Revise the ground truth in Roh's angulation dataset

Author: sunglok

benchmark_result/run_eval_roh(bearing,2d)/ex1_distribution.png

@@ -1,15 +1,15 @@
 ## MRCLAM Dataset
 UTIAS Multi-Robot Cooperative Localization and Mapping Dataset (a.k.a. MRCLAM dataset) consists of 9 sets of data including odometry, distance and bearing measurements of 15 landmarks from 5 robots. The dataset originally aims cooperative localization and SLAM, but we utilize it for landmark-based localization from instantaneously available observation.
 
-We convert MRCLAM dataset to easily apply to landmark-based localization. The instantaneously available data are defined as all measurements acquired during very small motion (0.01 meters and 0.01 radians) and at most 1 second. We reject the observation whose number of measurements is less than 3. Outlier measurements are also removed during conversion. Its conversion for each set is presented in [`run_conv_mrclam.m`](https://github.com/SunglokChoi/Triangulation-Toolbox/blob/master/run_conv_mrclam.m). The converted sets of data are written in MATLAB MAT files such as `mrclam1.mat`.
+We convert MRCLAM dataset to easily apply to landmark-based localization. The instantaneously available data are extracted as observed measurements during very small motion (0.01 meters and 0.01 radians) and short time (at most 1 second). We reject the observation whose number of measurements is less than 3. Outlier measurements are also removed during conversion. Its conversion for each set is presented in [`run_conv_mrclam.m`](https://github.com/SunglokChoi/Triangulation-Toolbox/blob/master/run_conv_mrclam.m). The converted sets of data are written in MATLAB MAT files such as `mrclam1.mat`.
 
 You can get more detail on MRCLAM dataset from its website and paper [1]. Please cite their paper if you use this dataset.
 
 #### Landmark Map, Ground Truth, and Measurements
-The data files, _mrclam?.mat_, contains three variables: _landmark_, _groundtruth_, and _measurement_. The variable _landmark_ contains 15 lines of information with 3 columns which sequentially mean landmark ID and 2D position. In contrast to the original dataset, the landmark IDs are rearranged to begin with 1. The variable _groundtruth_ contains 4 columns of data which are experiment ID and true 2D pose (position and orientation) of the robot. The experiment IDs start from 1 and its number is same with the number of lines. The variable _measurement_ includes 4 columns of data which represent the experiment ID, observed landmark ID, and its range and bearing measurements. Observations with same experiment ID mean a single set of experiments. Units of all values are meter for distance and radian for angle. In overall, three variables are aligned as follows.
+The data files, `mrclam?.mat`, contains three variables: _landmark_, _groundtruth_, and _measurement_. The variable _landmark_ contains 15 lines of information with 3 columns which sequentially mean landmark ID and 2D position. In contrast to the original dataset, the landmark IDs are rearranged to begin with 1. The variable _groundtruth_ contains 4 columns of data which are experiment ID and true 2D pose (position and orientation) of the robot. The experiment IDs start from 1 and its number is same with the number of lines. The variable _measurement_ includes 4 columns of data which represent the experiment ID, observed landmark ID, and its range and bearing measurements. Observations with same experiment ID mean a single set of experiments. Units of all values are meter for distance and radian for angle. In overall, three variables contains the following.
  * _landmark_: landmark ID, position X, position Y (of each landmark)
  * _groundtruth_: experiment ID, position X, position Y, orientation (of each robot's true pose)
  * _measurement_: experiment ID, observed landmark ID, observed distance, observed bearing (from each robot's true pose to a landmark)
 
 #### Reference
- * [1] K. Leung, Y. Halpern, T. Barfoot, and H. Liu, __The UTIAS Multi-Robot Cooperative Localization and Mapping Dataset__, The International Journal of Robotics Research, Vol. 30, No. 8, 2011 [Website](http://asrl.utias.utoronto.ca/datasets/mrclam/) [IJRR Online](http://ijr.sagepub.com/content/30/8/969)
+ * [1] K. Leung, Y. Halpern, T. Barfoot, and H. Liu, __The UTIAS Multi-Robot Cooperative Localization and Mapping Dataset__, The International Journal of Robotics Research, Vol. 30, No. 8, 2011 [IJRR Online](http://ijr.sagepub.com/content/30/8/969) [Website](http://asrl.utias.utoronto.ca/datasets/mrclam/)

benchmark_result/run_eval_roh(bearing,2d)/ex1_orientation.png

@@ -1,6 +1,6 @@
 ## Roh's Angulation Dataset
 
-_Roh's angulation dataset_ is a set of measurements acquired from an IR detector and four IRED landmarks. The sensor system aims at 2D bearing-based localization, which is described in Roh's thesis [1] and paper [2] in detail.
+_Roh's angulation dataset_ is a set of measurements acquired from a rotating IR detector and four IRED landmarks. The sensor system aims at 2D bearing-based localization, which is described in Roh's thesis [1] and paper [2] in detail.
 
 #### Landmark Map
 The map file, _landmark.csv_, contains position of four landmarks.

benchmark_result/run_eval_roh(bearing,2d)/ex1_position.png

@@ -1,5 +1,5 @@
 function [rad] = error_orientation(p, q)
-%ERROR_ORIENTATION  Calculate angular difference between two orientations.
+%ERROR_ORIENTATION  Calculate angular difference between two sets of Euler angles.
 %
 %   D = ERROR_ORIENTATION(P, Q)
 %       (matrix) P: An Euler angle (1x3 matrix)

dataset_mrclam/README.md

@@ -19,15 +19,15 @@
 };
 config.pose = ... % A list of ground truth
 [                            ...
-    1.5, 1.5, 0, 0, 0, pi/2; ...
-    1.5, 3.0, 0, 0, 0, pi/2; ...
-    1.5, 4.5, 0, 0, 0, pi/2; ...
-    3.0, 1.5, 0, 0, 0, pi/2; ...
-    3.0, 3.0, 0, 0, 0, pi/2; ...
-    3.0, 4.5, 0, 0, 0, pi/2; ...
-    4.5, 1.5, 0, 0, 0, pi/2; ...
-    4.5, 3.0, 0, 0, 0, pi/2; ...
-    4.5, 4.5, 0, 0, 0, pi/2; ...
+    1.50, 1.44, 0, 0, 0, tran_deg2rad(90.7); ...
+    1.50, 2.92, 0, 0, 0, tran_deg2rad(90.9); ...
+    1.58, 4.54, 0, 0, 0, tran_deg2rad(85.5); ...
+    3.01, 1.43, 0, 0, 0, tran_deg2rad(87.1); ...
+    3.05, 2.93, 0, 0, 0, tran_deg2rad(88.9); ...
+    3.09, 4.55, 0, 0, 0, tran_deg2rad(91.3); ...
+    4.50, 1.39, 0, 0, 0, tran_deg2rad(88.5); ...
+    4.53, 2.98, 0, 0, 0, tran_deg2rad(89.9); ...
+    4.51, 4.47, 0, 0, 0, tran_deg2rad(90.5); ...
 ];
 config.algorithm = ... % Description of localization algorithms
 {                                                                                                         ...

dataset_mrclam/noise_bearing.png

@@ -43,7 +43,7 @@
 fprintf(' %.1f [deg]\n', tran_rad2deg(errOri));
 
 % Visualize results
-figure();
+figure('Color', [1, 1, 1]);
 hold on;
     plot3(trueMap(:,1), trueMap(:,2), trueMap(:,3), 'b.', 'MarkerSize', 20);  % All landmarks
     plot3(obsMap(:,1), obsMap(:,2), obsMap(:,3), 'ro', 'LineWidth', 2);       % The observed landmarks

dataset_mrclam/noise_distance.png

@@ -103,10 +103,10 @@
 % observe_bearing
 disp('==== observe_bearing ====');
 [obsData, obsMap] = observe_bearing(map, pose);
-test_is_near(obsData(1,:), [tran_deg2rad( 135), acos(-5 / sqrt(75))]);
-test_is_near(obsData(2,:), [tran_deg2rad(-180), tran_deg2rad(  90)]);
-test_is_near(obsData(3,:), [tran_deg2rad(   0), tran_deg2rad( 180)]);
-test_is_near(obsData(4,:), [tran_deg2rad(  90), tran_deg2rad(  90)]);
+test_is_near(obsData(1,:), [tran_deg2rad( 135), -acos(sqrt(2/3))]);
+test_is_near(obsData(2,:), [tran_deg2rad(-180),       0]);
+test_is_near(obsData(3,:), [tran_deg2rad(   0), -pi / 2]);
+test_is_near(obsData(4,:), [tran_deg2rad(  90),       0]);
 
 % observe_pose
 disp('==== observe_pose ====');

dataset_roh/README.md

@@ -1,9 +1,9 @@
 function [result] = test_is_near(a, b, tol, verbose)
-%TEST_IS_TRUE  Check whether the given two values are near or not
+%TEST_IS_TRUE  Check whether two given values are near or not
 %
 %   RESULT = TEST_IS_NEAR(A, B, TOL, VERBOSE)
-%       (matrix) A      : The given value
-%       (matrix) B      : The given value
+%       (matrix) A      : The 1st given value
+%       (matrix) B      : The 2nd given value
 %       (scalar) TOL    : Tolerance (default: 1.0e-8)
 %       (scalar) VERBOSE: A flag to print its result (default: true)
 %       (scalar) RESULT : The test result
@@ -15,7 +15,7 @@
 %       t = test_is_near(pi, pi + eps)
 %       t = test_is_near([3, 29], [3 + eps, 29 - eps])
 %
-%   See also test_is_near.
+%   See also test_is_true.
 
 if nargin < 3
     tol = 1e-8;