Applying artificial potential fields to path planning for mobile robotics and to haptic rendering for minimally invasive surgery
Jing Ren · 2005
Artificial Potential Fields (APFs) are widely used in many areas of robotics, mainly due to their simplicity and computational efficiency, but they suffer from several inherent limitations and have many unresolved issues. The oscillation problem greatly limits the use of APFs for high speed mobile robots. In this thesis, we introduce Modified Newton's Method (MNM) into APFs and show that the oscillation problem can be eliminated or greatly reduced by using this technique. Another difficulty in applying APFs is the lack of a potential model that can accurately represent an arbitrary shape. Most of the existing potential models restrict the shape of objects to some particular geometries, which limits the application of the proposed method and can obstruct the goal or the path to the goal. This drawback becomes more pronounced when we apply APFs to haptic rendering for minimally invasive surgery. In this thesis, we propose a new potential model based on generalized sigmoid functions that can accurately represent a large class of shapes. New difficulties arise when we apply APFs to multi-robot path planning. One major challenge is to bridge the high-level task planning and the low-level path planning and integrate them into one framework. In this thesis, we propose a framework for a cooperative multi-robot team. By combining an intelligent agent architecture, a potential field-based hybrid navigation scheme and a built-in reconfiguration mechanism into one framework, the proposed architecture allows a robot team to allocate tasks among team members independent of human supervision and to accomplish a wide range of navigation tasks. Finally, we present a geometric transformation-based algorithm for avoiding moving obstacles. Our approach has low computational cost and is guaranteed to find the optimal solution if it exists. Algorithms are validated through simulation and experiments.