From biokinematics to a robotic active vision system.

Area of Science:

• Robotics and Computer Vision
• Biomimetics and Bio-inspired Engineering
• Animal Behavior and Sensory Ecology

Background:

• Barn owls possess fixed eyes, necessitating unique head movements for effective visual scanning and prey detection.
• These head motions are characterized by efficient, side-to-side peering, crucial for their predatory success.
• Understanding these natural behaviors offers insights into optimizing artificial vision systems.

Purpose of the Study:

• To investigate the efficacy of barn owl head movements as a model for enhancing machine vision.
• To bridge the gap between biological active vision and current machine vision capabilities.
• To test the hypothesis that biomotion-based scanning improves accuracy for static objects.

Main Methods:

• Barn owl head movements were measured and recorded.
• These biomotions were used to control a robot equipped with a depth camera for scanning tasks.
• Scan accuracy using barn owl-inspired trajectories was compared against intuitive scanning methods.

Main Results:

• Robot scanning trajectories based on barn owl biomotions demonstrated consistently higher accuracy for static objects.
• The biomotion-based approach outperformed intuitive scanning patterns in experimental trials.
• This study provides proof-of-concept for bio-inspired viewpoint manipulation in robotic vision.

Conclusions:

• Bio-inspired viewpoint manipulation, using barn owl head movements, can significantly enhance the scanning accuracy of robotic systems.
• Mimicking natural predatory visual strategies offers a promising avenue for advancing machine vision.
• Applications include improved inspection in manufacturing and enhanced perception for autonomous robots.