We proposed an algorithm for predicting friction
coefficient from visual information for mobile robot since
coefficient of friction is very important in driving on road and
traversing over obstacle. Our algorithm is based on terrain
classification for visual image. To predict friction coefficient
from given image, we divide an image into homogeneous
regions which have same material composition. The proposed
method, non-contacting approach, has advantage over other
methods that extract material characteristic of road by sensors
contacting road surface. Obtaining information about friction
coefficient before entering such terrain can be very useful for
path planning and avoiding slippery areas. And we form a
group of each terrain type. So, when new terrain is entered into
a system, the data of new terrain are classified into each group.
By grouping each terrain to use the same regression coefficients,
we can reduce the amount of processing time. The proposed
method will be verified by real outdoor environment with real
vehicles.