phd

CLOSED: 2012-04-20 Fri 07:14

• 需要哪些材料
• 联系方式

SCHEDULED: 2012-03-12 Mon CLOSED: 2012-03-13 Tue 10:42

• DONE put all figures in slides
  CLOSED: 2012-03-13 Tue 02:19
  

• DONE go through and struct the slides
  CLOSED: 2012-03-13 Tue 10:42
  

• DONE embed the video in slides
  CLOSED: 2012-03-13 Tue 10:42
  

• CANCELED write some notes for presentation
  CLOSED: 2012-03-13 Tue 10:42
  

• DONE create first draft slides with figures/equations
  SCHEDULED: 2012-03-11 Sun 16:15-17:15 CLOSED: 2012-03-12 Mon 12:04
  

• DONE create video / figure of kinect experiment [66%]
  SCHEDULED: 2012-03-11 Sun 17:15-18:15 CLOSED: 2012-03-13 Tue 10:42
  
  • [X] create a video from real robot
  • [ ] create a video from simulation
  • [X] create a point video from real robot
  • [X] create a point video from simulation
  • [ ] create a video from both real robot and simulation
  • [X] create a point video from both real robot and simulation
  • TODO the size of robot doesn't match!
    

draft

waiting for Ms. Tan

• DONE reply Ms. Tan:(send application and letter from Dieter)
  SCHEDULED: 2011-08-02 Tue 10:40-11:00 CLOSED: 2011-08-02 Tue 11:24
  

CLOSED: 2011-07-25 Mon 10:01

• 继续资助证明
  should be ready today

SCHEDULED: 2011-08-03 Wed 14:30-15:30 CLOSED: 2011-08-03 Wed 16:07
need some text

学生证

硕士证书、比赛成绩、发表论文清单

奖学金证明

CLOSED: 2011-07-25 Mon 10:01
Dieter

Ask Dr. Anna Brömmer

• HU template
  

• tikz
  
  • each figure in a separate file
    • use tikz2pdf for editing and preview
    • include with 'external' package

• todonotes
  
  • custom command
  • patch for working with pgf external package

• File structure
  
  • subfiles
    
    • include bibtex for each chapter
  • Makefile
    

• Other packages
  
  • cleveref
  • forloop
  • amsmath, isomath, …

• Other customized command
  
  • quote/quotation

• Editor: Emacs
  
  • forward/backward search
  • flyspell, predictive mode

• math: algin instead of eqnarray
  

• table number in appendix
  \renewcommand\thetable{A.\arabic{table}}

• Draft watermark
  

• git
• dropbox

CLOSED: 2012-02-07 Tue 17:36

• list of sensors
  

• related papers
  

CLOSED: 2012-03-14 Wed 14:12

SCHEDULED: 2012-02-13 Mon

• [X] check why the binary doesn't work in gruenau4: Because of restarting simspark in each run, the system is busy with reading files, so limit 5 process that don't make other guys crazy (The priority of my account is higher than students, so maybe my processes take over the whole system, so I use `nice +5` to start my program and limit to 5 processes.)
• [X] fix problem of fitness
  • check the lenght of file that if all data have been writen!
  • whether the old files are used?
  • the robot is just lucky! (because the random of simspark?) –> run simulation 3 times
    • the server crashed…!
    • the robot stood up in 0.5 second for 3 times!! but then I can not repeat the experiment!!
    • the server is not in 'sync' mode, set it and try again –> it doesn't matter, the robot still stands up so fast!
      • try to repeat it in 'sync' mode –> still stands too fast and can not repeat –> FIXED: problem of setting 'sync' mode
• [ ] check configuration of simspark (parameters)
  • default parameters: 742 ms
  • max speed -> nominal speed: 477 ms
  • [X] the max force for each joint!
    • [ ] the nominal force is too weak for robot
      • [ ] try to limit force in arm only
  • change ERP and CFM **
  • 改变碰撞参数
  • self collision
  • joint parameters: force, speed
• [ ] run the experiment
• [ ] check if the result is promising

SCHEDULED: 2012-02-29 Wed

• DONE create a python script to generate standing up
  SCHEDULED: 2012-02-21 Tue
  

• DONE save the logfile for creating a video later
  CLOSED: 2012-02-22 Wed 20:10
  

• DONE limit the joints range
  CLOSED: 2012-02-22 Wed 18:40
  

• DONE create a script to generate video [100%]
  CLOSED: 2012-02-29 Wed 17:42
  
  • DONE generate video from logfile
    CLOSED: 2012-02-29 Wed 00:26
    
    • grab frames from OpenGL
      
    • DONE turn frames to video
      CLOSED: 2012-02-29 Wed 00:26
      ffmpeg -r 25 -i %05d.png output.mp4
  • DONE find out logfiles with best fitness
    SCHEDULED: 2012-02-29 Wed 16:00-16:30 CLOSED: 2012-02-29 Wed 17:42
    
  • DONE create the script to generate the process of evolution (with generation number)
    SCHEDULED: 2012-02-29 Wed 16:30-17:00 CLOSED: 2012-02-29 Wed 17:42
    

• TODO better fitness function
  
  • the 'best' result just because of the server crashed!! -> check if the robot was on ground, otherwise the fitness is 0
    • the program crash because of empty file
  • The robot can only sit on the ground, but can not stand up
  • use the max height as fitness (bugfixed)
  • test fitness proposed in dissertation
    

CLOSED: 2012-03-11 Sun 13:38

• DONE read data from robot logfiles
  CLOSED: 2012-02-20 Mon 10:23
  

• DONE record long time logfile for joint sensor data and frameinfo
  CLOSED: 2012-03-03 Sat 19:48
  

• DONE system identify
  SCHEDULED: 2012-03-03 Sat 19:50-20:50 CLOSED: 2012-03-05 Mon 11:01
  
  • from NaoQi
    There are two different Devices that return a temperature: the MotorBoard, and the Battery. For both, the temperature returned in “Value” is in deg Celsius. The motor temperature is a simulated one, using electric current value of the motor. Even when the robot just starts, the temperature is quite high (50°C).The simulation starts at this value mainly due to the internal temperature value. The motor board implements a temperature limitation: as long as the simulated temperature reaches 75°C, the current limitation is decreased to stay under 85°C.

• DONE battery trajectory [100%]
  SCHEDULED: 2012-03-05 Mon 11:40-12:40 CLOSED: 2012-03-08 Thu 17:28
  
  • [X] write xabsl to repeat the behavior
  • [X] copy walk parameters from real robot and use old foot trajectory
  • [X] current feedback
    • [X] add code in perceptor (server)
    • [X] add code in agent (client)
  • [X] tempeature feedback
    • [X] add code in server
    • [X] add code in client
  • [X] set correct parameters of motors
    • [X] export function to ruby
    • [X] configure RSG files
  • [X] stiffness control
    • [X] code in server
    • [X] code in agent
  • [X] the feedback of hinge joint is not only motor force!
    • [X] AMotorEffector
    • [X] AMotorPerceptor
    • [X] using AMotor in rsg
  • [X] run the experiment and record logfile
  • [X] analysis logfile

SCHEDULED: 2012-02-23 Thu

• DONE create a static splines foot trajectory
  CLOSED: 2012-02-23 Thu 20:31
  

• DONE make the trajectory configurable
  CLOSED: 2012-02-23 Thu 20:54
  

• DONE run GA [100%]
  CLOSED: 2012-03-02 Fri 18:47
  
  • DONE battery sensoring
    CLOSED: 2012-03-01 Thu 17:03
    
  • DONE walk behavior and save the battery state to file [100%]
    SCHEDULED: 2012-03-02 Fri 16:00-17:00 CLOSED: 2012-03-02 Fri 17:02
    
    • [X] walk behavior: start and walk forward 60s and exit
    • [X] save the battery state
  • DONE create GA loop
    SCHEDULED: 2012-03-02 Fri 17:00-18:00 CLOSED: 2012-03-02 Fri 17:55
    
    • fitness = 1 - (1 - battery / distance)
  • DONE run in gruenau2 [100%]
    CLOSED: 2012-03-02 Fri 18:46
    
    • [X] install new server [100%]
      • [X] package new server
      • [X] upload
      • [X] install
    • [X] install new naoth [100%]
      • [X] package
      • [X] upload
      • [X] compile

• TODO check GA results
  
  • very strange trajectory!

SCHEDULED: 2012-03-08 Thu

• DONE learn how to program a Q learning in Python
  CLOSED: 2012-03-11 Sun 13:38
  
  • is there a Q learning lib in C++?
    

• TODO check the resulted QTable, and show if the learned policy is better in simulation
  

• TODO check the resulted policy in real robot
  

CLOSED: 2012-02-07 Tue 17:36

• choose parameters which will be ajusted automatically
  

• make the parameters ajustable
  

• do some experiement in real robot to get data
  

• do the experiement in simulation to adjust the parameters
  

• validation
  

SCHEDULED: 2012-05-17 Thu 09:00-09:30

• check the results
  
  • fix the problem: the fitness is always 0 2012-05-17 Thu
    
    • run single test: the robot should be able to stand up
      

• review my thesis, set the target pages for each chapter (or a new chapter)
  SCHEDULED: 2012-05-17 Thu 10:00
  

• State "DONE" from "" 2012-05-17 Thu 08:47

• general suggestion
  
  • English
    
  • general outline
    
  • evoluationary instead of genetic
    

• suggestion in each chapter
  

SCHEDULED: 2012-03-16 Fri CLOSED: 2012-04-04 Wed 16:03

• general suggestion
  
  • compare with other tools (webots)
    
  • English
    
  • own work in SimSpark
    
  • program, documentation (work reference to thesis)
    
  • how to make ML in SimSpark
    
  • comparsion bwteen results of ML in Sim + real ("close loop")
    

• suggestion in each chapter
  
  • extend the abstract a little
    
  • DONE chapter 1
    SCHEDULED: 2012-03-21 Wed 08:40-09:40 CLOSED: 2012-03-21 Wed 12:29
    
  • DONE chapter 2
    SCHEDULED: 2012-03-21 Wed 10:00-11:00 CLOSED: 2012-03-25 Sun 04:31
    
  • DONE chapter 3
    CLOSED: 2012-03-25 Sun 04:32
    
  • chapter 4
    
  • DONE chapter 5
    CLOSED: 2012-03-23 Fri 18:22
    

SCHEDULED: 2012-03-05 Mon 11:10-11:30 CLOSED: 2012-03-05 Mon 16:31
the reason (equation) for stiffness

/media/Data/git/Documents/Ph.D/dissertation/chapter3/better-model.tex::\urgent{the reason for stiffness}

• State "DONE" from "TODO" 2012-05-17 Thu 08:46

CLOSED: 2012-02-10 Fri 09:58

• some text from MS Robotics
  

DEADLINE: 2011-03-15 Tue CLOSED: 2012-02-07 Tue 17:33

• DONE V&V methods
  CLOSED: 2011-10-24 Mon 09:17
  
  • DONE clean/restruct the text copied(cited)
    SCHEDULED: 2011-08-03 Wed 10:00-11:00 CLOSED: 2011-08-03 Wed 11:52
    
  • DONE read through and remove some todos
    SCHEDULED: 2011-10-20 Thu 07:45-08:15 CLOSED: 2011-10-20 Thu 08:18
    
  • DONE consider which part are really necessary and make a plan for using in later sections
    SCHEDULED: 2011-10-21 Fri 07:30-08:00 CLOSED: 2011-10-21 Fri 08:17
    
  • DONE review section 1.1.2: evaluation based on system identification
    SCHEDULED: 2011-10-24 Mon 07:55-08:45 CLOSED: 2011-10-24 Mon 09:16
    

• conceptual model validation
  
  • physical model
    
    • DONE rigid body dynamcis
      SCHEDULED: 2011-05-16 Mon CLOSED: 2011-10-20 Thu 07:28
      
      • DONE finish rigid body dynamics citing(after this no new text/idea come in)
        SCHEDULED: 2011-08-02 Tue 14:30-15:30 CLOSED: 2011-08-02 Tue 22:34
        
    • engine validation experiment
      
      • CANCELED constraints -> still need to experiement
        SCHEDULED: 2011-10-20 Thu 08:15-09:15 CLOSED: 2012-02-07 Tue 17:33
        
      • DONE deterministic
        SCHEDULED: 2011-10-25 Tue 14:00-15:00 CLOSED: 2011-10-26 Wed 11:29
        
        • DONE experiement
          SCHEDULED: 2011-10-26 Wed 08:15-09:00 CLOSED: 2011-10-26 Wed 10:24
          
        • DONE add some text
          SCHEDULED: 2011-10-26 Wed 09:00-09:30 CLOSED: 2011-10-26 Wed 11:28
          
      • DONE cite some text
        SCHEDULED: 2011-08-05 Fri CLOSED: 2011-08-11 Thu 17:11
        
      • DONE run the experiement
        CLOSED: 2011-10-20 Thu 07:28
        
      • DONE write the section
        CLOSED: 2011-10-20 Thu 07:28
        
    • DONE collision
      CLOSED: 2011-10-06 Thu 09:40
      
    • DONE friction
      CLOSED: 2011-10-06 Thu 09:40
      
  • robot hardware model
    
    • sensors
      
      • DONE write some text about other sensors
        SCHEDULED: 2011-10-21 Fri 08:00-08:30 CLOSED: 2011-10-23 Sun 08:46
        
      • CANCELED reading more about sensors modeling
        CLOSED: 2012-02-10 Fri 09:50
        
    • actuators(motors)
      
      • DONE single joint test(cos), add the current test result and some text
        SCHEDULED: 2011-08-04 Thu 10:15-11:00 CLOSED: 2011-08-11 Thu 17:10
        
      • DONE do the experiment
        CLOSED: 2012-02-13 Mon 11:52
        
        • DONE run cos test again in rcssserver 2011
          SCHEDULED: 2011-10-19 Wed 07:25-07:55 CLOSED: 2011-10-19 Wed 11:32
          
        • DONE run step function in rcssserver 2011
          SCHEDULED: 2011-10-19 Wed 07:55-08:25 CLOSED: 2011-10-19 Wed 12:36
          
        • DONE how to test leg joints? –> remove ground collision
          CLOSED: 2011-10-19 Wed 11:58
          
        • DONE write some text about result in simulation
          SCHEDULED: 2011-10-19 Wed 16:00-17:00 CLOSED: 2011-10-19 Wed 17:34
          
        • DONE run experiment in real robot
          CLOSED: 2012-02-13 Mon 11:52
          
      • DONE replace the old figures
        CLOSED: 2012-02-09 Thu 10:57
        
      • DONE create a table for different type of joints
        CLOSED: 2011-10-19 Wed 15:48
        

• DONE results validation
  CLOSED: 2012-02-09 Thu 15:20
  
  • DONE Walk speed [2/2]
    CLOSED: 2012-02-09 Thu 15:20
    
    • [X] scrpit for extracting speed info from logfiles
    • [X] write some text, about algorithm and etc.
  • DONE repeat the old joint test(stand up), plot some figures and included in the thesis
    SCHEDULED: 2011-08-02 Tue 11:00-12:00 CLOSED: 2011-08-02 Tue 13:29
    
  • stand up motion in real robot
    
    • DONE prepare the code for run in the real robot
      SCHEDULED: 2011-08-04 Thu 15:00-16:00 CLOSED: 2011-12-06 Tue 12:30
      
    • DONE run some test in real robot
      SCHEDULED: 2011-08-04 Thu 16:00-17:00 CLOSED: 2011-12-06 Tue 12:30
      
    • kinect
      
      • DONE basic system frame work [2/2]
        SCHEDULED: 2011-11-05 Sat 15:00-16:00 CLOSED: 2011-11-07 Mon 10:40
        
        • [X] read/load kinect data
        • [X] read logfile data, rendering (another logplayer)
          • [X] why pyparse is so slow? -> it is
          • [X] try to load and play a game log
          • [X] make it faster -> it is still slow for 9 vs 9 game, but it should be fast enough for 1 robot
      • compare depth image
        
        • simulated kinect
          • [X] shader for depth image
          • [X] save image data
        • sync two cameras
          • time: log the data with robot's command, i.e. robot give information about time
          • camera position: use point cloud registeration at the beginning
        • comapre by image
          • [X] what kind of algorithm is avaiable
        • statistic data, plot
      • video comparsion
        
        • good calibrated kinect
        • make some cool video demo
      • advanced: gesture recognization
        
        • estimate chest rotation
        • estimate joint angle, compare
      • advanced: sync robot and kinect
        
      • DONE write the idea about using kinect
        SCHEDULED: 2011-10-25 Tue 07:00-08:00 CLOSED: 2011-10-25 Tue 09:02
        
      • DLR low-cost sensor workshop
        
        • DONE submit abstract
          DEADLINE: 2011-11-04 Fri CLOSED: 2011-11-05 Sat 11:29
          
      • DONE intergrate kinect driver in REP
        SCHEDULED: 2011-10-20 Thu 09:30-13:00 CLOSED: 2011-11-05 Sat 11:28
        
        • DONE calibration [1/1]
          SCHEDULED: 2011-10-24 Mon 12:30-14:00 CLOSED: 2011-11-04 Fri 23:58
          
          • [X] image show by matplotlib
          • single figure without main window –> Qt.exec has to be in main thread!!
          • [X] calibrate framework
          • [X] calibrate video camera of kinect
          • [X] calibrate IR camera of kinect
          • [X] calibrate relation between two cameras
          • DONE video camera
            CLOSED: 2011-11-10 Thu 06:09
            

• DONE read and restruct it in paper
  SCHEDULED: 2011-05-11 Wed CLOSED: 2011-05-16 Mon 15:03
  

CLOSED: 2012-02-10 Fri 09:58

• DONE clean old text with new structure
  SCHEDULED: 2011-11-15 Tue CLOSED: 2011-11-16 Wed 13:17
  

• DONE read paper "Extending Open Dynamics Engine for Robotics Simulation"
  SCHEDULED: 2011-08-04 Thu 11:00-12:00 CLOSED: 2011-11-15 Tue 09:17
  

CLOSED: 2012-02-10 Fri 09:58

• DONE some refercen to "reality gap" papers
  SCHEDULED: 2010-09-15 Wed CLOSED: 2010-10-04 Mon 10:30
  

• State "DONE" from "TODO" 2012-05-17 Thu 08:47

• search for physic simualtion engine
  

• download some documents and binaries of simulators(maybe some new one?)
  

• related work
  
  • UCHILSIM
    
    • UCHILSIM: A Dynamically and Visually Realistic Simulator for the RoboCup Four Legged League
      
      • the problem of evolution in real robot:
        • costs time
        • damage
      • reality gap: the difference between simulation and reality (no matter how accurate a simulator is)
        • representation by fitness
      • UCHILSIM is able to adapt its parameters automatically, by comparing robot controller behaviors in reality and in simulations.
      • there are few examples of the generation of complex robotic behaviors in a simulation with successful transfers to reality.
      • We believe that it is possible to generate accurate simulators at a level in which it is feasible to learn complex behaviors with the successful transfer of these behaviors to reality.
      • The Back to Reality paradigm combines into a single framework: learning from reality and learning from simulations.
    • Back to Reality Crossing the Reality Gap in Evolutionary Robotics
      
      • it is a promising approach for crossing the reality-gap among simulation and reality in evolutionary robotics
      • it allows to generate automatically models and theories of the real robot environment expressed as simulations
    • Learning to Kick the Ball Using Back to Reality
      
      • Our goal is to produce an optimal ball kick behavior for any given robot relative ball position and target direction that the robot might access without repositioning itself.
      • The difference in fitness obtained in simulation versus reality is minimized trough a learning process in which the simulator is adapted to reality. In other words the robot continuously generates an internal representation of their environment while adapting the simulator.
  • Minimal Simulation
    
    • Running across the reality gap: Octopod locomotion evolved in a minimal simulation
      
      • The basic idea of a minimal simulation is to model only those robot-environment interactions that are necessary to underpin a desired behavior.
      • Everything else is made unreliable by careful use of randomness. In this way an evolved controller is forced to use the minimal set of interactions picked out by the simulation designer. This set of interactions is modelled as simply (i.e. computationally cheaply) as possible by using an envelope of noise to mask the inaccuracies of the modelling. Again, by careful use of randomness, controllers evolve that are robust to these inaccuracies and hence will cross the reality gap.
    • Noise and the reality gap: The use of simulation in evolutionary robotics
      
      • Khepera robot
      • The evolved controllers were downloaded onto the real robot and the correspondence between behavior in simulation and in reality was tested.
      • It has been demonstrated that it is possible to develop successful robot controllers in simulation that generate almost identical behaviours in reality, at least for a particular class of robot-environment interaction dynamics.
      • Whereas these abstract models can be very useful in exploring some aspects of the problem of control in autonomous agents, great care must be taken in using them to draw conclusions about behavior in the real world.
      • Brooks was highly sceptical: There is a real danger (in fact, a near certainty) that programs which work well on simulated robots will completely fail on real robots because of the differences in real world sensing and actuation - it is very hard to simulate the actual dynamics of the real world.
      • it was argued that as the robot's sensory coupling with its environment becomes more complex, simulations would become extremely difficult to construct and would be slower than real time unless highly specialised hardware were available.
      • Noise also plays a part in determining the quality, The Noise Level Has to Be Right
      • great care must be taken in building the simulation and appropriate levels of noise must be included.
      • The robot-environment interactions modelled here are relatively simple.
  • Evolving Autonomous Biped Control from Simulation to Reality
    
    • Transferring an evolved control system from a simulated environment to the physical world poses a number of challenges. The difficulty in accomplishing such a task increases the complexity of the system that is being simulated and controlled increases. One of the most challenging control tasks is to generate a stable walking gait for a bipedal robot.
    • Dynamechs is an efficient rigid-body dynamic simulation library that is based on the Articulated Body Algorithm (ABA) developed by Featherstone
    • Various approaches have been tried to enable the seamless transition from simulated controllers to the physical world.
    • Jakobi argues that such approaches will always fail to transfer from simulation to reality, as the evolved controllers come to depend on certain aspects that are only present in the simulation, and are not reflected in the physical world.
    • A spline based control system is responsible for manipulating the servo inputs.
    • the problem observed in this system are believed to arise from the incomplete servo model and the shortcoming of the robot hardware.
  • Flying over the reality gap: From simulated to real indoor airships
    
    • Neuronal controllers are evolved in simulation to map visual input into motor commands in order to steer the flying robot forward as fast as possible while avoiding collisions.
    • The novelty in this first part does not lie in the mathematics of the model itself, which has been mainly adapted from fluid dynamics theory, but rather in the pragmatic methodology for swift identification of the different constitutive parameters of the model.
    • A major drawback of using artificial evolution to discover efficient robot controllers, that is the time it takes for evaluating a whole population of controllers over a number of generations (which is even more problematic with physical flying robots that may be damaged by collisions, have limited on-board energy, and have actuators performance that change over time), has been addressed in this article by demonstrating the use of physics-based simulation and successful transfer to reality.
    • However, systematic analysis of the importance of each parameter and component of the model has not been carried out and good correlation between simulation and reality has been demonstrated only within a specific navigational task where altitude was automatically kept constant.

CLOSED: 2011-05-03 Tue 12:23

CLOSED: 2011-02-23 Wed 12:05

• DONE copy text from proposal
  SCHEDULED: 2010-10-06 Wed CLOSED: 2010-10-12 Tue 11:51
  

SCHEDULED: 2010-09-11 Sat CLOSED: 2010-09-11 Sat 16:45

CLOSED: 2010-04-26 Mon 13:22

SCHEDULED: 2010-04-27 Tue 10:30 CLOSED: 2010-09-15 Wed 11:56

• robotics simulator (from Robotics Simulator in wikipedia)
  

  Simulator	Developer	Year	Licence	Physics Engine	Rendering Engine	Scene Description	Architecture	Robots	Programming Language	Application
  SimSpark	RoboCup Community	2003-	GPL	ODE	OpenGL	Ruby Scene Graph	TCP/IP	HOAP, Nao	C++	RoboCup 3D Simulation League
  SimRobot	B-Human	1994-	Open	ODE	OpenGL	XML	Monolithic	AIBO, Vehiele, Nao	C++	B-Human Nao Team
  Gazebo	researches	2007-	GPL	ODE, Physsim	OGRE	XML	TCP/IP, JINI, CORBA	Vehicle
  USARSim	University of California, University of Pittsburgh, National Institute of Standards and Technology USA	2003-	GPL	Unreal Engine	Unreal Engine	Unreal map	TCP/IP	Vehicle		RoboCup Rescue Simulation League
  OpenHRP	AIST, Japan	1998-	EPL1	own		VRML	CORBA	HRP	C, C++, Java	HRP
  Webots	Cyberbotics Ltd.	1996-	Commerical	ODE	OGRE	VRML	TCP/IP	Vehicle, Nao, ABIO, …	C++, Java, Python, Matlab
  MSRS	Microsoft	2007-	Commerical	PhysX	DirectX	XML	DSS	Arm, Vehiele, Dog, Nao	C#
  UCHILSIM	Universidad de Chile	2004	Demo only	ODE	OpenGL	VRML	Monolithic	AIBO	C++	RoboCup Four Legged League

  • features
    
    • fast robot prototyping
    • physics engine: ODE, PhysX
    • 3D rendering
    • scripting: URBI, Matlab, Python
  • Literature
    
    • OpenHRP
      
      • Open architecture humanoid robotics platform
        
        • This application doesn’t need porting of the controller thanks to the controller unification mechanism of OpenHRP. This mechanism is realized by introducing an adaptor which generalizes interfaces of the hardware body and the software body in the simulation world. Therefore, this application is done by simply replacing an adaptor for the simulated body with one for the real one.
      • OpenHRP: Open Architecture Humanoid Robotics Platform
        
        • the unification of the controllers for the simulated and real robot is realized by hardware abstraction and synchronization mechanism.
        • Thanks to the unification, the controllers can share a significant amount of software with simulator, including the parameter parser, kinematics and dynamics computation and collision detector.
        • The compatibility between the simulation and experiments is enhanced by introducing a new algorithm to simulate contact force and torque between objects. Although the analytical (or impulse-based) methods are more numerically stable and efficient than explicit methods which compute the contact force and torque, however the explicit methods have an advantage when a mechanical spring-damper is implemented on the feet of a robot such as HRP-1.
        • Design Concepts:
          • Modularity
          • Open architecture
          • Portability
    • Player/Stage/Gazebo
      
      • Player 2.0: Toward a Practical Robot Programming Framework
        
      • Design and Use Paradigms for Gazebo, An Open-Source Multi-Robot Simulator
        
      • Don't Let Player Get Stuck Increasing the Autonomy of Agents
        
      • A Player/Stage System for Context-Aware Intelligent Environments
        
  • Open Source
    
    • LpzRobots
      
    • OpenSim
      
    • Simbad
      
    • breve (Python)
      
    • Blender
      
    • Moby
      
  • Commercial
    
    • anyKode Marilou
    • Webots
    • MSRS
    • RoboLogix
    • Visual Components
    • SimplySim
    • Actin
    • Workspace

• 《帕尔格雷夫研究技巧系列-博士论文写作技巧》
  
  • 如果你在作品中只谈论你读过什么，那么没人会读你的作品
  • 很多时候，相信自己会取得很大成就是你能够取得任何成就的前提条件
  • 每当遇到难懂的文章，我们都需要做出抉择：要么认定作者无能，要么 认定我们自己愚蠢
  • 遇到好的表达方式时不要犹豫，把它拿过来以供己用
  • 对于创造性非小说作品，写作过程的实质就是一个人坐在桌子旁边，要 么周围摆满了各种信息、备注、潦草的笔记和资料，要么脑子里萦绕着 各种想法，然后挣扎着将这些想法落实到空白的屏幕或纸面上
  • 为了获得与在电脑上写作不同的视角，把你的材料打印出来。如果只在 屏幕上进行编辑工作，那么你对论文的改动只会局限于小规模的字面修 正和变化。在纸上上编辑论文时，你才会发现还可以进行更加根本的改 变，比如把大块文字移动到几页之外的位置上

• LaTeX template
  

SCHEDULED: 2010-10-15 Fri CLOSED: 2010-10-17 Sun 15:44

• Darmstadt Youth Hostel 25.0

SCHEDULED: 2010-11-01 Mon CLOSED: 2010-11-13 Sat 15:03

• [X] check
• [X] print

• DONE template (LaTeX?)
  CLOSED: 2010-11-10 Wed 10:13
  

CLOSED: 2011-01-03 Mon 15:58

• DONE prepare presentation TeX
  CLOSED: 2011-01-03 Mon 15:58
  

• DONE review the paper
  SCHEDULED: 2010-10-29 Fri CLOSED: 2010-11-01 Mon 12:31
  

CLOSED: 2010-11-05 Fri 12:06

• DONE book the ticket
  SCHEDULED: 2010-10-29 Fri CLOSED: 2010-11-01 Mon 14:16
  

SCHEDULED: 2010-08-22 Sun CLOSED: 2010-08-24 Tue 09:17

DEADLINE: 2010-07-16 Fri CLOSED: 2010-07-17 Sat 15:44

CLOSED: 2010-10-04 Mon 10:31

DEADLINE: 2010-08-15 Sun CLOSED: 2010-09-07 Tue 12:12

SCHEDULED: 2010-07-25 Sun CLOSED: 2010-09-07 Tue 12:11

SCHEDULED: 2010-07-18 Sun CLOSED: 2010-07-23 Fri 15:39