We describe a new approach to the problem of motion planning for mobil
e robots on natural, rough terrain, Our approach computes a multiresol
ution representation of the terrain using wavelets, and hierarchically
plans the path through sections which are well approximated on coarse
r levels and relatively smooth, Unlike most methods, the hierarchical
approximation errors are used explicitly in a cost function to disting
uish preferred terrain sections, The error is computed using the corre
sponding wavelet coefficients. We also propose a new nonscalar path co
st measure based on the sorted terrain costs along the path, This meas
ure can be incorporated into standard global path search algorithms an
d yields paths which avoid high cost terrain areas when possible, Addi
tional constraints for specific robots can be integrated into this app
roach for efficient hierarchical motion planning on rough terrain, We
present the algorithms and experimental results for real terrain data.