Modeling dynamic uncertainty in robot motions

A method for modeling uncertainties that exist in a robotic system, based on stochastic differential equations, is presented. The use of such a model permits the capture in an analytical structure of the ability to properly express uncertainty within the motion descriptions and the dynamic, changing nature of the task and its constraints. With respect to the dynamic nature of robotic motion tasks, the model of the environment uncertainty proposed is dynamic rather than static. The amount of knowledge about the environment is allowed to change as the robot moves. These results suggest that computational models traditionally found in the lower levels in robot systems may have application in the upper planning levels as well. Some experimental results using the model are presented.<
>