Nonlinear dynamics, feedback, decision, control, estimation, learning: these processes are everywhere in the natural and engineered systems. Examples include:

• Planning and Control: Robots and autonomous vehicles must sense the world around them and move safely through that world. In planning such movements, there are usually far too many possibilities to check them all, so we need clever search methods. This is the problem of planning. Also, the models used in planning never perfectly match reality, so we need to ensure that the desired motion is actually achieved. This is the problem of control.
• System identification and learning: Generating predictive mathematical/computational models from recorded data is a very common requirement in engineering and natural sciences, especially when first-principles models are unavailable or too complex. Major challenges when modelling dynamic systems include dealing with uncertainty, ensuring model stability, and identifying long-term dependencies between inputs and outputs.
• Dynamic walking robots: these are a class of humanoid bipedal robots that can walk very efficiently by taking advantage of natural physical dynamics. They are often underactuated, iIndeed, some can be completely passive. While this concept is fascinating and holds much promise (and human walking is an existence proof that it is feasible), there are many challenges for systematic design. The main challenges come from the complexity of the dynamics, which are highly nonlinear and involve impact events.