3D pose estimation of ground rigid target based on ladar range image

Area of Science:

• Robotics and Automation
• Computer Vision
• Sensor Fusion

Background:

• Accurate target pose estimation is crucial for simplifying target recognition in laser radar (ladar) systems.
• Existing methods can be complex and sensitive to occlusion and noise.

Purpose of the Study:

• To propose a novel and simplified algorithm for 3D pose estimation of rigid ground targets using ladar.
• To leverage the geometric characteristics of ground targets and ladar range images for robust pose estimation.

Main Methods:

• Utilized planar normals of rigid targets as directional vectors for model coordinate system axes.
• Developed a method to estimate 3D pose angles based on these normals and ladar 3D geometric features.
• Conducted simulation experiments with six military vehicle models.

Main Results:

• Achieved estimation errors of less than 2° in self-occlusion scenarios.
• Demonstrated accurate pose estimation (error < 2.5°) even with up to 80% occlusion for the tank model, provided upper and side planes are not fully obscured.
• Confirmed the method's robustness to noise and overall effectiveness.

Conclusions:

• The proposed ladar-based method offers a simplified and effective approach to 3D pose estimation for rigid ground targets.
• The technique shows high accuracy and robustness against occlusion and noise, enhancing target recognition capabilities.