Aerospace and Electronic Systems - March 2019 - 52

ARIES: An Autonomous Controller For Multirobot Cooperation

Figure 12.
Partial timeline-based plan translated from the partial PDDL plan in Figure 11 for the Leader and Follower Agents. The state variables are
represented on the left side. The green tokens represent the initial state of the mission. The blue tokens represent the actions taken from the
PDDL plan. The gray tokens are the effects of the tokens stored in the effects configuration file. The start of the mission is denoted by t, and
the execution frontier is denoted by t (line in red color).

ARIES a more robust execution against system failures.
Currently, ARIES is just able to manage the communication
lost and unachieved goal failures states, which is not enough
to provide a reliable autonomous controller. A possible way
to achieve a robust failure management system could be to
deploy a new T-REX reactor on every ARIES agent [leader
and follower(s)] that executes its own planner with the
world knowledge to deal with a wide range of opportunistic
failures. The current coordinated execution will keep
informed this reactor of every action execution, so it will be
able to execute the suitable contingency plan. Also, ARIES
does not allow the communication among the follower(s),
which limits the action plan when the leader is missing. A
possible solution could be to add to this failure manager
reactor the capabilities to
enable different operational modes into the TREX agent, which tells
the others reactors how to
operate in case of critical
failures, e.g., a missing
leader would trigger a
new operational mode
into the follower(s), in
which the COOP Reactor
starts the communication
with its nearest follower
partner, and so, they build
a communication network
to operate following a
swarm behavior.
The computational
results of executing
ARIES over the Nominal
scenario are presented in
Table 1, showing the
Figure 13.
communications between
Execution flow for taking two pictures, TP1 and TP2, with UGV at location SL1. Green tokens represent
agents for coordination
the initial states. Blue tokens represent the planned actions to be executed. Gray tokens are the effects of
(expressed in number of
accomplishing the planned actions. The execution frontier is denoted by t.
Thus, every token execution in a state variable implies an
effect, which means that ARIES requires two tokens
exchange to carry out the coordination between the agents.
Second, we can observe that both agents represent the
UAVCamera state variable evolution, which means that
both agents share information in order to accomplish a
proper coordination. This information sharing will enhance
the design of failure recovery models for future scenarios
where both agents coordinate a joint plan to solve anomalous situations. Finally, we observe how the dispatching
process has been split by a coordination process in which
both agents communicate to hold a truthful coordination.
The current coordination process is an essential step,
which requires a future deeper assessment to provide

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