Progress for week 42 (2014)

Fra Robin

(Forskjeller mellom versjoner)
Gå til: navigasjon, søk
(Budget)
 
(8 mellomrevisjoner ikke vist.)
Linje 16: Linje 16:
=== Accounting ===
=== Accounting ===
 +
* Read the NSGA-II paper
 +
* Made a test set up in Webots that works with the robot's SDK
 +
* Discovered abilities to make uneven terrain, measure walking distance etc. in Webots
== Stian ==
== Stian ==
Linje 23: Linje 26:
=== Accounting ===
=== Accounting ===
-
 
+
* Initial parameter test framework finished.
 +
* Ran a few GA powered tests on the parameters of the libelas algorithm.
 +
* Experimented with cvkit which allows viewing the original image in 3D by using the computed depth results. Useful for subjective quality evaluation.
 +
* Started writing new test and evaluation code so as to do all the steps in a single program rather than having to communicate to the premade programs through temporary files and IPC memory.
== Rune ==
== Rune ==
Linje 32: Linje 38:
=== Accounting ===
=== Accounting ===
 +
* Working on getting the pathfinder to work in roborobo
 +
* Set up git for backup
 +
* Started reading on voronoi/navigation mesh
== Else-Line ==
== Else-Line ==
Linje 39: Linje 48:
=== Accounting ===
=== Accounting ===
 +
* Finished LearningScenario class, enabling local search in the evaluation (baldwinian)
 +
* Implemented random walk
 +
* Ran a few tests on baldwinian learning scenario on control only (simulated annealing and one plus lambda), and a few without learning for comparison (example genes on evolved robot)
== Eivind ==
== Eivind ==
Linje 47: Linje 59:
* Start writing communication protocol (?)
* Start writing communication protocol (?)
-
 
+
=== Accounting ===
 +
* Purchased new motors, gears and motorcontroller.
 +
* Rewrote program for the new controller.
 +
* Started planning the game area
== André ==
== André ==
Linje 54: Linje 69:
* Look at the use of kinect vs omnidirectional cameras
* Look at the use of kinect vs omnidirectional cameras
* Look at omnidirectional cameras for triangulation
* Look at omnidirectional cameras for triangulation
-
* Look at some imagerecognition, for recognizeing the beacons
+
* Look at some imagerecognition, for recogneizing the beacons
=== Accounting ===
=== Accounting ===
 +
* Made a shopping list for the game-board
 +
* Used matlab to find elemnts in a picture
 +
** Figured out that matlab is to slow for realtime image processing

Nåværende revisjon fra 23. okt 2014 kl. 09:29

Innhold

Student template (copy this for your entry)

Budget

  • Todo 1
  • Todo 2

Accounting

  • Done 1
  • Done 2

Snorre

Budget

  • Finish reading paper about the NSGA-II algorithm
  • Experiment with Webots, (make objects etc.)
  • Make a decision about algorithm
  • Make a decision between walking on uneven terrain, crawling, reality gap

Accounting

  • Read the NSGA-II paper
  • Made a test set up in Webots that works with the robot's SDK
  • Discovered abilities to make uneven terrain, measure walking distance etc. in Webots

Stian

Budget

  • Finish initial test framework.
  • Use framework for parameter testing on the libelas algorithm to get a better understanding of its performance.

Accounting

  • Initial parameter test framework finished.
  • Ran a few GA powered tests on the parameters of the libelas algorithm.
  • Experimented with cvkit which allows viewing the original image in 3D by using the computed depth results. Useful for subjective quality evaluation.
  • Started writing new test and evaluation code so as to do all the steps in a single program rather than having to communicate to the premade programs through temporary files and IPC memory.

Rune

Budget

  • Finish A*
  • Improve usage of A*, navigation
  • Multiple agents

Accounting

  • Working on getting the pathfinder to work in roborobo
  • Set up git for backup
  • Started reading on voronoi/navigation mesh

Else-Line

Budget

  • Continue experimenting with the simulator
  • Implement a simple local search (random walk?)

Accounting

  • Finished LearningScenario class, enabling local search in the evaluation (baldwinian)
  • Implemented random walk
  • Ran a few tests on baldwinian learning scenario on control only (simulated annealing and one plus lambda), and a few without learning for comparison (example genes on evolved robot)

Eivind

Budget

  • Order motor+gear
  • Test ROS (Robotic Operating System)
  • Start planing/building of game area
  • Start writing communication protocol (?)

Accounting

  • Purchased new motors, gears and motorcontroller.
  • Rewrote program for the new controller.
  • Started planning the game area

André

Budget

  • Start building the game-boardR
  • Look at the use of kinect vs omnidirectional cameras
  • Look at omnidirectional cameras for triangulation
  • Look at some imagerecognition, for recogneizing the beacons

Accounting

  • Made a shopping list for the game-board
  • Used matlab to find elemnts in a picture
    • Figured out that matlab is to slow for realtime image processing
Personlige verktøy