visualization.m

%% Examples of visualization
% Its better to create some robot to add everything important to path
rob = Robot('loadNAO');
%%
% Most of the functions support optional arguments. To see the options try
% to view the help for the function, e.g.
help plotCorrections;

%% Corrections
% To show corrections, you just need to provide folder with results
folder = 'nao-right-arm';
plotCorrections(folder);

%%
% The function also supporst more arguments. For example, to use length in
% mm
folder = 'nao-right-arm';
plotCorrections(folder,'units','mm');

%% Error boxplots
% Boxplots of errors work very similar, but you can pass more than one
% folder
folder = 'nao-right-arm';
folder2 = 'self-touch-motoman';
plotErrorsBoxplots({folder,folder2});

%%
% Again, the function takes more optional arguments
% E.g. to change location of the legend and use logaritmic scale
folder = 'nao-right-arm';
folder2 = 'self-touch-motoman';
plotErrorsBoxplots({folder, folder2},'log',1,'location','northeast');


%% Error bars
% Errors can be also shown as bars
folder = 'nao-right-arm';
folder2 = 'self-touch-motoman';
plotErrorBars({folder,folder2})

%% Error Residuals
% Shows residual errors using quiver plot
% We need two set of points, which can be computed for example like this

load('Results/nao-right-arm/info'); %Load saved data
load('Results/nao-right-arm/datasets'); %Load saved data
dataset = datasets_out.selftouch{1};
extPoints=dataset.refPoints';
robPoints = zeros(4, size(dataset.joints, 1));
for i = 1:size(dataset.joints,1) % Transform all point to base frame
    link = rob.findLink(dataset.frame{i});
    robPoints(:,i) =  getTFtoFrame(rob.structure.defaultKinematics, link{1}, dataset.joints(i)) *[dataset.point(i,1:3),1]';
end
% And the function itself can be called
plotErrorResiduals(extPoints, robPoints(1:3,:))


%% Joint distribution
% Shows comparison of joint distributions
% Function takes 2 datasets, where both of them are cellArray of structs
close all
folder = 'self-touch-motoman';
load(['Results/',folder,'/info']);
load(['Results/',folder,'/datasets']);
% This example does not 'make sense', it just for demonstration of input.
% Argument with 'rightArm' determines which chain to show.
plotJointDistribution(rob, {datasets_out.selftouch{1},datasets_out.projection{1}},{datasets_out.projection{1}}, 'rightArm', '', '', 2);

%%
% Alternatively you can pass just one dataset
close all
folder = 'nao-right-arm';
load(['Results/',folder,'/info']);
load(['Results/',folder,'/datasets']);
% This example shows also title and legend settings
plotJointDistribution(rob, {datasets_out.selftouch{1}},[], 'rightArm', 'myFig', {'myDataset'}, 1);

%% Joint-error plot
% shows mean errors for each joint value of given joints
close all
folder = 'self-touch-motoman';
load(['Results/',folder,'/info']);
load(['Results/',folder,'/errors']);
load(['Results/',folder,'/datasets']);
% errors must be for whole dataset (train + test together)
plotJointsError(rob, [errorsAll{1}{1}, errorsAll{9}{1}], datasets_out.selftouch, group.rightArm, 'Joint-error plot', [1,8])


%% Histogram
% shows error for folder in histogram form
folder = 'nao-right-arm';
plotErrorsHistogram(folder);

%% Jacobian
% Shows Jacobians computed in each repetition of calibration. All the
% mandatory arguments can be loaded from results
folder = 'NAO-right-arm';
plotJacobian(folder)

%% Observability indexes
% Shows observability indexes for given folder

folder = 'nao-right-arm';
plotObserIndexes(folder);

%% Dataset points
% Shows model of robot with all points from dataset

folder = 'self-touch-motoman';
load(['Results/',folder,'/info']);
load(['Results/',folder,'/datasets']);
plotDatasetPoints(rob, datasets_out);


%% Moveable model
% This model uses 'showModel' method of Robot and allows to change
% activations by 'left' and 'right' arrow keys.
% You need to have datasets from calibration, which can be easile loaded
load('Results/self-touch-motoman/info');
load('Results/self-touch-motoman/datasets');
activationsView(rob,{datasets_out.selftouch{:}})

%%
% For Nao (**only**) robot it allows to show skin and info about activations
folder = 'nao-right-arm';
load(strcat('Results/',folder, '/info'));
load(strcat('Results/',folder, '/results.mat'));
loadKinfromMat(rob, folder, 'type', 'min');
[~,~,datasets] = rob.prepareDataset(optim, chains, approach, dataset_fcn,dataset_params);
activationsView(rob,{datasets.selftouch{:}},'info',1,'skin',1)