Over the past 10 years there has been an increasing interest in robotics to develop algorithms for stochastic control applicable to robotic and autonomous systems. Among other forms of stochastic control the path integral optimal control formulation provides a mathematically sound methodology for developing optimal control algorithms based on stochastic sampling of trajectories. In this talk I will present results in the area of sampling based control that go beyond the classical formulation of the path integral control framework. These results rely on information theoretic dualities and generalizations of the so called Feynman-Kac lemma. In addition I will present alternative methodologies to sampling based stochastic control using concepts from stochastic mechanics, polynomial chaos theory and non-parametric regression. Finally I will discuss various applications to autonomous systems in robotics and aerospace engineering.