| Robots increasingly resemble animals. For some, this is because they are designed to mimic animal features, but there is more to the story. Fundamentally, we seek for robots to achieve the same thing that animals do: to operate in the real world with intelligence, capable of movements which achieve environmentally relevant goals. They perform these movements using noisy sensors while the environment and even their own bodies undergo profound changes. Animals serve as the only existence proof of intelligent movement control for many domains, so it is natural for engineered systems to converge on the same behavioral and morphological solutions as those found by nature. Even if roboticists do not intentionally mimic biology, our engineered solutions often converge to resemble those found by nature. This thesis describes new methods for leveraging the insights of bio-logical control, rendering previously intractable problems, including those encountered by tendon-driven robotic hands, solvable. |
