<p>High-precision force control in robotic grinding and polishing is the pivotal technology for enhancing the surface quality and processing consistency of components with complex geometries. This paper provides a clear theoretical framework and a technological roadmap for the field. It systematically reviews the three major paradigms of active, passive, and hybrid active–passive compliant control, revealing an evolutionary logic driven by their performance contradictions. Active control is limited by the robot’s response bandwidth, while passive control lacks task adaptability. Consequently, hybrid active–passive compliant control fuses the advantages of both paradigms through task decoupling and frequency separation. Such an integrated strategy has emerged as a promising pathway. This review further distills the core challenges hindering industrial adoption, including the trade-off between performance and cost, the gap between theoretical models and physical reality, and the lack of generalized solutions. Finally, the paper provides an outlook, positing that the field is experiencing a paradigm shift from precise force control to intelligent process control. Future research will converge on the establishment of standardized benchmarking ecosystems, physics-informed sim-to-real transfer, and foundation model-driven embodied intelligence, to ultimately achieve autonomous and high-quality surface finishing.</p>