From falcons to robots : design principles and systems analysis for autonomous bio-inspired visually guided aerial grasping robots
Abstract/Contents
- Abstract
- The peregrine falcon's ability to pursue and grasp flying pigeons at high speed is a feat that is unparalleled in its demonstration of its supermaneuverable capabilities. It is thought that the falcon's pursuit and grasping ability is enabled by a fast, long-range visual tracking system, dynamic pursuit trajectories modeled by established guidance laws, as well as legs and claws that enable in-flight grasp and recovery following the capture of prey. The falcon's agility and the systems that enable it can therefore inspire the design of capable novel aerial grasping robots that capture rogue drones in flight in sensitive airspaces. Juxtaposed with the practical application is the opportunity to use bio-inspired aerial grasping robots to further understand the falcon's ability. This inspired the research direction for FalconQuad, an aerial grasping robot that visually tracks, pursues, and captures target quad-rotors in flight. In this thesis, I show the engineering considerations essential to implementing falcon-inspired pursuit and grasping for FalconQuad. This includes hardware development of the robot, a bio-inspired pursuit planning algorithm that considers hardware limitations, and a grasping system that minimizes disturbance on the pursuer throughout the pursuit flight. Using this robot, we understand the critical flight conditions for successful pursuit and capture by developing simulations and experiments that model the pursuit scenario using aerial robots as scale models of the biological example. The experiments and hardware implementation consequently informed a model that explains how the falcon might prioritize different factors in stages of aerial pursuit and grasping. Specifically at a distance, visual state estimation of the target is degraded and cannot be fully relied upon by the planning system. Therefore, the pursuer's objective is to minimize distance most effectively. Nearing interception, the pursuer must rely primarily on the mechanical redundancy of its gripper system rather than precise flight control algorithms. Following the experiments, a systems-level analysis shows that better visual tracking and pursuit logic could improve the performance of FalconQuad by precisely predicting target motion and informing when to attempt pursuit and interception. This exploration provokes future research to elucidate how depth to target and biological visual sensing limitations play a role in the falcon's ability to capture aerial prey. I hope this work will inspire novel multifunctional aerial robots enabled by developments in smart materials on platforms that could, in-turn, be used to evaluate new technologies for future aerospace vehicle concepts.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2023; ©2023 |
| Publication date | 2023; 2023 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Hoffmann, Kenneth Alexander Wolfgang |
|---|---|
| Degree supervisor | Cutkosky, Mark R |
| Thesis advisor | Cutkosky, Mark R |
| Thesis advisor | Follmer, Sean |
| Thesis advisor | Lentink, David, 1975- |
| Degree committee member | Follmer, Sean |
| Degree committee member | Lentink, David, 1975- |
| Associated with | Stanford University, School of Engineering |
| Associated with | Stanford University, Department of Mechanical Engineering |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Kenneth Hoffmann. |
|---|---|
| Note | Submitted to the Department of Mechanical Engineering. |
| Thesis | Thesis Ph.D. Stanford University 2023. |
| Location | https://purl.stanford.edu/ss900gp7410 |
Access conditions
- Copyright
- © 2023 by Kenneth Alexander Wolfgang Hoffmann
Also listed in
Loading usage metrics...