In this section, we'll outline two types of AV system from a software architecture perspective. The first type uses sequential architecture with multiple components, as illustrated in the following diagram:
The components of an AV system
The system resembles the reinforcement-learning framework we briefly discussed in Chapter 10, Meta Learning. We have a feedback loop where the environment (either the physical world or a simulation) provides the agent (vehicle) with its current state. In turn, the agent decides on its new trajectory, the environment reacts to it, and so on. Let's start with the environment-perception subsystem, which has the following modules (we'll discuss them in more detail in the following sections):
- Sensors: Physical devices, such as cameras and radars.
- Localization: Determines the exact position of the vehicle...
