Aerospace and Electronic Systems - March 2019 - 46

ARIES: An Autonomous Controller For Multirobot Cooperation

Figure 5.
Example of an effects file for a cooperative hybrid Rover-UAV
example.

token for every temporal action defined in the PDDL
domain file. That is, this file declares the effect and
parameters of every temporal action as a statement
with the format (StateVariable_Action M StateVariable_Predicate), where Predicate is the effect of
Action and M is the number of parameters that
PDDL-Lib has to assign to StateVariable_Predicate
from StateVariable_Action. Figure 5 shows an example of an effects file. For example, the UGVBase_At is
the effect of UGVBase_GoingTo, and it captures the
last two parameters of this last predicate, i.e., the x2
and y2 shown in Figure 4.
Once the sequence of tokens has been built for every
state variable, PDDL-Lib needs to get the duration of
every token to complete the timeline of every state variable. In this way, PDDL-Lib captures this duration from
the actions plan computed by the PDDL Planner (see Section "DEMONSTRATION."

ARIES CONTROLLER
The objective of the Autonomous coopeRatIve Execution
System (ARIES) is to provide an easily adjustable controller
for deploying cooperative heterogeneous robot teams. In
this way, we have implemented a multiagent architecture
based on the T-REX system and supported on standardized
technologies for enabling compatibility with other robotic

systems (flexible), adaptability to different problem domains
(scalable), and to be qualified to respond to some robot failures (fault tolerant). ARIES follows a hierarchical scheme
using the hybrid leader-follower approach in which the
leader centralizes the cooperative deliberation process of the
whole robot team, and then, it coordinates the execution of
each follower. The follower(s) execute(s) the goals received
from the leader and reply to it, as well as monitor(s) and
repair(s) (if needed) their status in order to be fully operative
for the team. The ARIES architectural concept is illustrated
in Figure 6(a).
ARIES represents the effort of bringing T-REX and
the leader-follower scheme together by splitting ARIES
in hierarchical agents, where each T-REX agent is executed on its own robotic platform. This effort results in
two types of T-REX agents: the Leader Agent and the
Follower Agent. Following the T-REX design criteria,
both Leader and Follower Agents have been built under
hierarchical layers. At the Leader Agent's top level, a
deliberative layer provides the cooperative planning capabilities. At the Follower Agent's top level, a deliberative
layer provides planning capabilities for supporting faulttolerant behaviors, expressed as series of rules that trigger
reactive behavior to overcome nonnominal situations.
Then, at middle and bottom levels, both agents have a
cooperative layer for synchronizing tasks execution and
an executive layer for execution control and monitoring.
Therefore, for a given plan provided by the deliberative
layer, the Leader Agent has to publish the mission goals
to the executive layer. If the target is a Follower Agent,
the cooperative layer forwards the goal to the executive
layer of the corresponding Follower. Then, the executive
layer is in charge of the action execution and monitoring.
This is the execution cycle followed in ARIES as shown
in Figure 6(b). In this description, we do not have included
the functional layer used to control the robotic platform.

Figure 6.
Autonomous coopeRatIve execution system (ARIES).

IEEE A&E SYSTEMS MAGAZINE

MARCH 2019

Aerospace and Electronic Systems - March 2019

Table of Contents for the Digital Edition of Aerospace and Electronic Systems - March 2019