Page 276 - Introduction to AI Robotics

7.2 Attributes of the Hybrid Paradigm
7.2
chical flavor with global world models, especially Saphira 77 and TCA. 131 259
Regardless of the bottom-up or top-down inspiration for including non-
behavioral intelligence, architectures which use reactive behaviors, but also
incorporate planning, are now referred to as being part of the Hybrid De-
liberative/Reactive Paradigm. At first, Hybrids were viewed as an artifact
of research, without any real merit for robotic implementations. Some re-
searchers went so far as to recommend that if a robot was being designed to
operate in an unstructured environment, the designer should use the Reac-
tive Paradigm. If the task was to be performed in a knowledge-rich environ-
ment, easy to model, then the Hierarchical Paradigm was preferable, because
the software could be engineered specifically for the mission. Hybrids were
believed to be the worst of both worlds, saddling the fast execution times of
reactivity with the difficulties in developing hierarchical models.
The current thinking in the robotics community is that Hybrids are the
best general architectural solution for several reasons. First, the use of asyn-
chronous processing techniques (multi-tasking, threads, etc.) allow deliber-
ative functions to execute independently of reactive behaviors. A planner
can be slowly computing the next goal for a robot to navigate to, while the
robot is reactively navigating toward its current goal with fast update rates.
Second, good software modularity allows subsystems or objects in Hybrid
architectures to be mixed and matched for specific applications. Applica-
tions which favor purely reactive behaviors can implement just the subset of
the architecture for behaviors, while more cognitively challenging domains
can use the entire architecture.
Attributes of the Hybrid Paradigm
The organization of a Hybrid Deliberative/Reactive system can be described
as: PLAN,then SENSE-ACT. It is shown in Fig. 7.1. The PLAN box includes
all deliberation and global world modeling, not just task or path planning.
The robot would first plan how to accomplish a mission (using a global world
model) or a task, then instantiate or turn on a set of behaviors (SENSE-ACT)
to execute the plan (or a portion of the plan). The behaviors would execute
until the plan was completed, then the planner would generate a new set of
behaviors, and so on.
The idea of PLAN,then SENSE-ACT evolved from two assumptions of
the Hybrid Paradigm. First, planning covers a long time horizon and re-
quires global knowledge, so it should be decoupled from real-time execution