Shape Control of Tendon-Actuated Tensegrity Structures
Open Access
- Author:
- Osikowicz, Nathaniel
- Area of Honors:
- Aerospace Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Puneet Singla, Thesis Supervisor
Robert G. Melton, Thesis Honors Advisor - Keywords:
- Tensegrity
Robotics
Nonlinear Control
Deployable Structures - Abstract:
- As we further extend our reach into outer space, there exists an unmet need for au- tonomous agents to carry out highly dexterous manipulation tasks such as on-orbit servicing and habitat construction. In order to be packaged efficiently for transport and autonomously deployed at a remote destination, these robotic mechanisms must be lightweight, yet highly articulated. Tensegrity structures, which comprise a continuous tendon network, are a suitable candidate for carrying out dexterous manipulation tasks in outer space. This thesis focuses on controlling the shape of tensegrity structures by changing the tension in the supporting tendons. A vector-based approach is used to model the multi-body dynamics of tensegrity structures in a non-minimal coordinate system. By modeling the dynamics of each bar member with 6 degrees of freedom rather than 5, we avoid the use of transcendental functions to improve the accuracy of numerical simulations. This methodology is further extended to handle Class-k structures by modeling bar contact forces as Lagrange constraint forces. A reduced-order model is then constructed to solve for the corresponding Lagrange multipliers in closed-form. Leveraging the vector-based dynamics model, a state feedback controller is developed to regulate the shape of a tensegrity structure to a desired reference trajectory. We define the control variable as the string force density to make the governing equations of motion linear in the control variable. This allows the required string force density to be solved for linearly at each time step by solving a convex linear programming problem. The developed control law is implemented in simulation on several Class-1 and Class-k tensegrity structures, clearly showing the effectiveness of the developed ideas in modeling and control of tensegrity structures. Combining our results, we develop a novel robotic manipulator by using self-similar iterations to yield a structure that is both highly dexterous and lightweight, proving that the modeling and control framework can be used to design complex engineering structures.