UC Berkeley

This dissertation presents the design, analysis and testing of various actuated modular spherical tensegrity robots for co-robotic and space exploration applications. Tensegrity structures are of interest in the field of soft robotics due to their flexible and robust nature. They have the ability to passively distribute forces globally providing shock protection from unexpected impact forces. This feature makes them a robust mobile platform suitable for uneven terrain and unpredictable environments in which traditional robots struggle. Robots built from tensegrity structures, which are composed of pure tensile and compression elements, have many potential benefits including high robustness through redundancy, many degrees of freedom in movement and flexible design. However, to fully take advantage of these properties a significant fraction of the tensile elements should be actuated, leading to a potential increase in complexity, messy cable and power routing systems and increased design difficulty.

The first part of this dissertation presents an elegant solution to a fully actuated tensegrity robot: the TT-3 (version three) tensegrity robot was developed at University of California Berkeley, in collaboration with NASA Ames. The TT-3 is a lightweight, low cost, modular, and rapidly prototyped spherical tensegrity robot. This dissertation describes in detail the novel design mechanisms, architecture and simulations of TT-3, the first untethered, fully actuated cable-driven six-bar tensegrity spherical robot ever built and tested for mobility. Furthermore, this dissertation discusses the controls and preliminary testing performed to observe TT-3’s system behavior and performance and was evaluated against previous models of tensegrity robots developed at University of California at Berkeley and elsewhere.

The second part of this dissertation will present a new platform for prototyping spherical tensegrity robots that significantly reduces the time required for manufacturing and assembly. This simplified tensegrity system design allows for more scientific experiments to be performed in less time. This work describes the design architecture of the TT-4mini, an example of a robot that uses this prototyping platform. In order to demonstrate the platform’s use for scientific experiments, the TT-4mini was shown to achieve uphill climbing, which has not been performed by any other spherical tensegrity robot in hardware. This dissertation discusses preliminary observations on the system’s performance in uphill climbing from simulations and testing, including evidence of climbing surfaces with an incline up to 13 degrees. Furthermore, this new prototyping platform demonstrated the ability to create three complex 12-bar tensegrity structures with simplified procedures.

Lastly, the dissertation presents the improved tensegrity robot using the rod-centered architecture: the TT-4 (version four). It is a larger robot; its rod length is 1 meter compared to the 0.65 meter rod length of TT-3. The goal of the TT-4 robot prototype is twofold: to evaluate both the design improvements and the interaction of the robot with a center payload. Because the size of the robot has increased, the volume at the center of the robot available to hold payload has also increased. This allows researchers to further study the configuration, method of attachment, and the dynamics of a payload.

These robots are based on a ball-shaped six-bar tensegrity structure and features a unique modular rod-centered distributed actuation and control architecture. This revolutionary new architecture has been demonstrated in both software and hardware testing to increase the performance of tensegrity robots and has the potential to be extensible to a wide range of tensegrity configurations.