Efficient Vision-based Navigation - Learning about the Influence of Motion Blur

This video shows the result of our work on learning vision-based navigation. Two wheeled robots use visual features to navigate to a goal location. With reinforcement learning, they learned when to drive slowly to avoid motion blur, and when to drive fast to reach the goal quickly. More information and publications at http://hrl.informatik.uni-freiburg.de/animations-vnav.html

Category:

Science & Technology

Tags:

• mobile
• robot
• robotics
• navigation
• vision
• reinforcement learning
• motion blur
• Freiburg
• efficient
• SURF
• learning
• camera
• Albert-Ludwigs-Universität