Fast diffeomorphic matching to learn globally asymptotically stable nonlinear dynamical systems
Abstract
We propose a new diffeomorphic matching algorithm and use it to learn nonlinear dynamical systems with the guarantee that the learned systems have global asymptotic stability. For a given set of demonstration trajectories, we compute a diffeomorphism that maps forward orbits of a reference stable time-invariant system onto the demonstrations, thereby deforming the whole reference system into one that reproduces the demonstrations, and is still stable.
Origin : Files produced by the author(s)
Loading...