...
 
Commits (2)
......@@ -3,7 +3,7 @@
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\section{Introduction to Evolutionary Computing}
The work presented in this thesis reflects the result of an ambitious goal to progress the vision of a robotic ecosystems which evolves in real time and real space\cite{2014-invivo, eiben2012embodied, nature-2015}.
The work presented in this thesis reflects the result of an ambitious goal to progress the vision of a robotic ecosystem which evolves in real time and real space\cite{2014-invivo, eiben2012embodied, nature-2015}.
This implies that the robot morphologies (body, hardware), as well as the controllers (mind, software), are evolvable, i.e., subject to reproduction and selection.
In other words, the primary concern is on robots that can produce offsprings more adaptive to their environment.
......@@ -21,24 +21,24 @@ Last but not least, significant advancements are recently being made in material
All these potential platforms share the ability to create programmable and customised hardware with ease.
Embedding evolution in hardware design is likely going to grow as a much more interesting topic in the future, allowing objects to become more resistant, but also more personalised to the functional and aesthetic needs of the single person.
Over the course of the past six years, a group of researchers at Vrije University Amsterdam have worked on developing the entire system\cite{nature-2015,eiben2012embodied}, in simulation and in hardware, as well as its individual components e.g. particular modules design solutions and specific algorithms for gait learning\cite{Evosphere-2015,jelisavcic2016improving,jelisavcic2017analysis}.
The result is a population of robotic organisms that adapts to the given environment using evolution.
Over the course of the past six years, a group of researchers at Vrije University Amsterdam have worked on developing the entire system\cite{nature-2015,eiben2012embodied}, in simulation and hardware, as well as its individual components, e.g. particular modules design solutions and specific algorithms for gait learning\cite{Evosphere-2015,jelisavcic2016improving,jelisavcic2017analysis}.
The result is a population of robotic organisms which adapts to the given environment using evolution.
However, the primary concern is still the automation of a robotic learning process, autonomous walking in the first place, specifically gait learning and targeted locomotion.
This thesis represents an attempt to find a possible automated solution for autonomous walking learning that can be implemented in a real world scenario with physical robots.
This thesis represents an attempt to find a possible automated solution for autonomous walking learning that can be implemented in a real-world scenario with physical robots.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{\tol}
The grand vision of a system where robots can evolve, learn, interact, and reproduce in real-time and real-space has been introduced, at first, as a theoretical model, namely called \tol\cite{triangle-2013}.
The \tol concept is generic, based on the principles of \EAs (EA), which assumes genotype-phenotype dichotomy.
The \tol concept is generic, based on the principles of \EAs (EA), which assumes a genotype-phenotype dichotomy.
The main difference from the typical \EA approach is the inclusion of co-evolution of minds and bodies, and noisy evaluation process conditioned with an \textit{online} evolutionary process.
Apart from these two main differences, \tol is similar to any other \EA considering the conceptual stages it consists of;
although the evolution is not organised in generations, as individual are independently reproducing without a global control by the system.
although the evolution is not organised in generations, as individuals are independently reproducing without a global control by the system.
The \tol concept distinguishes three principal stages: Morphogenesis, Infancy, and Mature Life, as illustrated in Figure \ref{fig:ToL}.
The birth process, i.e. Morphogenesis, is the first stage of life of an individual robot that spans from the moment of activation of a new genome (circle 1 on Figure \ref{fig:ToL}) up until the physical completion of a body encoded with the genome (circle 2).
Infancy, the second stage, is the period that starts from the genesis of a new robot (circle 2) until the robot is capable of acting independently in its environment (circle 3).
Infancy, the second stage, is the period that starts with the genesis of a new robot (circle 2) and ends when the robot is capable of acting independently in its environment (circle 3).
In the third stage, robots are programmed to conceive a ``child'', i.e. produce a new genome through mutation and recombination of the robot' genomes.
It should be noted that at this point we switch perspectives: the beginning of a new life cycle marks the beginning of another \tol belonging to the new robot encoded by the newly created genome.
......@@ -55,7 +55,7 @@ It should be noted that at this point we switch perspectives: the beginning of a
There are many possible implementations of the general \tol framework, distinguishable by different morphologies and controller architectures, but in all of these newborn robots are random combinations of the bodies and minds of their parents.
This raises a problem: new robots are born with new bodies that can and will be different from the bodies of the parents.
This situation implies that every newborn robot must acquire a new controller that matches the new body quickly after birth\footnote{Even if the parents had well matching bodies and minds, there is no general guarantee that recombination and mutation will keep the good match. See also \cite{Cheney2016}}.
This situation implies that every newborn robot must acquire a new controller that matches the new body quickly after birth\footnote{Even if the parents had well-matching bodies and minds, there is no general guarantee that recombination and mutation will keep the good match. See also \cite{Cheney2016}}.
This leads us to focus on the main problem, located in the Infancy stage, which is essential for any future implementation: \textit{locomotion learning}.
......@@ -109,13 +109,13 @@ As a first step to reach our main goal, we establish that the controller should
\textbf{\textit{Train the proposed controller to achieve targeted locomotion, in the form of phototaxis.}}
In the second step, we establish whether it is possible to train the controller to move in the direction of a target.
If possible, to stop the robot when the target is within reach.
The target of choice is represented by a light source.
A light source is the target of choice.
The robot can estimate the position of the light by using two light sensors (photoresistors) positioned symmetrically in the head of the robot.
These light sensors are used in addition to the previously mentioned set of sensors.
The behaviour of locomotion towards a light is called phototaxis.
\end{itemize}
% In order to answer this question, research was divided into two parts.
% In order to answer this question, the research was divided into two parts.
% The first part addresses the problem of simple gait learning, to determine if the new design is a justifiable solution for the main task.
% Experiments were conducted on a set of symmetrical and asymmetrical morphologies, which are used as a benchmark, in order to test their movement capabilities.
% Details about the implementation and experimental setup are described in Chapter \ref{ch:Method} and \ref{ch:Experiments}.
......
This diff is collapsed.
......@@ -245,12 +245,12 @@
% Include an abstract of your report
\abstract{
% explain enviroment
Invisioning the development of an evolvable robotic ecosystem called \tol, where robots can evolve both software and hardware, many components still need to be developed.
% explain environment
Envisioning the development of an evolvable robotic ecosystem called \tol, where robots can evolve both software and hardware, many components still need to be developed.
% explain problem
One of the main missing components is teaching newborn robots to move, not just randomly, but towards a target of choice. The target could be an area to explore, a food source or a mating area.
% explain aim of the project
The aim of this thesis is to create a combination of learner and controller algorithms capable of genereting targeted locomotion for a variety of robot morphologies.
% explain the aim of the project
This thesis aims to create a combination of learner and controller algorithms capable of generating targeted locomotion for a variety of robot morphologies.
}
% \begin{abstract}
......