Return to search

Design Optimization and Motion Planning For Pneumatically-Actuated Manipulators

Soft robotic systems are becoming increasingly popular as they are generally safer, lighter, and easier to manufacture than their more rigid, traditional counterparts. These advantages allow an increased sense of freedom in both the design and operation of these platforms. In this work, we seek methods of leveraging this freedom to both design and plan motions for two different serial-chain, pneumatically actuated manipulators developed by Pneubotics, a small startup company based in San Francisco. In doing so, we focus primarily on two related endeavors: (1) the optimal kinematic design of these and other similar robots (i.e., choosing link lengths, base positioning, etc.), and (2) the planning of smooth paths in joint space that enable these robots to perform useful tasks. Our method of design optimization employs a genetic algorithm in combination with maximin multi-objective optimization techniques to efficiently generate a diverse set of Pareto optimal designs. This set represents the optimal region of the design space and highlights inherent tradeoffs that designers must make when choosing a particular set of design parameters for manufacture. In our work, we have chosen to optimize inflatable robots to be both dexterous, and to be able to support loads near the ground with limited deflection. We have also applied our framework to optimize a plastic manipulator to perform painting motions. In our approach to motion planning we simultaneously optimize the base position and joint motions of a robot in order to enable its end effector to follow a smooth desired trajectory. While this method of path planning generalizes to any kind of robot, we envision it to be especially applicable to soft robots and other mobile robots that can be quickly and easily repositioned to perform tasks in varying environments. Our method of path planning works by moving a set of virtual robot arms (each representing a single configuration in a sequence) branching from a common base, to a number of assigned target poses associated with a task. Additional goals and hard constraints (including joint limits) are naturally incorporated. The optimization problem at the core of this method is a quadratic program, allowing constrained high-dimensional problems to be solved in very little time. We demonstrate our method by planning and performing painting motion on two different systems. We also demonstrate in simulation how our planner could be used to perform several common tasks including those involving, pick-and-place, wiping and wrapping motions.

Identiferoai:union.ndltd.org:BGMYU2/oai:scholarsarchive.byu.edu:etd-7289
Date01 April 2017
CreatorsBodily, Daniel Mark
PublisherBYU ScholarsArchive
Source SetsBrigham Young University
Detected LanguageEnglish
Typetext
Formatapplication/pdf
SourceAll Theses and Dissertations
Rightshttp://lib.byu.edu/about/copyright/

Page generated in 0.0017 seconds