Cooperative and robust object manipulation by multiple robots via linear estimated state feedback

In modern industrial automation, the deployment of multiple robotic manipulators for cooperative operations has become increasingly common, offering enhanced system flexibility and responsiveness. However, as the number of manipulators increases, the system dynamics become substantially more nonlinear and complex, giving rise to unmodeled dynamics and various sources of uncertainty. Moreover, external disturbances can further degrade control performance, while the lack of a comprehensive sensing infrastructure may result in incomplete state information. To address these challenges, this study proposes a robust, model-independent control framework based on function approximation techniques. The methodology leverages linear differential equations with unknown coefficients to capture the aggregated system uncertainties under the assumption that such uncertainties can be effectively described using this structure. The approximation capability of the proposed model is then justified via the Stone-Weierstrass theorem, establishing the role of linear differential equations as universal approximators. Notably, the control strategy does not require velocity measurements, thereby simplifying its practical implementation. Stability analysis based on Lyapunov’s direct method ensures that tracking errors remain uniformly ultimately bounded. The controller is validated within a dual-arm cooperative manipulation scenario involving a rigid object, and its performance is benchmarked against three contemporary approximation-based control methods. Simulation results confirm the efficacy and robustness of the proposed approach.