Ge Shuzhi Sam, Lai Xuecheng, Al Mamun Abdullah
Electrical and Computer Engineering Department, National University of Singapore, Singapore 117576.
IEEE Trans Syst Man Cybern B Cybern. 2005 Apr;35(2):240-54. doi: 10.1109/tsmcb.2004.842368.
In this paper, an Instant Goal approach is proposed for collision-free boundary following of obstacles of arbitrary shape and globally convergent path planning in unknown environments. Firstly, for effective knowledge representation and manipulation, a vector representation is presented, which not only saves much space but also conforms to the physical properties of range sensors. Secondly, the concept of Instant Goals is introduced enabling the robot to perform boundary following in a "natural" human-like manner, with additional measures taken to ensure that the robot is moving "forward" along the boundary, even if the obstacle is of arbitrary shape and disturbing obstacles are present. Collision checking is performed simultaneously and, when needed, collision avoidance is efficiently incorporated in. Based on the approach of boundary following, a realistic sensor-based path planner with global convergence property is designed for the robot capable of acquiring discrete and noisy range data. Realistic simulation experiments validate the effectiveness of the proposed approaches.
本文提出了一种即时目标方法，用于在未知环境中对任意形状的障碍物进行无碰撞边界跟踪和全局收敛路径规划。首先，为了实现有效的知识表示和操作，提出了一种向量表示法，它不仅节省了大量空间，而且符合距离传感器的物理特性。其次，引入了即时目标的概念，使机器人能够以类似人类的“自然”方式进行边界跟踪，并采取额外措施确保机器人沿边界“向前”移动，即使障碍物形状任意且存在干扰障碍物。同时进行碰撞检查，并在需要时有效地纳入碰撞避免措施。基于边界跟踪方法，为能够获取离散和噪声距离数据的机器人设计了一种具有全局收敛特性的基于传感器的现实路径规划器。现实的仿真实验验证了所提方法的有效性。
