Behaviors#
A behavior is the core abstraction of navground. It describes a generic local navigation algorithm that takes into account the local environment and its current ego-state (position, velocity, parameters, …) to compute a velocity command to move towards a target. The types of ego-state, command, and target are shared by all behaviors, while the local environment type is specific. For instance, navigation algorithms may represent the environment using images (like grid maps) or list of symbolic objects (like neighbors); therefore, we do not try to find a shared local environment abstractions.
In simulation, the behavior’s ego-state is updated from the agent’s state at each step, while
targets are updated by Tasks
local environments are updated by State Estimations.