Abstract: Direct brain-controlled robot interface (BCRI) is a new type human-robot interface which is an important research and development direction for brain-machine interface (BMI)/brain-computer interface (BCI) in the robot control field. Many experimental researches and developments for BCRI were reported by Nature, Science and other important international journals and it has become an international frontier research hotspot. The paper mainly discussed the control strategies for BCRI, the adaptation and fusion between BMI/BCI module and robot multilayer control architecture module, the adaptive classification algorithms for brain signal used in BCRI and the trilateral adaptation among human, BMI/BCI module and robot control system. The current situation and limitation for BCRI and some important problems faced by BCRI were analyzed and the further research ideas and directions were also pointed out.