Low-velocity tracking capability is a key performance of flight motion simulator (FMS), which is mainly affected by the nonlinear friction force. Though many compensation schemes with ad hoc friction models have been proposed, this paper deals with low-velocity control without friction model, since it is easy to be implemented in practice. Firstly, a nonlinear model of the FMS middle frame, which is driven by a hydraulic rotary actuator, is built. Noting that in the low velocity region, the unmodeled friction force is mainly characterized by a changing-slowly part, thus a simple adaptive law can be employed to learn this changing-slowly part and compensate it. To guarantee the boundedness of adaptation process, a discontinuous projection is utilized and then a robust scheme is proposed. The controller achieves a prescribed output tracking transient performance and final tracking accuracy in general while obtaining asymptotic output tracking in the absence of modeling errors. In addition, a saturated projection adaptive scheme is proposed to improve the globally learning capability when the velocity becomes large, which might make the previous proposed projection-based adaptive law be unstable. Theoretical and extensive experimental results are obtained to verify the high-performance nature of the proposed adaptive robust control strategy. Low-velocity tracking capability is a key performance of flight motion simulator (FMS), which is mainly affected by the nonlinear friction force. While many paper folding with ad hoc friction models have been proposed, this paper deals with low-velocity control without friction First, a nonlinear model of the FMS middle frame, which is driven by a hydraulic rotary actuator, is built. Noting that in the low velocity region, the unmodeled friction force is substantial by a changing-slowly part, thus a simple adaptive law can be employed to learn thisching-slowly part and compensate it. To guarantee the boundedness of adaptation process, a discontinuous projection is utilized and then a robust scheme is proposed. a prescribed output tracking transient performance and final tracking accuracy in general while driving asymptotic output tracking in the absence of modeling errors. In additio n, a saturated projection adaptive scheme is proposed to improve the globally learning capability when the velocity becomes large, which might make make the previous proposed projection-based adaptive law be unstable. Theoretical and extensive experimental results are obtained to verify the high-performance nature of the proposed adaptive robust control strategy.