Toward autonomous avian-inspired grasping for micro aerial vehicles

Area of Science:

• Robotics
• Aerial Robotics
• Control Systems

Background:

• Micro aerial vehicles (MAVs), especially quadrotors, have broad applications.
• Existing aerial manipulation literature primarily uses quasi-static models, limiting dynamic capabilities.
• Agile aerial grasping of moving objects remains a significant challenge.

Purpose of the Study:

• To develop dynamic grasping capabilities for quadrotors, mimicking agile avian predators.
• To enable fast pick-and-place operations, object transportation, and sensor deployment/retrieval.
• To present a robust methodology for aerial robots to grasp objects dynamically.

Main Methods:

• Developed a dynamic grasping approach that explicitly models robot and gripper dynamics.
• Implemented trajectory planning and control algorithms for dynamic grasping.
• Utilized both motion capture systems and onboard sensors (monocular camera, IMU) for grasping.
• Mapped quadrotor dynamics to a virtual image plane for image-based trajectory planning and control.

Main Results:

• Demonstrated successful dynamic grasping of a cylindrical object using a quadrotor.
• Validated the approach using both external motion capture and purely onboard sensing.
• Presented a vision-based controller with guaranteed convergence properties for dynamic grasping.
• Showcased experimental results with a quadrotor equipped with an articulated gripper.

Conclusions:

• Dynamic grasping is feasible and effective for quadrotors, significantly advancing aerial manipulation.
• The proposed vision-based control and trajectory planning methods enable agile and precise object capture.
• This research opens possibilities for more sophisticated autonomous aerial robotic operations.