Manipulator path planning in unknown environments using model based planners: Conversion criteria and IPA sensor implementation

Path planning in unknown environments such as Mars or underwater are not only challenging but also daunting tasks. While significant advances are made for mobile platforms, manipulator motion planning in unknown environments still falls short of full fledged solutions due to the uncertainty and complexity of higher order configuration spaces. Most feasible solutions proposed so far are either modeling unknown environments in realtime or applying randomized model based planners with sensitive skin type sensors. Although solving unknown environment planning problems with randomized model based approaches seems very promising, no systematic study has been reported as to how to adopt model based planners into the sensor based problem domain effectively. In this paper, we study the adoption issues of model based planners to tackle sensor based planning problems, and hence provide conversion guidelines from one to other problem domain. In addition, we discuss the range effect of the sensitive skin type sensor, IPA, for on-line planning in unknown environments. Some experimental results in an unknown environment are presented in the paper as well.