Attachment 'Index.html'
Download| Page 1 Page 1 |
Indice Index
Abstract Abstract
Ros, un sistema in grado di ricostruire a Ros, a system able to reconstruct in
rgbd la mappa di un luogo e di creare rgbd map of a place and create
Introduzione Introduction
ambiente 3D in precedenza ricostruito 3D environment previously reconstructed
sensori rgbd montati a bordo. sensors mounted on board rgbd.
ricostruzione tridimensionale di un ambiente e la reconstruction of a three-dimensional environment and
navigazione all'interno di esso. navigation within it. Dal punto di vista pratico, invece, From the practical point of view, instead,
quadrirotore all'interno di un ambiente ricco di ostacoli di cui ha quadcopter in an environment full of obstacles that have
mappa con lo scopo di confrontare all'interno di map with the purpose of comparing within
SHERPA (Smart collaboration between Humans and (Smart collaboration between Humans and
“ROS” come base. "ROS" as a base.
“Gazebo”. "Gazebo".
Capitolo 1 Chapter 1
( “mapping” ) e definire all'interno di esso un percorso privo di ostacoli per ("Mapping") and defining within it an obstacle-free route for
( “path planning” ). ("path planning").
1.1 Mapping 1.1 Mapping
Possibilità di modellare spazio libero, occupato e non ancora esplorato: Ability to model space, occupied and not yet explored:
Rapidità e semplicità di creazione e aggiornamento: a seconda della Quick and easy creation and modification: depending on
Rappresentazione probabilistica delle informazioni: tipicamente i dati per Probabilistic information: typically the data for
Efficienza: sia in termini di tempi di accesso per ottenere informazioni su Efficiency: both in terms of access time to obtain information on
Multi-resolution: potrebbe risultare comodo avere una mappa in grado di Multi-resolution: it can be handy to have a map that can
1.1.1 Fonte di informazioni 1.1.1 Source of information
pointcloud , 3D. PointCloud 3D. Queste rappresentano la porzione di mondo These represent the portion of the world
immagini RGBD : ad ogni pixel sono associati quattro images RGBD: to each pixel are associated four
(rgb) e uno ( d ) per la distanza del punto rispetto (RGB) and one (d) to the distance of the point from
rgbd ad rgbd to
pointcloud da fonti molto diverse, ad esempio proiettori PointCloud from very different sources, such as projectors
Microsoft kinect o Asus prime. Gli Microsoft Kinect or Asus first. The
pointcloud ad un basso costo ed ingombro, PointCloud at a low cost and dimensions,
pointcloud mediante l'uso di voxel : cubi di dimensione fissa PointCloud by the use of voxels: cubes of fixed size
, a ciascuno di to each of
1.1.2 Creazione della mappa 1.1.2 Creating the map
pointcloud legate fra di loro PointCloud connected to them
pointcloud manca un'informazione fondamentale: dove si missing information PointCloud fundamental: where
pointcloud per poterle quindi posizionare nello spazio ed ottenere così una PointCloud for them then place in space and thus to obtain a
odometria del sensore: come si è mosso nello sensor: as he moved in
visual odometry , che tenendo traccia di visual odometry, that keeping track of
features, in immagini successive permettono di ricostruire lo features in successive images allow us to reconstruct the
VSLAM (visual simultaneous localization and mapping). VSLAM (visual simultaneous localization and mapping).
stream di immagini stream of images
rgbd per odometria e mapping , ma richiede una notevole potenza rgbd for odometry and mapping, but requires considerable power
pointcloud (necessita anche delle PointCloud (needs both
features . features.
pointcloud in un sistema di riferimento assoluto e ottenere così PointCloud in an absolute reference system and thus obtain
1.1.3 Mantenimento della mappa 1.1.3 Maintenance of map
Point-Cloud: Point-Cloud:
sensori rgbd sono in grado di generare nuvole dell'ordine dei rgbd sensors are capable of generating clouds of the order of
frame lavorando a frequenze di 30 Hz; frame working at frequencies of 30 Hz; è perciò evidente It is therefore clear
1.1 - Rappresentazione di 1.1 - Representation
un albero mediante a shaft by
pointcloud PointCloud
Voxel Grid [1]: Voxel Grid [1]:
voxel. L'ambiente è rappresentato mediante una griglia 3D di cubi, voxel. The environment is represented by a 3D grid of cubes,
voxel . voxels.
Elevation Map [2]: Elevation Map [2]:
1.2 - Rappresentazione di 1.2 - Representation
un albero mediante a shaft by
elevation map elevation map
Multi Level Elevation Grid [3]: Multi Level Elevation Grid [3]:
voxel occupati che si voxels that are occupied
voxel liberi per ogni cella). voxel free for each cell). L'aggiornamento The update
voxel , inoltre non è ancora possibile also not yet possible
Octomap [4] [5]: Octomap [4] [5]:
voxel più voxel
1.3 - Rappresentazione di 1.3 - Representation
un albero mediante multi a tree by means of multi
level elevation grid level elevation grid
1.4 - Decomposizione ricorsiva di un cubo 1.4 - recursive decomposition of a cube
sfruttando larappresentazione ad albero ottale exploiting larappresentazione tree octal
1.5 - Rappresentazione di 1.5 - Representation
un albero mediante a shaft by
octomap octomap
1.2 Path Planning 1.2 Path Planning
goal . Questa azione goal. This action
path planning ed è oggetto di ricerca scientifica con ottimi path planning and is the subject of scientific research with excellent
planning non necessariamente tengono planning does not necessarily hold
goal; goals; i n questo i n this
path planning geometrico . Esistono metodi che tengono già geometric path planning. There are methods that already take
goal , si parla in questo caso di path planning orientato al goal, we speak in this case of path planning oriented
controllo. control.
1.2.1 Definizione del problema [6] 1.2.1 Definition of the problem [6]
W l'insieme di tutti i punti che la W the set of all points that the
O la regione in cui Or the region where
F la regione libera di spazio. F region free space.
C composto da tutte le C consists of all the
Cfree il suo sottoinsieme contenente tutte le configurazioni appartenenti a F. Cfree its subset containing all configurations belonging to F.
start) e una finale ( goal) appartenenti and an end (goal) belonging
Cfree calcolare un percorso tra di esse equivale a cercare la sequenza di Cfree calculate a route between them is equivalent to the search sequence
Cfree che permettono di passare dalla partenza all'arrivo. Cfree that allow you to switch from start to finish.
1.2.2 Soluzioni Geometriche 1.2.2 Geometric Solutions
start a goal , senza nessuna informazione sui comandi da impartire al start in goal, without any information on the commands that shall be given to
path planning geometrico. Per poi muovere il robot sarà path planning geometric. To then move the robot will
D urante gli anni molte soluzioni sono state D uring the years many solutions have been
potential fields [7]: il potential fields [7]: the
goal goal
goal. Ha però un grave goal. He, however, a serious
punti di minimo ”, punti in cui la somma delle forze prodotte è 0, se il points", points in which the sum of the forces produced is 0, if the
uso di grafi in cui il problema of graphs in which the problem
start e goal alla struttura generata che prende quindi il the start and goal to the structure generated which then takes the
roadmap . roadmap. A questo punto il problema di planning è stato semplificato At this point the problem of planning has been simplified
routing su un grafo già ampiamente on a graph already widely
Scomposizione in celle dell'intera mappa: Break it down into cells of the entire map:
start e goal per ottenere una roadmap consistente. start and goal to achieve a consistent roadmap. Questo This
completo . complete. Necessita però di un'analisi esaustiva di tutta la mappa, However, it requires a comprehensive analysis of the whole map,
combinatorie . combinatorial.
1. Celle omogenee: 1. Cells homogeneous:
2. Celle poligonali [6]: 2. Cells polygonal [6]:
1.6 - Decomposizione in celle 1.6 - decomposition in cells
poligonali e grafo ottenuto polygonal and graph obtained
3. QuadTree/Octree [8]: 3. quadtree / octree [8]:
4. Diagramma di Voronoi [9]: 4. Diagram of Voronoi [9]:
roadmap ottimale considerando come nodi i roadmap optimal considering how the nodes
1.8 - Decomposizione in zone di 1.8 - decomposition in areas of
Voronoi di una mappa Voronoi of a map
bidimensionale two-dimensional
1.7 - Decomposizione di una 1.7 - decomposition of a
mappa bidimensionale mediante two-dimensional map by means of
quadtree quadtree
5. Visibility Graph [10]: 5. Visibility Graph [10]:
roadmap che permette di roadmap that allows
Probabilistic roadmap (PRM) [11]: Probabilistic Roadmap (PRM) [11]:
approcci probabilistici alla risoluzione del problema: i probabilistic to the resolution of the problem:
1.9 - visibility graph ottenuto 1.9 - visibility graph obtained
partendo da una mappa starting from a map
bidimensionale two-dimensional
1.10 - Probabilistic roadmap 10.01 - Probabilistic roadmap
ottenuta su una mappa obtained on a map
bidimensionale two-dimensional
Cfree, la parte computazionalmente più complessa diventa creare le Cfree, the part becomes more computationally complex create
n n
n definito a priori, e raggiungibili attraverso un n defined a priori and can be reached through a
k nodi più vicini indipendentemente dalla distanza a cui k nodes closer regardless of distance to which
roadmap è considerata abbastanza roadmap is considered enough
start e and start
goal alla roadmap connettendole ai nodi più vicini e sfruttare il grafo per goals by connecting them to the roadmap nodes closer and exploit the graph for
completa : non è possibile stabilire con complete: it is not possible to determine
goal, ne trovare sempre in goal, you always find in
probabilisticamente completa : la probabilità di non probabilistically complete: the probability of not
roadmap può essere sfruttata per molte roadmap can be used for many
query , di conseguenza il costo della fase di creazione viene ammortizzato . queries, accordingly the cost of being created is amortized.
1. PRM* [12]: il valore del parametro n, definito nell'algoritmo 1. PRM * [12]: the value of the parameter n, defined in the algorithm
2. LazyPRM [13]: Funziona come il PRM normale, ma l'assenza di 2. LazyPRM [13]: Functions as the PRM normal, but the absence of
start e goal viene verificata the start and goal is verified
query is slower.
Rapidly exploring random trees [14]:
start , at each
Cfree looking for the nearest node and if it is
goal, if
roadmap in this case the data structure is created single-
query : whenever a request is made to Schedule I is
roadmap : the number of nodes and especially of branches is less and the
local planners, the component that has the
1.11 - RRT obtained from a map
two-dimensional, the pink zone
is the goal
1. RRT-connect [15]:
start , the other in the goal . The expansion is alternating, after a certain number
2. RRG ( Rapidly exploring random graph ) [12]:
roadmap : the expansion in the algorithm proceeds as the basis, but to
Road map.
3. Optimal RRT (RRT *) [12]:
EST (Expansive Search Tree) [19]:
Cfree and track configurations already explored: cells
path in the tree; the points
the start in the other goal . There are variants posticipates
SBL (single-query bi-directional lazy collision checking) [16]. In this
Search algorithms on trees and graphs:
start and goal, come in search algorithms on graphs or
A * [17] . The idea
start , until
goal . Then you can determine a path
start , just calculated, and the goals , obtainable only
goal .
open = list of accessible nodes to visit;
closed = list of nodes already visited;
start = start node;
goal = node to be achieved;
goal_non_visitato = true;
goal_non_visitato ) {
primaEsecuzione ) {
x = start;
x = estraiNodoConScorePiùBasso ( open ) ;
effettuavisita ( x ) ;
x == goal ) {
goal_non_visitato = false;
closed .aggiungi ( x );
open .rimuovi ( x );
aggiuntaNuoviNodiRaggiungibili ();
aggiornamentoScoreDeiNodi ( closed );
1.2.3 Solutions based on control:
path planning the
query ,
t, starting from the configuration of start apply all the
t , and calculate all the configurations
Cfree , repeat the procedure for each state so
start,
goals.
KPIECE (kinodinamic plannign by interior-exterior cell exploration) [18].
1.12 - Create a tree
a two-dimensional map
by algorithm KPIECE
outside the cells and internal : the
goal within a reasonable time and with a data structure very light if
R R
1.3 Conclusion:
octomap it appears as the solution
the start, goal and situation of the space around the robot.
Chapter 2:
open-source.
ROS; the role of planning and mapping was
MoveIt !, based on many libraries between
OMPL for path planning and octomap for mapping .
2.1 ROS [20]
ROS , which stands for "robot operating system", is one of the solutions open source
ubuntu . It also allows you to facilitate the management of systems
rpc , through the use of services .
topics .
parameter server .
stacks or package easily distributable and reusable, so as to
run-time even in
package or stack for easy
2.1.1 Key Concepts
ROS realize is the "ROS Computation Graph" that is a peer network
Nodes: the processes that run within the framework ROS .Sono I'm
Master: ( or ROS-master ) is the naming system integrated in ROS that associates
Parameter Server: currently integrated in the master, is a system
framework may at any time and read
Messages: Format through which the nodes communicate with each other, are
packet of applications ROS.
Topics: Mechanism through which it makes possible communication between nodes
pub / sub : a node produces messages and makes them available
topic ( publish), those interested in those
topic ( subscribe) whenever there is a
subscribers are notified by the framework in
topic must be strongly typed, ie
topic , as well as more nodes can
topics : each is a separate entity, a
Services: Mechanism through which communication is made between
Client / Server synchronous blocker. A node is able
remote procedure call .Il The
clients will have to know what features are available and how to invoke them,
signature of the functions are stored in special files descriptive
.srv " and the list of possible functions is maintained in the Master.
Bags: I am a format used to store and play back at a later
ROS-messages.
ROS also provides a set of tools for the management of all
Package: the main unit for the organization of the software, within
Stack: is a set of packages logically related in some way between
Manifest: it is a file xml in each package, which provides a
Stack Manifest: realizes the function of the manifest at the level of the stack .
Message Type: description file messages ROS used by nodes
package , typically found in the " msg "within the
package itself or other external. The file format is purely textual, you
Service Type: File description of the services provided by the nodes,
srv " inside the package . The file format is
ROS provides a convenient full suite
framework . In the latest versions all
rqt ".
2.1.2 Operation
topics .Per For
re-mapping , that is, give a significant value to these parameters,
re-
mapping arguments ) or alternatively use a launch files : a file with syntax
launch file may be
server parameter so that it can be modified dynamically
run-time ; typically the mapping is not among them: Once a node is
topic as a publisher or subscriber is not able to detach and
topics in the framework will inform the various subscriber so that
callback associated with the management of messages
topic . Similarly if there is a change of the parameters in the parameter
server will always be the task of the framework to inform the nodes of the new value.
services instead the system behaves with a semantics of
client / server : the master provides a list of available services,
client after a lookup on
PRC .
pub / sub is obtained by exploiting the
socket direct between publishers and subscribers with variants of protocols TCP or
Client / Server of services is
socket to exchange requests and responses; again the Master does only
packages that
framework is able to solve only the dependencies of a package
manifest.xml ) and then proceed to installation. As
tools to compile two versions of the command make :
ROSmake and catkin-make . As the command from which these also
makefile ) present within each
package , which specifies where to find the libraries and sources to compile.
2.2 MoveIt! [21] [21]
Moveit! is a framework software developed by ROS and created with the idea
framework
ROS multiple external libraries that implement state of the art
path planning and mapping, coordinating the
framework was developed in
2.1 - Diagram of a high level of the various components of the
framework, the elements in gray are external components to
MoveIt
custom to be able to easily adapt to your needs. All is
framework octomap for the part of reconstruction maps, the
OMPL (Motion Planing Open Library) for the part of path planing a
planning .Il The
ROS "Move Group" that
(Planning Scenes), the system
(Pipeline Planning) and controller (Trajectory
Execution Manager) in charge of implementing the defined trajectories. The Move
Group also has the task of receiving the commands given by users,
2.2.1 Octomap [22]
MoveIt! uses an implementation of costumizzata
Octomap framework, which implements a representation of the map
Octree and uses probabilistic functions to update it in
PointCloud and dall'odometria robot.
framework can provide a very detailed map of the places observed
Queries are also facilitated by the hierarchical structure that stores the
voxel and go to update the value of employment
voxel color.Ogni Each
MoveIt! has the objective the obtaining of
rgb relating thereto. Furthermore, to ensure modularity is
plugin to use
PointCloud.
2.2.2 OMPL (Open Motion Planning Library) [23]
open
source that implements many algorithms based on sampling for the motion
planning. The library was established for use in both research environment, both as
path planning based on sampling:
R R
self-collision . The validity of the path between two nodes neighbors occurred
Motion Validator, also only defined as
benchmark comparison it is possible to compare the
start , goals and state space, the system can determine
queries .
MoveIt! it got through the package ROS
"Ompl_ROS" adapting the API library in the world ROS. Through MoveIt! You
benchmark , without having to worry about defining
motion validator .
octomap that will be
collision checking , FCL and the PCD
2.2.3 Move Group:
Move Group is the main process of the application MoveIt !, is responsible
joint, links and sizes in
.urdf [24] standard in ROS .
tf
[25] of ROS, in which for each link is indicated which is the transformation
link to an absolute reference system fixed.In questo In this
PointCloud from rooms
1. Creation of the configuration file from the file system ".urdf"
setup assistant , which creates an automated
. A robot is
links, that is, static elements, connected by the joint ,
mesh .In In
MoveIt !, that is, sets of links and joint for which the system will plan
link that static
2. Configuration files previously produced are finished
launch file for
3. At this point everything is ready for the actual use, the user has three
plugin in the viewer ROS "Rviz" ;
query of planning by specifying start and
Goals graphically and to get a representation of planning
scenes , that is, the map reconstructed from the system with the sensor data.Per For
plugin graphic.
start and goal application calculates the distance to be traversed,
link and joint that
launch file , the trajectory execution manager that is responsible for
feedback
MoveIt !. In the case where there are problems, such as a
ROS Action : a particular interface and library ROS, made
topic, which allows to obtain a semantics of
Client / Server with non-blocking ability to monitor
client. In practice
servers will exhibit a range of topics on which it will be possible to send requests
client to perform trajectories, get feedback or cancel
topics for any new
2.3 Conclusions
ROS is derived from the large spread that it had
MoveIt! as
path planning ;
benchmark between different algorithms.
alpha test , MoveIt! represents the component that
planning within the
ROS.
Capitolo 3 Chapter 3
Ros for the simulation of
3.1 Gazebo & Hector_Quadrotor
MoveIt! he was chosen for the simulator
open source Gazebo [26], which began as official simulator integrated with the system
Ros , if they have recently posted becoming an independent component, but
Ros both for ease of integration, and especially for the large
Package for the simulation of the robot and sensors of various kinds are already created
.
plugin. What we can do is to simulate the behavior of a robot
mesh . The robot behaves exactly like that
plugins for other functions. This way you can test the full
Ros ; at the time of interfacing with the real-
package " Hector_quadrotor " [27] ,
Ros a complete simulation of a
Ros. Besides FVO itself (structure and engines), they are simulated
Extended Kalman Filter ) to obtain a single
inner loop for controlling the attitude, speed
outer loop controls the speed
[28] .
package are the presence in the model
Kinect ) chosen
PointCloud in the final design and two nodes for the issue
tf [25] is required for the low level
. The medium is
Ros, mainly through three
link : the base_link that represents the body of quadcopter, the camera_link that
base_footprint that
frame fixed
odom_combined .
Ros means publishing in
tf the transformation required to convert a given by
links to a reference system
odom_combined.
links are categorized in a hierarchical tree structure with the
link fixed, so you need to publish only the
link to its parent. For each link just then
3.1 - Tree tf relative to simulated quadcopter
package simulation, so that
MoveIt! also it needs a topic in which they are explicitly published
link as coordinates in space relative to the fixed frame, this is
plug- in gazebo.
MoveIt! all the data it needs
Hector_Quadrotor in the right format urdf , the odometry is published by
PointCloud is generated by the sensor simulated
plugin .
3.2 - Simulation of a quadcopter equipped with a Kinect inside
simulator Gazebo
3.2 Configuration
MoveIt !; these can be created through a simple graphical interface,
SetupAssistant, starting from the file ".urdf" descriptor of the robot.
joint virtual so
tf; in the case of a joint quadcopter was added
MoveGroup , ie groups of links and joint for
MoveIt! it was created for the management of any
MoveGroup separate within the same robot to generate trajectories
launch files sufficient to start the application in "mode
rgbd in our case, and
oint which controller is in charge of implementing the trajectory produced
.
plugin , so it was necessary
Ros to dialogue with the room rgbd such OpenNI and its
Ros , who deal with retrieving data from the chamber and publish
plugin.
MoveIt! provides
plug- in MOVEit !, which dialogues
planning and receives from it the trajectories
MoveGroup . According to the definition of them and any
joint to be controlled and forwards
joint .Per For
actionlib and a file action costum that allow to obtain
topic. In
action is called a kind of contract of function,
actionServer that implements the
action. They are normally defined formats topic of goal, result,
feedback, cancel and status.
SimpleControllerManager, designed as plug- in
Moveit! , and a
ActionController, realized as independent node. For the realization
ActionController was necessary to define an action because those custum
the joint with only one degree of
gripper;
action
MultiDofFollowJointTrajectoryAction and provides for the issue as a goal of a
mapping and path planning, the ' ActionController was made in a manner
actionlib, so everything is pre-configured for a realization competitor
MoveIt! .
planning
goal . Thanks to the system entirely
3.3 - Diagram final simplified communication between the nodes of the application,
ovals represent independent processes, arrows topics through which
communicate
/ move_group , node generated by MoveIt! ,
/ Robot_state_publisher and / ground_truth_to_tf on the topic / tf, and according to them
/ my_controller_node , action controller
topics / cmd_vel. Finally
/ gazebo is the process that is running the simulation itself, while
/ Quadrotor_teleop controller is the handling of the airplane. All this is visible in
topics related to data from
planning is via a
plugin viewer Rviz that allows you to select graphically
goals that the robot has to reach and which algorithm to use planning as well
3.4 - Graphical interface for query generation of planning and visualization
map rebuilt
3.3 Test application
planning among those available: Kpiece, Bidirectional Kpiece, Lazy Bidirectional
Kpiece, EST, PRM, PRM *, RRT, RRT-Connect, RRT *, SBL. Alternatively,
start, goal and situation of the planning stage.
planning geometry, are not
query, both from the point of view of
3.5 - Working environment in the simulator on the left and its reconstruction
within the plugin MoveIt! on the right
goal by
.
path planning library instead it has proved capable of
3.6 - Example of queries submitted to the system on the left (the goal is
represented by the marker orange) and trajectory of solution to the right
3.7 - Another example of a query submitted to the system and resulting path, this time
been limited work area so as not to allow the overcoming of obstacles
sorvolandoli
MoveIt! versions
alpha.
query will follow a simplification of
path planning. There being no particular restrictions
3.4 Conclusion
mapping everything works fine provided you have
path planning instead is
Chapter 4
mapping and planning in an automated system
4.1 Integration of the application in real environment
topic / tf and posting
joint of the robot on another.
topics in a format
Ros and MoveIt !, typically a PointCloud.
MoveIt! must always be
ActionServer, which will have the
joint controller
queries to the system and possibly rerun until
smoothing to make the trajectory
hardware least comparable to that of a
notebook midrange, but the physical weight of this architecture can represent
on-board only controls low-level and the
path planning could instead be
Ros, in particular by the possibility it offers to put into
RosMaster centralized,
SHERPA 4.2 (
Smart collaboration between Humans and ground-Aerial Robots for
Improving rescuing activities in Alpine environments
) [28]
animals "and have
Inteligent Donkey ): It serves as a base for operations
Sherpas box , also it allows you to charge on-site aircraft
(Trained Wasps) : Probably made
Sherpas box . They represent the "eyes
(Patrolling hawks) : Increase range
payload, and a helicopter able to replace the terrestrial vehicle
sherpa box in situation where access to the area
animals, contain all
docking / recharge / deployment station for aircraft at low altitude.
animals need to communicate to
wifi , while for communication with the components more
4.3 Possible integration in SHERPA
software presented in this thesis:
layer unifying between machines; if this
software. Use
ROS framework appears as a possible solution to the
socket, but also makes
Sherpas each
streams are then gathered to
stream of data coming from the sensors. The framework octomap allows
MoveIt! considered, the
plugin for each source of information. Publishing data
topics it is therefore possible not only to
plugin appropriate
MOVEit! thanks to its high configurability
OMPL that abstracting
MoveIt!
urdf robot.
MoveIt! overly complex: view
the start and goal. On the other hand there is only one component
ad hoc for the system using
open source with the possibility of
MoveIt! has also been developed to adapt to situations and
plugins : program components independent
s s
plug external indicated in the appropriate file
plug of planning with a
plugins of planning, so you can compare the
Ros as everything else
topic
4.4 Conclusions
complexity
Computational . With the technology currently available is very
Mobile computing will achieve
Schedule : If you want to get immediate answers to queries you
dynamics of the medium or use optimized algorithms for
mapping , however, the proposed algorithm
rgbd succeeding to
Ros and in particular of
MoveIt! is constantly increasing, and with it the number of solutions and plugins
fully self a real
planning
custom optimized for the control of this specific means, algorithms in
Bibliografia Bibliography
[1] Tabak Roth-Y, R Jain - "Building an environment model using depth information." (1989).
[2] M Hebert, Caillas C, Krotkov And, Kweon IS, Kanade T - "First Results for Terrain Mapping
[3] J Ryde, Hu H - "3D mapping with multi-resolution voxel occupied lists." (2010) Autonomus
[4] D Meagher - "Geometric modeling using octree encoding." (1982)
Computer Graphics and
Image Processing , Vol. 19, No. 2. pp. 129-147
[5] Armin Hornung, Kai M. Wurm, Maren Bennewitz, Cyrill Stachniss, Wolfram Burgard -
[6] Steven M. LaValle - "Motion Planning: The Essentials" (2011) Robotics & Automation
[7] J. Barraquand, B. Langlois, and JC Latombe - "Numerical techniques for potential field
[8] Alex Yahja, Antony (Tony) Stentz, Sanjiv Singh and Barry Brumit - "Framed-Quadtree Path
[9] Priyadarshi Bhattacharya, Marina L. Gavrilova - "Voronoi diagram in optimal path
[10] Tomás Lozano-Pérez; Wesley Michael A. - "An algorithm for planning collision-free paths
[11] THE Kavraki, Svestka P., Latombe JC, MH Overmars - "Probabilistic roadmaps for path
[12] S. Karaman and E. Frazzoli - "Sampling-based Algorithms for Optimal Motion Planning"
[13] R. Bohlin and LE Kavraki - "Path Planning Using Lazy PRM" (2000), and Robotics
[14] SM LaValle - "Rapidly-exploring random trees: A new tool for path planning." (1998)
[15] J. Kuffner and SM LaValle - "RRT-Connect: An efficient approach to single-query path
[16] G. Sánchez and JC Latombe - "A single-query bi-directional probabilistic roadmap
[17] Hart, PE; Nilsson, NJ; Raphael, B. - "A Formal Basis for the Heuristic Determination of
[18] AI Şucan and LE Kavraki - "Kinodynamic motion planning by interior-exterior cell
[19]
[21] A. Ioan Sucan and Sachin Chitta, "MoveIt!" - [Online] Available:
[22]
[23] A. Ioan Şucan, Mark Moll, Lydia E. Kavraki - "The Open Motion Planning
[25] http://wiki.ros.org/tf
[26] http://gazebosim.org
[27] http://wiki.ros.org/hector_quadrotor , Johannes Meyer and Alexander and Sendobry
[28] Naldi, Furci, Sanfelice, Marconi "Global Trajectory Tracking for Underactuated
[29] www.sherpa-project.eu/sherpa/
Original text
Contribute a better translation
Attached Files
To refer to attachments on a page, use attachment:filename, as shown below in the list of files. Do NOT use the URL of the [get] link, since this is subject to change and can break easily.You are not allowed to attach a file to this page.
