Abstract: To address the challenges of low intelligence, poor process coordination, and delayed abnormal condition regulation in solid backfilling working faces, this study aims to achieve full-process autonomous decision-making and adaptive control for filling support robots, advancing green and safe coal mining technologies. An intelligent “group collaboration-process self-driving-abnormal regulation” solution is proposed. First, a flexible full-process model defines spatiotemporal constraints and transition conditions for five core processes—support, moving frame, pushing conveyor, unloading, and tamping—using finite state machines. This model covers 12 manipulator actions to address adaptability challenges under geological mutations, including inclination changes of ±30° and roof collapse. Second, an abnormal critical equation system is established through seven derived equations for frame alignment, pushing timing, unloading interference, and guard plate--shearer collision, overcoming single-machine control limitations. Finally, ADAMS/Simulink co-simulation validates algorithm robustness in complex scenarios such as sudden inclination shifts, roof collapse, and pitching mining. Key outcomes include: Moving frame control achieves a response time ≤1 s with support height error <2.5%; The system dynamically identifies critical angles for unloading and tamping interference under actual working conditions and enables precise regulation. For example, the critical tamping angle for horizontal unloading interference increased from 30° to 45°, and for pitching unloading interference (pitching mining and upward filling) increased from 45° to 70°. Tamping arm stroke control accuracy reached 98.3%, with horizontal interference condition regulation completed in only 3.1 s; Tamping anomaly avoidance demonstrated 100% recognition of bottom interference (critical stroke: 976 mm, corresponding to Condition II) and outer interference (critical stroke: 1021 mm, corresponding to Condition III), with full post-adjustment task completion. The proposed full-process self-driving control mechanism enables precise regulation of key parameters like support height, manipulator stroke, and tamping angle under complex geological conditions. This breakthrough overcomes technical bottlenecks in multi-machine coordination and abnormal condition regulation, providing foundational theories and technical support for intelligent backfilling faces.