The Accompanying Behavior Model and
Implementation Architecture of Autonomous Robot
Software
Shuo Yang, Xinjun Mao, Zhe Liu, Sen Yang, Jiangtao Xue, and Zixi Xu
College of Computer,
National University of Defense Technology
Email: {yangshuo@nudt.edu.cn}
Abstract—Autonomous robots are increasingly applied in real-
world environments, and expect to execute plans robustly to
accomplish the assigned tasks in the presence of dynamics
and uncertainties of the changing environment. The robustness
of plan execution requires the robot to keep aware of plan
execution status and to adapt the plan towards possible execution
contingencies. Such requirements pose a great challenge for
autonomous robot software in terms of the abstraction model
over robot behavior patterns. Conventional abstraction mod-
els for robot behaviors generally follow the sense-model-plan-
act and behavior-based paradigms, which show limitations in
tight integration with sensory inputs and tracking execution
traces of robot plans. This paper proposes an accompanying
behavior model that considers robot behaviors as task-oriented
and observation-based types with diverse aims, and develops
the run-time mechanisms to facilitate collaboration between two
types of behaviors. Additionally, we implement the model by
the multi-agent approach which develops the robot software
as a multi-agent system. To demonstrate the feasibility and
applicability of proposed model, we conduct a case study by
implementing a typical example of service scenarios, e.g., a robot
that autonomously picks up and drops off dishes for remote
guests in the open and dynamic environment.
I. INTRODUCTION
Autonomous robots are increasingly applied in real-world
environments to provide services as partners and tools for
human daily life [1], such as manufacture, field exploration,
domestic care, etc. The working environments for autonomous
robots are typically open, dynamic, and non-structured, which
may present uncertainties and dynamics for robot plan exe-
cution. Moreover, the inherent instability of robot hardware
effectors may produce biases between actual operation results
and expected results, which may also make the plan execution
fail. Considering the case when a service robot is grasping
a cup of tea, the pick plan will not succeed if the cup is
moved to somewhere else or the robot gripper doesn’t follow
the pre-computed arm trajectory exactly. In the presence of
such uncertainties, the autonomous robot expects to carry out
plans both robustly and adaptively to accomplish its assigned
task. The issue of robust plan execution requires the robot
to keep aware of plan execution status and to adapt the plan
in case of possible run-time contingencies, which is of great
significance for autonomous robot to operate robustly in highly
dynamic environments.
Moreover, the software for autonomous robot plays an
important role in making high-level decisions and guiding the
robot to accomplish complex tasks. The autonomous robot
software typically incorporates behavior abstraction models
to specify the robot behavior patterns, together with resultant
software architectures to implement and realize the abstraction
models. However, the robustness of robot plan execution
presents a great challenges for the behavior abstraction models,
which expects the software to continuously sense the situated
environment, monitor the status of plan execution and adapt
the plan in case of run-time emergencies.
Conventional behavior models generally follow the
paradigms of sense-model-plan-act [2] and behavior-based [3],
both of which view the robot behavior pattern from different
aspects. The sense-model-plan-act model specifies a simple
and sequential behavior pattern for robot control. Its main
features are that the sensing behavior outputs the environment
information into a world model, which is then used by the
planning behavior, and that the plan is executed by the
acting behavior [4], [5]. However, this model is established
under assumptions of static or predicable environments, and
it appears rather dangerous for autonomous robot to act in
a dynamic world because no sensing behaviors are involved
while the robot is acting to execute the plans [6]. Another
model in behavior-based paradigm defines a number of levels
of competence that organizes in a subsumption architecture,
with each level of competence specifying a desired class of
robot behaviors. The behavior-based model incorporates the
arbitration schemes that enable the high-level behaviors to
override signals from the lower-level behaviors. Although it
facilitates environmental awareness and strong reactivity, there
are no explicit specification on the interactions between sens-
ing behaviors and acting behaviors, because it proves difficult
to plan robot behaviors in an optimal manner to achieve long-
range and complex goals [7]. As can be analyzed, both of
these behavior models present the limitations to support a
tight, concurrent and scheduled connection between the robot
sensing and acting behaviors, so as to facilitate the robust plan
execution in open and dynamic working environments.
This paper proposes a novel paradigm of accompanying
behavior model to specify the robot behavior pattern that
is able to carry out plan execution process robustly. More