Optimization and Kinematics Prediction Modeling of a Spiral-Tensegrity Continuum Robot

In confined space operations, traditional rigid robotic arms are too bulky and inflexible, while flexible robotic arms lack sufficient load capacity. Tensegrity structures offer an effective solution due to their lightweight nature and high structural efficiency. Building on this, this article innovatively introduces spiral curves to design a spiral tensegrity continuum robot. The spiral curves provide large-range agility, and the tensegrity structure ensures lightweight construction with adjustable dynamic stiffness. To overcome traditional design methods’ difficulty in balancing multiple performance metrics, this article proposes a rapid design framework using intelligent optimization algorithms, significantly shortening the design cycle and improving overall performance. To improve kinematic prediction under cable friction, spring deformation, and limited training data, a proper orthogonal decomposition (POD)-backpropagation neural network (BPNN) model is developed to predict the robot shape and end-effector position. This model effectively predicts the robot’s overall shape and end effector position. Prototype experiments validated the effectiveness and accuracy of these models, achieving a kinematic modeling precision of 98%. This article supports expanding the applications of continuum robots in complex multitask scenarios.