When a droplet is placed on a surface with a temperature higher than Leidenfrost temperature, it will be levitated on its own vapor cushion, which makes the droplet have amazing mobility. Generally, the vapor flow under the droplet is corrected by constructing asymmetry micro/nano textured surfaces to realize the self-propulsion of the droplet. However, the control of droplet dynamics becomes uncertain due to the complicated interaction among liquid-vapor-solid phases, and the direction of droplet motion and droplet transport velocity (10－40 cm/s) have limitations. Leidenfrost heat transfer surface and impact surface are constructed in this experiment. When the surface of Leidenfrost heat transfer for droplet levitation contacts the droplet which the sufficient energy flows towards, the rough ring surface acts as an igniter. When a warm Leidenfrsot droplet (fuel) contacts skirt ring (igniter), abundant micro/nano cavities of rough skirt ring not only generate additional radiation heat flux towards droplet but also provide nucleation sites to trigger explosive boiling on a ～10 ms time scale. The thrust force F_th generated by periodic explosive boiling realizes the self-propulsion of droplets. In the initial stage of droplet motion, the inertial force F_i is dominant, and the droplet impact is mostly specular reflection, and the droplet trajectory is chaotic. With the decrease of droplet diameter, the pulsed thrust F_th is dominant, and the droplet trajectory passes through the center of the Leidenfrost heat transfer surface. Our experimental results show that the droplet passes through the center of the Leidenfrost heat transfer surface in a wide diameter range (D = 0.671－1.576 mm). For the last (150^th) collision with the rough ring, the curved trajectory exists instead of line trajectory with D \sim 0.105 mm. For a very small droplet, the drag force F_d is dominant, which prevents the droplet from continuing to move. The droplet is followed by sessile droplet evaporation until it totally disappears. At the same time, the droplet transport velocity is as high as 68.91 cm/s, which has not been realized before. The present work provides a new method to manipulate droplet motion at a high temperature. Our Leidenfrost system is simple, cost-effective and lasts long-term operation because it does not depend on complicated micro/nano fabrications, which is expected to be used in microfluidic and heat transfer two-phase systems.