual SLAM and Smooth Speed Control for Realizing Real-Time Autonomous System
Mao-Jen Ko, Yuchen Lin, Chih-Hung Cheng, Thanh-Tan Nguyen · 2024
In recent years, the rapid development of autonomous system technology has increasingly positioned it as a crucial player in modern transportation. Appropriate speed planning emerges as one of the key elements for achieving safe and efficient autonomous system operations. Speed planning not only requires adherence to road speed limits and ensuring safe distances to avoid collisions but also necessitates consideration for ensuring passenger comfort. This paper proposes a method that restricts acceleration and jerk to achieve smooth speed control, enabling autonomous system to emulate the driving style of human drivers. Utilizing visual SLAM for feature extraction and matching on images, a sparse point cloud three-dimensional map of the environment is constructed. The recorded path is tracked using the Pure Pursuit algorithm. Finally, this paper will validate the proposed approach through an embedded system deployed in a real vehicle. The integration of visual SLAM and smooth speed control technology realizes a real-time and smooth autonomous system, providing a safer autonomous system experience.