We present techniques for constructing approximate stochastic models of complicated dynamical systems for applications in interactive computer graphics. The models are designed to produce realistic interaction at low cost. We describe two kinds of stochastic models: continuous state (ARX) models and discrete state (Markov chains) models. System identi cation techniques are used for learning the input-output dynamics automatically, from either measurements of a real system or from an accurate simulation. The synthesis of behavior in this manner is several orders of magnitude faster than physical simulation.We demonstrate the techniques with two examples: (1) the dynamics of candle ame in the wind, modeled using data from a real candle and (2) the motion of a falling leaf, modeled using data from a complex simulation. We have implemented an interactive Java program which demonstrates real-time interaction with a realistically behaving simulation of a cartoon candle ame. The user makes the ame animation icker by blowing into a microphone.

We describe a new approach to the problem of motion planning for mobile robots on natural rough terrain. Our approach computes a multiresolution representation of the terrain using wavelets, and hierarchically plans the path through sections which are well approximated on coarser levels and relatively smooth. Unlike most methods, the hierarchical approximation errors are used explicitly in a cost function to distinguish preferred terrain sections. The error is computed using the corresponding wavelet coefficients. We also propose a new nonscalar path cost measure based on the sorted terrain costs along the path. This measure can be incorporated into standard global path search algorithms and yields paths which avoid high cost terrain areas when possible. Additional constraints for specific robots can be integrated into this approach for efficient hierarchical motion planning on rough terrain. We present the algorithms and experimental results for real terrain data.

We describe a new approach to the problem of motion planning for mobile robots on natural, nonhomogenous terrain. Our approach computes a multiresolution representation of the terrain using wavelets, and hierarchically plans the path through sections which are well approximated on coarser levels and relatively smooth. Unlike most methods, the hierarchical approximation errors are used explicitly in a cost function to distinguish preferred terrain sections. The error is computed using the corresponding wavelet coefficients. The path planning algorithm uses a new nonscalar path cost measure based on the sorted terrain costs along the path. This measure can be incorporated into standard global path search algorithms and yields intuitively good paths. Additional constraints for specific robots can be integrated into this approach for efficient hierarchical motion planning on rough terrain. We present experimental results for real terrain data.