Asphalted road temperature variations due to wind turbine cast shadows

Area of Science:

• Robotics and Computer Vision
• Sensor Fusion
• Autonomous Systems

Background:

• Dynamic shadows from wind turbine blades pose challenges for autonomous vehicle navigation.
• Shadows in visible spectrum images can be misclassified as obstacles, requiring removal.
• Existing shadow removal techniques may not be fully effective in all conditions.

Purpose of the Study:

• To develop a technique for avoiding dynamic shadow removal in visible spectrum images.
• To leverage infrared imaging to overcome limitations of visible spectrum analysis.
• To improve the reliability of autonomous navigation systems in wind farm environments.

Main Methods:

• Utilizing infrared spectrum images where dynamic shadows disappear.
• Performing sensor fusion between thermal and visible spectrum bands.
• Developing a technique to avoid explicit shadow removal stages.

Main Results:

• Demonstrated the effectiveness of infrared imaging in eliminating dynamic shadows.
• Successfully fused thermal and visible data to enhance perception.
• Showcased a method to bypass the need for complex shadow removal preprocessing.

Conclusions:

• The proposed technique effectively addresses the challenge of dynamic shadows in autonomous navigation.
• Infrared imaging combined with visible spectrum data offers a robust solution for perception in complex environments.
• This approach enhances the safety and efficiency of autonomous vehicles in wind farms.