rippleSLAM()

March 1st, 2007 by research

v0.1-Combined EKF and fullslam.

v0.2-Includes smoothing constraint

v0.4-Includes Cholesky, Levenberg-Marquardt. (works apart frm when no features obsereved at a pose…)

v0.5- Working code with the constrained formulation (Odo formulation is probably not complete-> problem with the det(Jh'*Q'Jh) being zero)

v0.6 - formulated odo to be three dimensional [u, w, 0]'

loop_for_nees()

February 21st, 2007 by research

This runs multiple runs of various SLAM algoritms and generate the average NEES and bounds.

Few modifications are needed in each slam code importantly where the read_odometry(…) called. Here give the write path to read the robot_odometry.dat. Also for each slam algo the resulting NEES per each trial is saved in the respective data directory und ***_nees_results.txt.

Also change fn: plot_stat_cov(…, data_path)

Currently on cluster:

  • drive_xyzslam() - tested with data_path='/home/dhera/_Aus2005/synced/data/rectangle_cubicle_simulation2/lc_noise_ng_134_153_065_sigma';
    xyz_nees_results.txt
  • drive_uvdslam() - tested with above data set. uvd_nees_results.txt
  • drive_UT_uvdslam_v_0_6() - tested with above data set. utf_nees_results.txt. Data set reduced to eliminate several trials with badly scaled results. (trial size 36)

fullslam3()

February 12th, 2007 by research

%—————————————————-
% damith.herath@eng.uts.edu.au
% LS Optimization - Full SLAM integrating odometry
% v0.1 - 12 Feb 2007
%
%—————————————————-

mapper3D_analyzer(path,n_of_images)

January 30th, 2007 by research

mapper3D_analyzer(path,n_of_images)
%—————————————————————
% Written by damith.herath@eng.uts.edu.au
% v0.2
%
%—————————————————————

 derivative of mapper3D, Current version uses scan_obs (from drive_uvdslam_sd) to read xxx_output.txt files and will plot individual ftr behaviour.

drive_uvdslam_sd()

January 22nd, 2007 by research

function drive_uvdslam_sd();
%—————————————————-
% uvd SLAM v0.6 with save observation data for full slam
% 03 Aug 2006
% initialisation error fixed
% all graphics handled in one routine
% uvd update only
% NEES, Q in prediction
% To work on real data (save_estimates)
% NOTE: this will drop the last non-obsrvd ftrs frm ftr_lst!!
%—————————————————-

fullslam2_nf()

January 22nd, 2007 by research

function fullslam2_nf()
%—————————————————-
% LS Optimization - Full SLAM integrating odometry
% v0.5 - 05 Jan 2007 Reads data from the new format generated from
% drive_uvdslam_sd()
% derived from fullslam v0.9 - 21 Nov 2006 on the cluster
% bug fix: Remove ftrs without ekf estimates being used in observation list
% New formulation with [v,w] as observation
% bug fix: miss-aligned covariance plot
% bug fix: turn-rate(w) error normalizing angel
%—————————————————-

drive_uvdslam()

January 16th, 2007 by research

current version
% uvd SLAM v0.6
% 03 Aug 2006
% initialisation error fixed
% all graphics handled in one routine
% uvd update only
% NEES, Q in prediction
% To work on real data (save_estimates)

Full SLAM2

January 16th, 2007 by research

This code does the LS optimisation on both observations and odometry

% LS Optimization - Full SLAM integrating odometry
% v0.4 - 05 Jan 2007
% derived from fullslam v0.9 - 21 Nov 2006 on the cluster
% bug fix: Remove ftrs without ekf estimates being used in observation list
% New formulation with [v,w] as observation
% bug fix: miss-aligned covariance plot
% bug fix: turn-rate(w) error normalizing angel


Planned books:

Current books:

  • Probabilistic Robotics (Intelligent Robotics and Autonomous Agents)

    Probabilistic Robotics (Intelligent Robotics and Autonomous Agents) by Sebastian Thrun, Wolfram Burgard, Dieter Fox Edit

Recent books:

None

View full Library