Modeling and Estimation of Bat Flight for Learning Robotic Joint Geometry from Potential Fields
| dc.contributor.author | Bender, Matthew Jacob | en |
| dc.contributor.committeechair | Mueller, Rolf | en |
| dc.contributor.committeechair | Kurdila, Andrew J. | en |
| dc.contributor.committeemember | Southward, Steve C. | en |
| dc.contributor.committeemember | Woolsey, Craig A. | en |
| dc.contributor.committeemember | Furukawa, Tomonari | en |
| dc.contributor.department | Mechanical Engineering | en |
| dc.date.accessioned | 2018-11-01T08:00:38Z | en |
| dc.date.available | 2018-11-01T08:00:38Z | en |
| dc.date.issued | 2018-10-31 | en |
| dc.description.abstract | In recent years, the design, fabrication, and control of robotic systems inspired by biology has gained renewed attention due to the potential improvements in efficiency, maneuverability, and adaptability with which animals interact with their environments. Motion studies of biological systems such as humans, fish, insects, birds and bats are often used as a basis for robotic system design. Often, these studies are conducted by recording natural motions of the system of interest using a few high-resolution, high-speed cameras. Such equipment enables the use of standard methods for corresponding features and producing three-dimensional reconstructions of motion. These studies are then interpreted by a designer for kinematic, dynamic, and control systems design of a robotic system. This methodology generates impressive robotic systems which imitate their biological counter parts. However, the equipment used to study motion is expensive and designer interpretation of kinematics data requires substantial time and talent, can be difficult to identify correctly, and often yields kinematic inconsistencies between the robot and biology. To remedy these issues, this dissertation leverages the use of low-cost, low-speed, low-resolution cameras for tracking bat flight and presents a methodology for automatically learning physical geometry which restricts robotic joints to a motion submanifold identified from motion capture data. To this end, we present a spatially recursive state estimator which incorporates inboard state correction for producing accurate state estimates of bat flight. Using these state estimates, we construct a Gaussian process dynamic model (GPDM) of bat flight which is the first nonlinear dimensionality reduction of flapping flight in bats. Additionally, we formulate a novel method for learning robotic joint geometry directly from the experimental observations. To do this, we leverage recent developments in learning theory which derive analytical-empirical potential energy fields for identifying an underlying motion submanifold. We use these energy fields to optimize a compliant structure around a single degree-of-freedom elbow joint and to design rigid structures around spherical joints for an entire bat wing. Validation experiments show that the learned joint geometry restricts the motion of the joints to those observed during experiment. | en |
| dc.description.abstractgeneral | In recent years, robots modeled after biological systems have become increasingly prevalent. Such robots are often designed based on motion capture experiments of the animal they aim to imitate. The motion studies are typically conducted using commercial motion capture systems such as ViconTM or OptiTrackTM or a few high-speed, high-resolution cameras such as those marketed by PhotronTM or PhantomTM. These systems allow for automated processing of video sequences into three-dimensional reconstructions of the biological motion using standard image processing and state estimation techniques. The motion data is then used to drive robotic system designs such as the SonyTM AiboTM dog and the Boston Dynamics Atlas humanoid robot. While the motion capture data forms a basis for these impressive robots, the progression from data to robotic system is neither algorithmic nor rigorous and requires substantial interpretation by a human. In contrast, this dissertation presents a novel experimental and computational framework which uses low-speed, low-resolution cameras for capturing the complex motion of bats in flight and introduces a methodology which uses the motion capture data to directly design geometry which restricts the motion of joints to the motions observed in experiment. The advantage of our method is that the designer only needs to specify a general joint geometry such as a ball or pin joint, and geometry which restricts the motion is automatically identified. To do this, we learn an energy field over the set of kinematic configurations observed during experiment. This energy field “pushes” system trajectories towards those experimentally observed trajectories. We then learn compliant or rigid geometry which approximates this energy field to physically restrict the range of motion of the joint. We validate our method by fabricating joint geometry designed using both these approaches and present experiments which confirm that the reachable set of the joint is approximately the same as the set of configurations observed during experiments. | en |
| dc.description.degree | Ph. D. | en |
| dc.format.medium | ETD | en |
| dc.identifier.other | vt_gsexam:17269 | en |
| dc.identifier.uri | http://hdl.handle.net/10919/85615 | en |
| dc.publisher | Virginia Tech | en |
| dc.rights | In Copyright | en |
| dc.rights.uri | http://rightsstatements.org/vocab/InC/1.0/ | en |
| dc.subject | Bat Flight | en |
| dc.subject | Motion Capture | en |
| dc.subject | Motion Estimation | en |
| dc.subject | Bayesian Filtering | en |
| dc.subject | Empirical Potentials | en |
| dc.subject | Manifold Estimation | en |
| dc.subject | Bioinspired Design | en |
| dc.title | Modeling and Estimation of Bat Flight for Learning Robotic Joint Geometry from Potential Fields | en |
| dc.type | Dissertation | en |
| thesis.degree.discipline | Mechanical Engineering | en |
| thesis.degree.grantor | Virginia Polytechnic Institute and State University | en |
| thesis.degree.level | doctoral | en |
| thesis.degree.name | Ph. D. | en |
Files
Original bundle
1 - 1 of 1