Chapter 2 Simulation and Numerical

Robotics is, among other things, the study of complex hardware-software systems. In a complex system, many components interact, giving rise to emergent behaviors that can be hard to explain, anticipate, or design. Rather than just evaluating the system as a whole, we would often prefer first to evaluate components of the system individually. For instance, robotic mechanism design incorporates simulation to help select and predict the performance of a mechanism before fabrication. Similarly, software is designed to operate in harmony with hardware that provides specific interface guarantees, but we would like to evaluate software apart from the mechanisms so that we’re not debugging hardware and software at the same time. During the course of software development, we would like to have assurances that software modules will meet their own specification guarantees apart from the hardware. In robotics, we use simulators in this and many other capacities. Uses occur both offline for design and testing purposes and online for planning and control purposes. Offline simulators often (but not always) provide a visualization of the simulated system, and they also provide access to raw data and analysis. They can be interactive or can be driven forward by the simulated passage of time. In this chapter, we examine how and why simulators are used for robotics. In addition, we delve briefly into the numerical methods that make them work.