High-Precision Self-Localization for Autonomous Driving: Visual Odometry According to Velocity Ranges

摘要

In order to run safely, autonomous driving requires the ability to self-localize. However, the many existing road conditions (obstacles, velocities, light, etc) make self-localization become very challenging. This research aims to develop self-localization methods to deal with such variable conditions, specially moving velocity. We present visual odometry methods for each range of low, medium and high velocities.