Bayesian Optimization-Based Ideal Landing Planning for Ramp Jump of Single-Track Two-Wheeled Robots

For the ramp jump of Single-Track Two-Wheeled (STTW) robots, determining an appropriate landing attitude to ensure a smooth landing is a challenging issue. In this paper, we first conduct a comprehensive analysis of the governing dynamics of the ramp jump, including the gyroscopic effect caused by steering during the flight phase and the impact forces upon landing in the landing phase. Subsequently, we define an ideal landing attitude from which the system naturally returns to its equilibrium point without requiring active control. Building on the dynamics and constraints of the ideal landing, we propose an Ideal Landing-Based Planning (ILBP) method to achieve these ideal landing attitudes. To mitigate the method's dependence on precise modeling, we employ Bayesian Opti-mization (BO) to learn the parameters of the ILBP, enabling the robot to withstand larger disturbances than those managed by the original ILBP. Finally, we validate the effectiveness of the proposed approach within the MuJoCo simulation environment, where the robot effectively masters landing control in fewer than ten trials under parameter perturbations.