.png){kind=icon}
.png){kind=icon}
Return
Guided Sampling-Based Motion Planning Algorithm for Dynamic and Obstacle Cluttered Environments
.png){kind=icon}
PRE
AI
DOI：10.2514/1.G009138.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper presents a guided sampling-based algorithm for motion planning that is designed to allow safe navigation in dynamic and cluttered environments. Sampling-based algorithms address motion planning problems by sampling random trajectories in the configuration space and evaluating their performance. However, their effectiveness is heavily dependent on the predefined sampling distributions. In dynamic environments, predefined sampling distributions often fail to adapt to real-time changes, leading to safety violations as environmental conditions deviate from initial assumptions. To address this issue, this paper proposes a guided sampling algorithm based on the model predictive path integral (MPPI) control algorithm, combining with the merits of the rapidly exploring random trees (RRT) algorithm and the control barrier function (CBF) method. The proposed RRT-CBF-guided MPPI (RC-MPPI) algorithm leverages both RRT and CBF to autonomously update the mean and covariance of the sampling distributions, utilizing current environmental information, including obstacle locations and safety constraints. The RRT initially guides the determination of the mean and covariance by considering obstacle avoidance and the target position. Subsequently, the CBF refines this preliminary determination to ensure adherence to safety constraints. The proposed RC-MPPI improves sample efficiency and enhances safety by integrating the guided sampling into the planning process. In particular, this study calculates an upper bound for the required sample size in MPPI and uses this bound to demonstrate how RC-MPPI improves sample efficiency. To assess the effectiveness of RC-MPPI, numerical simulations are conducted using both an unmanned ground vehicle model and an unmanned aerial vehicle model. The results demonstrate significant improvements in sample efficiency and safety compared to conventional MPPI.
Keywords:
Robot Kinematics
Unmanned Aerial Vehicle
Nonlinear Model Predictive Control
Numerical Simulation
Hamilton Jacobi Bellman (HJB) Equations
Optimization Algorithm
Journal
J
IF:
2.8
Papers:
192
Citations:
1.3W
View more organizations