【报告摘要】Observer design for nonlinear systems remains a classical yet persistently challenging problem in both the control and robotics communities. Unlike linear systems, where observability is a structural property, nonlinear observability is inherently trajectory dependent. Furthermore, most existing observer constructive tools for nonlinear systems rely on restrictive structural assumptions. These requirements pose significant limitations in practice, particularly in robotics applications where system states often evolve on nonlinear manifolds. To address these challenges, we have collaborated with researchers to develop a radically new approach: parameter estimation-based observers (PEBO). By carefully designing dynamic extensions, this method transforms the estimation of system states into the estimation of constant quantities, significantly simplifying the problem and relaxing the excitation conditions required for nonlinear observer design. We have further extended the PEBO framework from Euclidean space to matrix Lie groups, demonstrating its effectiveness in solving a series of challenging estimation problems in robotics. In this talk, I will present a comprehensive tutorial on the recent progress of the PEBO approach and its applications in robotics.