Indoor navigation for mobile robots using predictive fields

A predictive field based path planner for mobile robot navigation in indoor areas is described. Predictive fields are used for incorporating moving obstacle information into the navigation function framework. Navigation functions have been limited by geometric restrictions for robot workspaces and cannot easily be used in everyday environments. A methodical description of indoor workspaces such that they follow the constraints of a navigation function based path planner is proposed. Typical use of navigation function based path planners as a gradient based control input requires large scaling factors in practical use. A direction based control input, which eliminates the scaling factor altogether, is proposed and its stability and convergence is proved using Lyapunov-type analysis. The proposed algorithm improves the practicability of navigation functions and makes it possible to envisage a person following mobile robot operating in indoor environments using predictive field navigation.