Optimizing the structure and movement of a robotic bat with biological kinematic synergies
Abstract
In this article, we present methods to optimize the design and flight characteristics of a biologically inspired bat-like robot. In previous, work we have designed the topological structure for the wing kinematics of this robot; here we present methods to optimize the geometry of this structure, and to compute actuator trajectories such that its wingbeat pattern closely matches biological counterparts. Our approach is motivated by recent studies on biological bat flight that have shown that the salient aspects of wing motion can be accurately represented in a low-dimensional space. Although bats have over 40 degrees of freedom (DoFs), our robot possesses several biologically meaningful morphing specializations. We use principal component analysis (PCA) to characterize the two most dominant modes of biological bat flight kinematics, and we optimize our robot's parametric kinematics to mimic these. The method yields a robot that is reduced from five degrees of actuation (DoAs) to just three, and that actively folds its wings within a wingbeat period. As a result of mimicking synergies, the robot produces an average net lift improvesment of 89% over the same robot when its wings cannot fold.
Additional Information
© The Author(s) 2017. Article first published online: November 11, 2018; Issue published: September 1, 2018. We would like to thank the team of graduate and undergraduate students from aerospace, electrical, computer, and mechanical engineering departments at the University of Illinois at Urbana-Champaign for their contribution to construct the initial prototype of B2. The biological motion capture data set was provided by Kenneth Breuer and Sharon Swartz from Brown University. We greatly appreciate their insights and helpfulness in this project. We would like to thank them in their assistance with this, as well as José Iriarte-Díaz for compiling the data. The experimental results were performed in the Intelligent Robotics Laboratory (IRL) of the Coordinated Science Laboratory Studio. This research was supported by NSF Grant 1427111.Attached Files
Accepted Version - IJRR2017_v14.pdf
Files
Name | Size | Download all |
---|---|---|
md5:454d04c3c2a1dfac429d23a5acff461d
|
3.0 MB | Preview Download |
Additional details
- Eprint ID
- 87816
- Resolver ID
- CaltechAUTHORS:20180713-074446939
- NSF
- CMMI-1427111
- Created
-
2018-07-13Created from EPrint's datestamp field
- Updated
-
2021-11-15Created from EPrint's last_modified field
- Caltech groups
- GALCIT