Biomimetic adaptive control of a deployable tensegrity structure

Biomimetic behavior includes aspects such as learning from previous experience, self-diagnosis, and adaptation. This thesis describes control methodologies that are essential to development towards biomimetic behavior of a complex deployable structure. Simulations of the structure are improved from previous work to include friction effects of cables sliding over joints. Despite improved simulations, testing shows that significant uncertainties in the behavior of the structure remain. Therefore, control is not effective with predetermined actuation movements. Special methodologies for feedback control using simulations and measurements are required.

A near-full-scale deployable tensegrity structure is used to test methodologies. Active control methods are proposed for deployment, midspan connection of both halves of the structure, and self-stress. This thesis presents a method using feedback to compare measurements and simulations to modify control commands. Since collision of elements is possible in the folded state and produce undesirable bending stresses, a path-planning algorithm is implemented for the first stage of deployment. Error in nodal positions at midspan is successfully reduced through the use of the path-planning method and deployment time is significantly reduced compared with previous work. Lastly, algorithms for self-stress, involving penalty and rejection constraints on element stress, are useful for correcting nodal positions after deployment.

Damage is detected in this thesis using vibration measurements. The method uses dynamic behavior of the structure to determine whether or not the structure is damaged. Using parameters of the structure and a set of candidate locations for the damaged element, candidates are successively excluded until few candidates remain, successfully including the true location of damage.

Adaptation and learning are demonstrated by mitigation methods after damage and in-service loading (such as pedestrians). Active control is useful to manipulate the shape of the tensegrity structure to reduce the member stresses and vertical downwards displacement caused by a damaged element. Though the response improves the condition of the structure to respect the serviceability limit for vertical downwards displacement, the tensegrity structure cannot be fully restored to the design configuration.

Since correction of end-node coordinates can be grouped by the direction of correction and resulting cable-length changes, case-based reasoning is useful to reduce time of execution and to reduce unnecessary cable-length changes. Single pedestrian and crowd loading configuration is applied analytically and experimentally to the tensegrity structure. Application of mitigation techniques is useful beyond serviceability thresholds for a moving load used to simulate in-service loading.

The research question of this thesis: "Is it feasible for a deployable tensegrity structure to improve movement and service performance through behavior biomimetics?". The answer is yes. This work presents methods inspired by those observed in nature for efficient movement that is generalizable for future deployable structures.