This study presents a hybrid trajectory optimization method that generates a collision-free smooth trajectory for autonomous mobile robots. The hybrid method combines sampling-based model predictive path integral (MPPI) control and gradient-based interior-point differential dynamic programming (IPDDP) exploiting their advantages of exploration and smoothing. The proposed method, called MPPI-IPDDP, consists of three steps. The first step generates a coarse trajectory by MPPI control, the second step constructs a collision-free convex corridor, and the third step smooths the coarse trajectory by IPDDP using the collision-free convex corridor computed in the second step. For demonstration, the proposed algorithm was applied to trajectory optimization for differential-driving wheeled mobile robots and point-mass quadrotors. A supplementary video of the simulations can be found at https://youtu.be/-oUAt5sd9Bk.
翻译:本研究提出了一种混合轨迹优化方法,为自主移动机器人创造无碰撞平稳轨道。混合方法结合了基于取样的模型预测路径集成(MPPI)控制和基于梯度的内点差异动态编程(IPDDP),利用探索和平滑的优势。拟议方法称为MPPI-IPDDP,由三个步骤组成。第一步通过移动电话伙伴关系控制产生粗糙的轨迹,第二步建立无碰撞锥形走廊,第三步通过第二步计算的无碰撞锥形走廊平滑了IPDDP的粗体轨迹。为示范,拟议的算法被用于对不同驾驶轮式移动机器人和点质量孔式机器人的轨迹优化。模拟的辅助视频可在https://youtu.be/-oUAt5sd9Bk找到。