Model-Based Design, Control, and Planning for Legged Microrobots
Doshi, Neel Deepak
MetadataShow full item record
CitationDoshi, Neel Deepak. 2019. Model-Based Design, Control, and Planning for Legged Microrobots. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
AbstractLegged animals of different sizes demonstrate impressive locomotion capabilities in unstructured and challenging environments. Consequently, scientists and engineers have sought to understand biological locomotion and replicate its versatility in numerous legged robots. A key aspect of this research is the development of dynamical models that identify principles of biological locomotion and inform the development of increasingly effective legged robots. Indeed, these models will play an important role in the eventual development of robotic platforms capable of reliable locomotion in a variety of environments.
Though research in legged-robotics is primarily focused on large (~100 cm in length) robots, advances in manufacturing have enabled the development of a number of insect-scale (0.1 cm to 10 cm in length) legged microrobots. These smaller robots can be used as at-scale physical models to understand the remarkable locomotion capabilities of insects or to explore the scaling laws of legged locomotion. Moreover, these microrobots enable a new set of applications, including inspecting confined spaces, monitoring the environment, and exploring hazardous areas.
Given the nascent stage of the field of legged microrobotics, most of the research is focused on the characterization of experimental performance and the development of suitable manufacturing paradigms. On the other hand, drawing inspiration from the research on larger legged robots, this dissertation focuses on using model-based techniques to both gain a deeper understanding and the improve the performance of legged microrobots.
This is achieved via three separate case-studies that demonstrate the utility of our model-based approach. First, we perform a model-based redesign to improve payload capacity and enable novel applications. Second, we design model-based estimators and controllers to enable locomotion at a wide range of stride frequencies. Finally, we develop and evaluate a framework for planning whole-body locomotion trajectories that improve performance and enable dynamic behaviors. More importantly, these case studies enable a deeper understanding of the flexure-based transmissions, uncover effective locomotion principles for legged microrobots, and improve the state-of-the-art for trajectory optimization through contact.
Citable link to this pagehttp://nrs.harvard.edu/urn-3:HUL.InstRepos:42029637
- FAS Theses and Dissertations