We investigate how natural dynamics can yield stable, agile, and energy efficient robotic systems. Firstly, we cover a design with a single passive rolling element to stabilize frontal plane dynamics for a 3D biped walking across a range of forward velocities and/or step lengths. We examine aspects of the non-linear dynamics that contribute to the energy efficiency and stability of the system through simulations. Secondly, we examine switching controllers that allow for agile foothold selection in 5-link walkers. We leverage dynamic programming and discretization of the reachable space to walk across intermittent footholds. We utilize our meshing techniques to quantify stability and agility of these switching controllers. Finally, we provide experimental data on the effect of extra mass and power on humans at a variety of locations and forward velocities. This allows robot and exoskeleton designers to optimize for energy performance by understanding mass placements and power densities required for high performing legged locomotion. Finally, we present experimental data for an exoskeleton capable of assisting across running and walking speeds.