Robust attitude control for NASA Astrobee robots operating in the ISS

Free-flying robots have been recently developed to operate on-board the International Space Station (ISS) as semi-autonomous robotic assistants. Free-flyer robots can be designed as modular base for integration of a wide range of hardware and software as monitoring and maintenance tools of ISS systems. Thus, the free-flyers are the ideal platform for manual observation of the ISS by ground control, autonomous sensor readings and surveying of ISS conditions, and human-robot interaction during long duration human missions. The NASA Astrobee project aims to develop a highly capable robot that can operate for long periods of time without crew supervision or operation. The goal of this research is to realize a model and a robust attitude controlsystem for the NASA Astrobee, which is a free-flyer equipped with a 3 degree of freedom (DoF) manipulator. Two subsystem, the Astrobee’s main body and the manipulator, have been modelled and controlled independently, considering the torques and the disturbances due to the manipulator motion. For the attitude control and stabilization of the main body, the twisting sliding mode controller (TW-SMC) and the back-stepping controller are proposed, while for the manipulator motion a first order sliding mode controller is proposed. Robustness and performance are analyzed to show the effectiveness of the proposed control system. First, the TW-SMC is designed to achieve a trade-off among control law flexibility, robustness and precision attitude control. Among robust control strategies, SMC are characterizes as low complexity, low computational and low cost control methods. Including the chattering attenuation introduced with the second order SMC and the hyperbolic tangent in the reaching law, the TW-SMC is a suitable approach for the Astrobee’s main body attitude control. The TW-SMC is compared with a back-stepping approach, which consist in an adaptive controller based on Lyapunov functions, that use an iterative algorithm for the control design. The controller outputs are torque taking into account attitude error, the arm torque and motion, and the inertia variation, leading the Astrobee to track the desired attitude. Controllers’ performance are evaluated and analyzed both in MATLAB/Simulink enviroment and with the NASA’s ROS/Gazebo Astrobee simulator. Simulink-ROS combined simulations have been run to test the arm motion stabilization, attitude changing and their combination. To show the robustness of the proposed control methodologies, simulations with variable body mass, links masses and end-effector mass have been run to test the controllers in off-design conditions.