Supermaneuverable Autonomous Pursuit - Peregrine Falcon Versus Pigeon Inspired UAVs
Abstract
A peregrine falcon pursuing a pigeon in flight provides compelling inspiration for unmanned aerial vehicle autonomy, flight control, grasping, perching, and rudderless morphing flight system design, because of the multifunctional capabilities of the falcon. The Highly Maneuverable Autonomous UAV DESI project involved a close collaboration among two Stanford university labs (Lentink and Cutkosky) and their industry partner (Skydio), to develop aerial vehicles capable of rudderless morphing flight in turbulence, perching on complex surfaces such as tree branches and catching airborne targets such as small UAVs in flight. To translate falcon capabilities to the UAV realm, we focused our bioinspired robotic multifunctional structures and autonomous pursuit research on four key challenges during two years of collaborative development. First, we developed the Stereotyped Nature inspired Aerial Grasper, SNAG, a robotic leg and end effector mounted to a quadrotor that can perch on a range of dirt, moss and lichen covered branches and can catch objects dynamically. Next, we developed the first rudderless morphing flyer, PigeonBot 2, a biohybrid rudderless morphing flyer with underactuated soft feathered wings and tail. It not only uncovered the previously unknown principles of morphing rudderless flight, it also demonstrated them in a high turbulence in a wind tunnel and during extensive autonomous flight testing outdoors. Then, we focused on achieving in-flight dynamic grasping with a quadcopter that can grasp target airborne drones in flight and recover from the collision that triggers the grasp. Finally, we demonstrate an aerial grasping mechanism integrated onto the autonomous Skydio X2 platform. We then used the Skydio vision-based autonomy engine to test grasping targets in flight. This work inspired the development of a trajectory generator that combines differential flatness with trajectory optimization to satisfy state constraints more directly and quickly.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 14, 2022
- Accession Number
- AD1230355
Entities
People
- Scott Delp
Organizations
- Stanford University