SummaryThe 12-tetrahedral robot is an addressable reconfigurable technology (ART)-based variable geometry truss mechanism with twenty-six extensible struts and nine nodes arranged in a tetrahedral mesh. The robot has the capability of reconfiguring shape and dimension for environment sensing requirements, which makes it suitable for space exploration and environmental perception. In this paper, we have derived a dynamics model and presented a decentralized adaptive controller for a 12-tetrahedral robot. First, the robot is divided into the node and the strut subsystems, and the kinetic and the potential energy are calculated for the two subsystems. Then, the dynamics model is achieved by applying the Lagrangian formalism on the total energy of the robot. Since the dynamics is too complicated for implementing model-based controllers, a two-layer controller is presented to control the robot, in which the planning layer determines gait and trajectory of the robot, and the executive layer adopts the decentralized adaptive control strategy and consists of twenty-six strut controllers. Each strut controller regulates the movement of the corresponding strut without information exchange with other struts. Co-simulations based on ADAMS and Matlab have been conducted to verify the feasibility and effectiveness of the proposed controller.