Autonomous Navigation for Animats
The animat approach would physically simulate the body and give it the ability to sense the environment like animals do. Instead of passing data directly to the animats, active perception allows them to select the information they are interested in, which increases efficiency in dynamic situations.
This resolves most of the issues (faced by bots) by providing characters with fresh information about the local surroundings (see Figure 6.2). Dynamically moving obstacles are trivial to identify when the world is perceived continuously, and the characters act upon these perceptions.
Figure 6.2. Using whisker-like sensors to perceive the environment. Areas of empty space are detected, as well as obstacles within range.
When we get to the creation of movement in Chapter 10, "Steering Behaviors for Obstacle Avoidance," and Chapter 12, "Synthesizing Movement with Rule-Based Systems," we'll quickly realize that this approach is much easier too! Indeed, the problem can be interpreted as a reactive task—one of the many benefits of embodiment. This makes it particularly efficient, because no planning is required.
However, although many aspects of navigation are improved upon by this nouvelle AI approach, some aspects are not up to standard of the typical planning approaches. Notably, the reactive approach has trouble reaching particular points in space; it's good at the low-level details, but the higher levels can lack purpose.