Autonomous Robotics

Misyurov D.A. Dialectical formulas based on the binary notation as the development formulas // Credo New. 2012. №2

by Dmitry Misyurov

The article suggests dialectical formulas based on the binary notation as the development formulas: formula with... more The article suggests dialectical formulas based on the binary notation as the development formulas: formula with dominant and the non-dominant elements; universal formula; formula with symbolic weight of elements; tautological formula. For example, it suggests an opportunity to use the dialectical formulas for modeling and artificial intelligence creation, etc.

Autonomy and Desire in Machines and Cognitive Agent Systems

by Kevin Magill

Co-authored with Yasemin J. Erden, published 'Online First' in 'Cognitive Computation', 2012, DOI DOI 10.1007/s12559-012-9140-9. Available through subscribing institutions. Drop me a line if you can't access.
Substantially revised from 2011 paper of same title.

The development of cognitive agent systems relies on theories of agency, within which the concept of desire is key.... more The development of cognitive agent systems relies on theories of agency, within which the concept of desire is key. Indeed, in the quest to develop increasingly autonomous cognitive agent systems, desire has had a significant role. We argue, however, that insufficient attention has been given to analysis and clarification of desire as a complex concept. Accordingly, in this paper, we will draw on some key philosophical accounts of the nature of desire, including what distinguishes it from other mental and motivational states, in order to identify some key characteristics of desire as a complex concept. We will then draw on these in order to investigate the role, definition and adequacy of concepts of desire within applied theoretical models of agency and agent systems.

Cooperative Behavior of Artificial Neural Agents Based on Evolutionary Architectures

by Piero Savastano

The Morality of Lethal Autonomous Machines

by Emmanuel Goffi

Published in CRUSER News, issue 14, April 2012.

Hydrodynamic pressure sensing with an artificial lateral line in steady and unsteady flows

by Maarja Kruusmaa

Roberto Venturelli, Otar Akanyeti, Francesco Visentin, Jaas Ježov, Lily D Chambers, Gert Toming, Jennifer Brown, Maarja Kruusmaa, William M Megill and Paolo Fiorini, Bioinspiration & Biomimetics Volume 7 Number 3

With the overall goal being a better understanding of the sensing environment from the local perspective of a situated... more With the overall goal being a better understanding of the sensing environment from the local perspective of a situated agent, we studied uniform flows and Kármán vortex streets in a frame of reference relevant to a fish or swimming robot. We visualized each flow regime with digital particle image velocimetry and then took local measurements using a rigid body with laterally distributed parallel pressure sensor arrays. Time and frequency domain methods were used to characterize hydrodynamically relevant scenarios in steady and unsteady flows for control applications. Here we report that a distributed pressure sensing mechanism has the capability to discriminate Kármán vortex streets from uniform flows, and determine the orientation and position of the platform with respect to the incoming flow and the centre axis of the Kármán vortex street. It also enables the computation of hydrodynamic features which may be relevant for a robot while interacting with the flow, such as vortex shedding frequency, vortex travelling speed and downstream distance between vortices. A Kármán vortex street was distinguished in this study from uniform flows by analysing the magnitude of fluctuations present in the sensor measurements and the number of sensors detecting the same dominant frequency. In the Kármán vortex street the turbulence intensity was 30% higher than that in the uniform flow and the sensors collectively sensed the vortex shedding frequency as the dominant frequency. The position and orientation of the sensor platform were determined via a comparative analysis between laterally distributed sensor arrays; the vortex travelling speed was estimated via a cross-correlation analysis among the sensors.

Models of Cockroach Shelter Seeking Implemented on a Robotic Test Platform

by Brian Tietz

Master's Thesis, Case Western Reserve Univeristy

Animal behavior is often a model for robotic control, with benefits for both robotics and biology. This research... more Animal behavior is often a model for robotic control, with benefits for both robotics and biology. This research covers a new animal behavior for this category: cockroach shelter-seeking.
Cockroach behavior was tracked in a 91 cm by 91 cm arena, and significant trends were identified that form a stochastic navigation algorithm called RAMBLER. Components of RAMBLER were then implemented on a mobile robot, and compared with a deterministic model of the same cockroach behaviors. In the process of programming the robotic model, an interesting behavior was discovered when the cockroach loses contact with a barrier in the arena, posing new questions about animal behavior.

Self‐supervised learning to visually detect terrain surfaces for autonomous robots operating in forested terrain

by Phil Salesses

Evolution of autonomous and semi-autonomous robotic surgical systems: a review of the literature

by George Moustris

Co-authored with S. C. Hiridis, K. M. Deliparaschos, K. M. Konstantinidis

Background

Autonomous control of surgical robotic platforms may offer enhancements such as higher... more Background

Autonomous control of surgical robotic platforms may offer enhancements such as higher precision, intelligent manoeuvres, tissue-damage avoidance, etc. Autonomous robotic systems in surgery are largely at the experimental level. However, they have also reached clinical application.

Methods

A literature review pertaining to commercial medical systems which incorporate autonomous and semi-autonomous features, as well as experimental work involving automation of various surgical procedures, is presented. Results are drawn from major databases, excluding papers not experimentally implemented on real robots.

Results

Our search yielded several experimental and clinical applications, describing progress in autonomous surgical manoeuvres, ultrasound guidance, optical coherence tomography guidance, cochlear implantation, motion compensation, orthopaedic, neurological and radiosurgery robots.

Conclusion

Autonomous and semi-autonomous systems are beginning to emerge in various interventions, automating important steps of the operation. These systems are expected to become standard modality and revolutionize the face of surgery.

Learning Behaviours for Robot Soccer

by James Brusey

PhD Thesis, Awarded Australasian Distinguished Doctoral Dissertation 2004 (www.core.edu.au)

A central problem in autonomous robotics is how to design programs that determine what the robot should do next.... more A central problem in autonomous robotics is how to design programs that determine what the robot should do next. Behaviour-based control is a popular paradigm, but current approaches to behaviour design typically involve hand-coded behaviours. The aim of this work is to explore the use of reinforcement learning to develop autonomous robot behaviours automatically, and specifically to look at the performance of the resulting behaviours.

This thesis examines the question of whether behaviours for a real behaviour-based, autonomous robot can be learnt under simulation using the Monte Carlo Exploring Starts, -soft On Policy Monte Carlo or linear, gradient-descent Sarsa algorithms. A further question is whether the increased performance of learnt behaviours carries through to increased performance on the real robot. In addition, this work looks at whether continuing to learn on the real robot causes further improvement in the performance of the behaviour.

A novel method is developed, termed Policy Initialisation, that makes use of the domain knowledge in an existing, hand-coded behaviour by converting the behaviour into either a reinforcement learning policy or an action-value function. This is then used to bootstrap the learning process.

The Markov Decision Process model is central to reinforcement learning algorithms. This work examines whether it is possible to use an internal world model in the real robot to suit the requirements of the Markov Decision Process model.

The methodology used to answer these questions is to take three realistic, non-trivial robotic tasks, and attempt to learn behaviours for each. The learnt behaviours are then compared with hand-coded behaviours that have either been published or used in international competition. The tasks are based on real task requirements for robots used in a RoboCup Formula 2000 robot soccer team. The first is a generic movement behaviour that moves the robot to a target point. The second requires the robot to dribble the ball in an arc so that the robot maintains possession and so that the final position is lined up with the goal. The third addresses the problem of kicking the ball away from the wall.

The results show that for these three different types of behavioural problem, reinforcement learning on a simulator produced significantly better performance than hand-coded equivalents, not only under simulation but also on the real robot. In contrast to this, continuing the learning process on the real robot did not significantly improve performance.

The Policy Initialisation technique is found to accelerate learning for tabular Monte Carlo methods, but makes minimal improvement and is, in fact, costly to use in conjunction with linear, gradient-descent Sarsa. This approach, unlike some other techniques for accelerating learning, does not appear to bias the solution.

Finally, the evidence from this thesis is that internal world models that maintain the requirements of Markov Decision Processes can be constructed, and this appears to be a sound approach to avoiding problems connected with partial observability that have previously occurred in the use of reinforcement learning in robotic environments.

Robotic Lake Lander Test Bed for Autonomous Surface and Subsurface Exploration of Titan Lakes

by Ramaprasad Kulkarni

"Co-authored with Wolfgang Fink, Markus Tuller, Alexander Jacobs, Mark Tarbell, R Furfaro, V R Baker","published in 'IEEE Aerospace Proceedings',2012"

We introduce a robotic lake lander test bed that can be operated either stand-alone or as part of a Tier-Scalable... more We introduce a robotic lake lander test bed that can be operated either stand-alone or as part of a Tier-Scalable Reconnaissance mission architecture to study and field test an integrated hardware and software framework for fully autonomous surface and subsurface exploration and navigation of liquid bodies. The lake lander is equipped with both surface and subsurface sensor technologies. Our particular focus is on Saturn's moon Titan with its hydrocarbon lakes with respect to future missions involving lake landers (e.g., Titan Mare Explorer (TiME) mission), potentially in conjunction with balloons/airships and orbiter-support overhead. This test bed serves as an analog to a Titan unpiloted surface vessel equipped with its own onboard realtime navigation and hazard avoidance system, surface and subsurface exploration sensor suite, and autonomous science investigation software system. As such the test bed helps map out a technical path toward true autonomy for the robotic exploration of the Solar System.

Birth of the object: Detection of objectness and extraction of object shape through object action complexes

by Nicolas Pugeault

Novel central pattern generator elements for autonomous modular robots

by Fernando Herrero-Carrón

Supervised by Prof. Pablo Varona Martínez and Prof. Francisco de Borja Rodríguez Ortiz

Central Pattern Generators are neural circuits that are found in living
beings, from simple animals like... more Central Pattern Generators are neural circuits that are found in living
beings, from simple animals like mollusks, up to higher mammals and human
beings. They are responsible for the control of rhythmic motor activities
including walking, breathing and chewing among others. Their patterns of
activity are robust yet flexible, meaning that some intrinsic, measurable properties
of the rhythm are kept invariant, while at the same time modulation of
the rhythm is possible, for instance in frequency or amplitude. For all these
reasons, CPG models have been used as locomotion controllers in different
types of robots. In particular, the modular nature of CPGs has fit very well
in the specific field of robotics dealt with in this thesis: that of modular
robots.
Following recent research on living CPGs, we propose a new concept of
bio-inspired controller for a modular worm-like robot. The central biological
principles from which this thesis draws inspiration are: rich intrinsic dynamics
for neurons and synapses, mutual inhibition in non-open topologies
and winnerless competition dynamics, all of which are characteristic to living
CPGs.
In this thesis we first introduce and study the neuron and synapse models
that will later be used to build different CPG circuits. We analyze the
relationship between individual parameters and neural and synaptic activity.
Using these models, we build a simple oscillator to control a single module
of the robot. Again, we study its behavior and the influence of different control
parameters. Finally, we solve the problem of decentralized, autonomous
rhythm coordination. With the right connectivity topology and dynamics,
each module negotiates with its neighbor to establish a rhythmic oscillation
and an overall frequency and phase difference. For this, we propose four
different inter-modular connectivity patterns. These patterns are scalable
and, just by repetition, CPGs with an arbitrary number of modules can be
constructed.
The result is a set of new bio-inspired strategies that can be used to
build effective CPG controllers for autonomous modular robots. The right
combination of intrinsic dynamics and topological connectivity is the key to
adaptability and self-organization of the locomotive rhythm. We show that
the dynamical properties of the models make the proposed modular oscillator
more resilient to noise than other state of the art works. Then, we couple our
oscillator to different simulated servomotors without changing its parameters,
and show that it can adapt its frequency of oscillation over some orders of
magnitude, depending on servo responsiveness. Finally, we show that the
four proposed connectivity patterns generate the desired rhythm, and CPGs
built this way effectively generate forward locomotion in a real robot.
4

Emergence of Safe Behaviours with an Intrinsic Reward

by Maarja Kruusmaa

Y. Gavshin, M. Kruusmaa. Emergence of Safe Behaviours with an Intrinsic Reward. In: Adaptive and Intelligent Systems: 2011 International Conference on Adaptive and Intelligent Systems, Klagenfurt, Austria, September 6-8, 2011. A. Bouchachia. Berlin/Heidelberg: Springer Verlag, 2011, LNAI 6943, pp. 180-191.

This paper explores the idea that robots can learn safe be- haviors without prior knowledge about its environment nor... more This paper explores the idea that robots can learn safe be- haviors without prior knowledge about its environment nor the task at hand, using intrinsic motivation to reverse actions. Our general idea is that if the robot learns to reverse its actions, all the behaviors that emerge from this principle are intrinsically safe. We validate this idea with experiments to benchmark the performance of obstacle avoidance behavior. We compare our algorithm based on an abstract intrinsic re- ward with a Q-learning algorithm for obstacle avoidance based on exter- nal reward signal. Finally, we demonstrate that safety of learning can be increased further by first training the robot in the simulator using the intrinsic reward and then running the test with the real robot in the real environment.
The experimental results show that the performance of the proposed algorithm is on average only 5-10% lower than of the Q-Learning algo- rithm. A physical robot, using the knowledge obtained in simulation, in real world performs 10% worse than in simulation. However, its perfor- mance reaches the same success rate with the physically trained robot after a short learning period. We interpret this as the evidence confirm- ing the hypothesis that our learning algorithm can be used to teach safe behaviors to a robot.

Assessing Safety of Object Pushing Using the Principle of Reversibility

by Maarja Kruusmaa

Gavshin, Y.; Kruusmaa, M. (2011). Assessing Safety of Object Pushing Using the Principle of Reversibility. In: Hybrid Artificial Intelligent System: 6th International Conference, HAIS 2011, Wroclaw, Poland, May 23-25, 2011. (Toim.) E. Corchado, M. Kurzynski, M. Wozniak. Berlin/Heidelberg: Springer-Verlag?, 2011, (Lecture Notes in Artificial Intelligence, Subseries of Lecture Notes in Computer Science; 6678), 313 - 320.

This article presents an implementation of an innovative safety module for a robot control architecture. It applies... more This article presents an implementation of an innovative safety module for a robot control architecture. It applies the principle of reversibility to assess intrinsic safety of actions and to adapt robot’s be- havior. The underlying idea is that all reversible actions are intrinsically safe. A practical experiment is conducted to demonstrate that a robot control architecture can develop complex safe behaviors. This is accom- plished by using the safety module in conjunction with human-based knowledge and sufficiently high level of perception. A robot is placed in a room with a movable object while the safety module analyzes move- ments of the robot and the object. As the result, the robot can identify, for example, that pushing object into a corner is irreversible and thus unsafe.

Identification of Reverse-Action Pairs using Reversibility of Actions

by Maarja Kruusmaa

Gavshin, Yuri; Kruusmaa, Maarja (2011). Identification of Reverse-Action Pairs using Reversibility of Actions. In: 2011 IEEE International Conference on Systems, Man, and Cybernetics (SMC 2011), Anchorage, AK, USA, October 9-12, 2011. IEEE, 2011.

This paper presents a practical approach to identi- fication of reverse-action pairs for reversibility-based learning... more This paper presents a practical approach to identi- fication of reverse-action pairs for reversibility-based learning to develop safe behaviors. The approach is to analyze reversibility of consecutive actions by introducing metrics defined on the actions and the states. The experiment is conducted in the Player/Stage simulator to test applicability of the approach. The robot is allowed to move freely in the environment by executing two- dimensional actions – moving forward/backward or rotating left/right. Reversibility of a sum of every consecutive pair of actions is then analyzed to identify the ones that undo each other. As a result, a set of action pairs are identified with a general rule “if moved/rotated by X, move/rotate by -X to undo”.

Against the flow: A Braitenberg controller for a fish robot

by Maarja Kruusmaa

T. Salumäe, I. Rano, O. Akanyeti, M. Kruusmaa, "Against the flow: A Braitenberg controller for a fish robot", IEEE International Conference on Robotics and Automation, St. Paul, USA, May 14-18, 2012, to appear

Underwater vehicles do not localise or navigate with respect to the flow, an ability needed for many underwater tasks.... more Underwater vehicles do not localise or navigate with respect to the flow, an ability needed for many underwater tasks. In this paper we implement rheotaxis behaviour in a fish robot, a behaviour common to many aquatic species. We use two pressure sensors on the head of the robot to identify the pressure differences on the left and right side and control the heading of the fish robot by turning a servo-motor actuated tail. The controller is inspired by the Braitenberg vehicle 2b, a simple biological model of tropotaxis, that has been used in many robotic applications. The experiments, conducted in a flow pipe with a uniform flow, show that the robot is able to orient itself, and keep the orientation, to the incoming current. Our results demonstrate that guidance of a fish robot relative to a flow can be implemented as a simple rheotaxis behaviour using two sensors and a Braitenberg 2b controller.

The Machine Question: Ethics, Alterity and Technology

by David Gunkel

Chapter 6 of "Thinking Otherwise: Philosophy, Communication, Technology" (Purdue University Press, 2007)

This chapter, published in "Thinking Otherwise: Philosophy, Communication, Technology" (2007) represents... more This chapter, published in "Thinking Otherwise: Philosophy, Communication, Technology" (2007) represents something like a manifesto for the rights of machines. It considers the machine (AIs, robots, autonomous systems, etc.) as both moral agent and moral patient and argues for the ethical standing and appropriate treatment of artificial entities. The title refers to and expands on the "animal question," which has had considerable influence in moral philosophy during the later half of the 20th century. The machine question takes this moral innovation one step further by demonstrating that the machine, the other of the animal other, remains one of the last socially acceptable moral prejudices and arguing for a thinking of ethics that is able to proceed otherwise.

Controladores Fuzzy para o Posicionamento Sem Bola do Goleiro no Futebol de Robôs Simulado 2D.

by Jessica Meyer

In: 5th WTICG-BASE - 5th Workshop de Trabalhos de Iniciação Científica e Graduação Bahia, Alagoas e Sergipe, 2008, Vitória da Conquista. Proceedings of the 8th Escola Regional de Computação Bahia, Alagoas e Sergipe (ERBASE), 2008. Sociedade Brasileira de Computação (SBC), Porto Alegre, 2008, vol. 1.

This paper describes a research about fuzzy controllers for the positioning of the goalkeeper without the ball in the... more This paper describes a research about fuzzy controllers for the positioning of the goalkeeper without the ball in the 2d simulated robot soccer. The goal of this research was to improve the behavior of the goalkeeper, raising the number of catches and decreasing the number of goals, improving its positioning to catch the ball. To validate the research, 30 matches were simulated against some of the best teams of the world that participate of the RoboCup. The achieved results are presented, the conclusions are discussed and future works are suggested.

Controladores Fuzzy para Agentes Robôs Jogadores de Futebol

by Jessica Meyer

In: 4th WTICG-BASE - 4th Workshop de Trabalhos de Iniciação Científica e Graduação Bahia, Alagoas e Sergipe, 2007, Vitória da Conquista. Proceedings of the 7th Escola Regional de Computação Bahia, Alagoas e Sergipe (ERBASE), 2007. Sociedade Brasileira de Computação (SBC), Porto Alegre, 2007, vol. 1.

Bahia2D: Descrição do Time

by Jessica Meyer

In: CBR - Competição Brasileira de Robótica, 2007, Florianópolis. Proceedings of the 8th Simpósio Brasileiro de Automação Inteligente (SBAI). Sociedade Brasileira de Automática (SBA), Florianópolis, 2007, vol. 1.