14 Commits

134 changed files with 450 additions and 1679 deletions

3
.gitignore vendored
View File

@@ -45,6 +45,3 @@ sandbox/*
# Videos
*.mp4
*.avi
# Figures
*.fig

View File

@@ -1,14 +1,22 @@
classdef agent
properties (SetAccess = public, GetAccess = public)
properties (SetAccess = private, GetAccess = public)
% Identifiers
index = NaN;
label = "";
% Sensor
sensorModel;
sensingLength = 0.05; % length parameter used by sensing function
% Guidance
guidanceModel;
% State
lastPos = NaN(1, 3); % position from previous timestep
pos = NaN(1, 3); % current position
% Sensor
sensorModel;
vel = NaN(1, 3); % current velocity
pan = NaN; % pan angle
tilt = NaN; % tilt angle
% Collision
collisionGeometry;
@@ -17,26 +25,15 @@ classdef agent
fovGeometry;
% Communication
commsGeometry = spherical;
lesserNeighbors = [];
% Performance
performance = 0;
comRange = NaN;
% Plotting
scatterPoints;
plotCommsGeometry = true;
end
properties (SetAccess = private, GetAccess = public)
initialStepSize = NaN;
stepDecayRate = NaN;
end
methods (Access = public)
[obj] = initialize(obj, pos, pan, tilt, collisionGeometry, sensorModel, guidanceModel, comRange, index, label);
[obj] = run(obj, domain, partitioning, t, index, agents);
[partitioning] = partition(obj, agents, objective)
[obj] = initialize(obj, pos, vel, pan, tilt, collisionGeometry, sensorModel, guidanceModel, comRange, index, label);
[obj] = run(obj, sensingObjective, domain, partitioning);
[obj, f] = plot(obj, ind, f);
updatePlots(obj);
end

View File

@@ -1,34 +1,33 @@
function obj = initialize(obj, pos, collisionGeometry, sensorModel, comRange, maxIter, initialStepSize, label, plotCommsGeometry)
function obj = initialize(obj, pos, vel, pan, tilt, collisionGeometry, sensorModel, guidanceModel, comRange, index, label)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'agent')};
pos (1, 3) double;
vel (1, 3) double;
pan (1, 1) double;
tilt (1, 1) double;
collisionGeometry (1, 1) {mustBeGeometry};
sensorModel (1, 1) {mustBeSensor};
comRange (1, 1) double;
maxIter (1, 1) double;
initialStepSize (1, 1) double = 0.2;
sensorModel (1, 1) {mustBeSensor}
guidanceModel (1, 1) {mustBeA(guidanceModel, 'function_handle')};
comRange (1, 1) double = NaN;
index (1, 1) double = NaN;
label (1, 1) string = "";
plotCommsGeometry (1, 1) logical = false;
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'agent')};
end
obj.pos = pos;
obj.vel = vel;
obj.pan = pan;
obj.tilt = tilt;
obj.collisionGeometry = collisionGeometry;
obj.sensorModel = sensorModel;
obj.guidanceModel = guidanceModel;
obj.comRange = comRange;
obj.index = index;
obj.label = label;
obj.plotCommsGeometry = plotCommsGeometry;
obj.initialStepSize = initialStepSize;
obj.stepDecayRate = obj.initialStepSize / maxIter;
% Initialize performance vector
obj.performance = [0, NaN(1, maxIter), 0];
% Add spherical geometry based on com range
obj.commsGeometry = obj.commsGeometry.initialize(obj.pos, comRange, REGION_TYPE.COMMS, sprintf("%s Comms Geometry", obj.label));
% Initialize FOV cone
obj.fovGeometry = cone;
obj.fovGeometry = obj.fovGeometry.initialize([obj.pos(1:3)], tand(obj.sensorModel.alphaTilt) * obj.pos(3), obj.pos(3), REGION_TYPE.FOV, sprintf("%s FOV", obj.label));
obj.fovGeometry = obj.fovGeometry.initialize([obj.pos(1:2), 0], tan(obj.sensorModel.alphaTilt) * obj.pos(3), obj.pos(3), REGION_TYPE.FOV, sprintf("%s FOV", obj.label));
end

View File

@@ -1,35 +0,0 @@
function [partitioning] = partition(obj, agents, objective)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'agent')};
agents (:, 1) {mustBeA(agents, 'cell')};
objective (1, 1) {mustBeA(objective, 'sensingObjective')};
end
arguments (Output)
partitioning (:, :) double;
end
% Assess sensing performance of each agent at each sample point
% in the domain
agentPerformances = cellfun(@(x) reshape(x.sensorModel.sensorPerformance(x.pos, [objective.X(:), objective.Y(:), zeros(size(objective.X(:)))]), size(objective.X)), agents, 'UniformOutput', false);
agentPerformances{end + 1} = objective.sensorPerformanceMinimum * ones(size(agentPerformances{end})); % add additional layer to represent the threshold that has to be cleared for assignment to any partiton
agentPerformances = cat(3, agentPerformances{:});
% Get highest performance value at each point
[~, idx] = max(agentPerformances, [], 3);
% Collect agent indices in the same way as performance
indices = 1:size(agents, 1);
agentInds = squeeze(tensorprod(indices, ones(size(objective.X))));
if size(agentInds, 1) ~= size(agents, 1)
agentInds = reshape(agentInds, [size(agents, 1), size(agentInds)]); % needed for cases with 1 agent where prior squeeze is too agressive
end
agentInds = num2cell(agentInds, 2:3);
agentInds = cellfun(@(x) squeeze(x), agentInds, 'UniformOutput', false);
agentInds{end + 1} = zeros(size(agentInds{end})); % index for no assignment
agentInds = cat(3, agentInds{:});
% Use highest performing agent's index to form partitions
[m, n, ~] = size(agentInds);
[jj, kk] = ndgrid(1:m, 1:n);
partitioning = agentInds(sub2ind(size(agentInds), jj, kk, idx));
end

View File

@@ -30,12 +30,6 @@ function [obj, f] = plot(obj, ind, f)
% Plot collision geometry
[obj.collisionGeometry, f] = obj.collisionGeometry.plotWireframe(ind, f);
% Plot communications geometry
if obj.plotCommsGeometry
[obj.commsGeometry, f] = obj.commsGeometry.plotWireframe(ind, f);
end
% Plot FOV geometry
maxAlt = f.Children(1).Children(end).ZLim(2); % to avoid scaling the FOV geometry as the sim runs, let's just make it really big and hide the excess under the floor of the domain. Check the domain altitude to figure out how big it needs to be to achieve this deception.
[obj.fovGeometry, f] = obj.fovGeometry.plot(ind, f, maxAlt);
[obj.fovGeometry, f] = obj.fovGeometry.plot(ind, f);
end

View File

@@ -1,92 +1,28 @@
function obj = run(obj, domain, partitioning, timestepIndex, index, agents)
function obj = run(obj, sensingObjective, domain, partitioning)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'agent')};
sensingObjective (1, 1) {mustBeA(sensingObjective, 'sensingObjective')};
domain (1, 1) {mustBeGeometry};
partitioning (:, :) double;
timestepIndex (1, 1) double;
index (1, 1) double;
agents (:, 1) {mustBeA(agents, 'cell')};
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'agent')};
end
% Collect objective function values across partition
partitionMask = partitioning == index;
if ~unique(partitionMask)
% This agent has no partition, maintain current state
return;
end
objectiveValues = domain.objective.values(partitionMask); % f(omega) on W_n
% Do sensing
[sensedValues, sensedPositions] = obj.sensorModel.sense(obj, sensingObjective, domain, partitioning);
% Compute sensor performance on partition
maskedX = domain.objective.X(partitionMask);
maskedY = domain.objective.Y(partitionMask);
% Compute agent performance at the current position and each delta position +/- X, Y, Z
delta = domain.objective.discretizationStep; % smallest possible step size that gets different results
deltaApplicator = [0, 0, 0; 1, 0, 0; -1, 0, 0; 0, 1, 0; 0, -1, 0; 0, 0, 1; 0, 0, -1]; % none, +X, -X, +Y, -Y, +Z, -Z
C_delta = NaN(7, 1); % agent performance at delta steps in each direction
for ii = 1:7
% Apply delta to position
pos = obj.pos + delta * deltaApplicator(ii, 1:3);
% Compute performance values on partition
if ii < 5
% Compute sensing performance
sensorValues = obj.sensorModel.sensorPerformance(pos, [maskedX, maskedY, zeros(size(maskedX))]); % S_n(omega, P_n) on W_n
% Objective performance does not change for 0, +/- X, Y steps.
% Those values are computed once before the loop and are only
% recomputed when +/- Z steps are applied
else
% Redo partitioning for Z stepping only
partitioning = obj.partition(agents, domain.objective);
% Recompute partiton-derived performance values for objective
partitionMask = partitioning == index;
objectiveValues = domain.objective.values(partitionMask); % f(omega) on W_n
% Recompute partiton-derived performance values for sensing
maskedX = domain.objective.X(partitionMask);
maskedY = domain.objective.Y(partitionMask);
sensorValues = obj.sensorModel.sensorPerformance(pos, [maskedX, maskedY, zeros(size(maskedX))]); % S_n(omega, P_n) on W_n
end
% Rearrange data into image arrays
F = NaN(size(partitionMask));
F(partitionMask) = objectiveValues;
S = NaN(size(partitionMask));
S(partitionMask) = sensorValues;
% Compute agent performance
C = S .* F;
C_delta(ii) = sum(C(~isnan(C)));
end
% Store agent performance at current time and place
obj.performance(timestepIndex + 1) = C_delta(1);
% Compute gradient by finite central differences
gradC = [(C_delta(2)-C_delta(3))/(2*delta), (C_delta(4)-C_delta(5))/(2*delta), (C_delta(6)-C_delta(7))/(2*delta)];
% Compute scaling factor
targetRate = obj.initialStepSize - obj.stepDecayRate * timestepIndex; % slow down as you get closer
rateFactor = targetRate / norm(gradC);
% Compute unconstrained next position
pNext = obj.pos + rateFactor * gradC;
% Determine next planned position
nextPos = obj.guidanceModel(sensedValues, sensedPositions, obj.pos);
% Move to next position
% (dynamics not modeled at this time)
obj.lastPos = obj.pos;
obj.pos = pNext;
obj.pos = nextPos;
% Calculate movement
d = obj.pos - obj.collisionGeometry.center;
% Reinitialize collision geometry in the new position
d = obj.pos - obj.collisionGeometry.center;
if isa(obj.collisionGeometry, 'rectangularPrism')
obj.collisionGeometry = obj.collisionGeometry.initialize([obj.collisionGeometry.minCorner; obj.collisionGeometry.maxCorner] + d, obj.collisionGeometry.tag, obj.collisionGeometry.label);
elseif isa(obj.collisionGeometry, 'spherical')
obj.collisionGeometry = obj.collisionGeometry.initialize(obj.collisionGeometry.center + d, obj.collisionGeometry.radius, obj.collisionGeometry.tag, obj.collisionGeometry.label);
else
error("?");
end
obj.collisionGeometry = obj.collisionGeometry.initialize([obj.collisionGeometry.minCorner; obj.collisionGeometry.maxCorner] + d, obj.collisionGeometry.tag, obj.collisionGeometry.label);
end

View File

@@ -5,13 +5,6 @@ function updatePlots(obj)
arguments (Output)
end
% Find change in agent position since last timestep
deltaPos = obj.pos - obj.lastPos;
if all(isnan(deltaPos)) || all(deltaPos == zeros(1, 3))
% Agent did not move, so nothing has to move on the plots
return;
end
% Scatterplot point positions
for ii = 1:size(obj.scatterPoints, 1)
obj.scatterPoints(ii).XData = obj.pos(1);
@@ -19,6 +12,9 @@ function updatePlots(obj)
obj.scatterPoints(ii).ZData = obj.pos(3);
end
% Find change in agent position since last timestep
deltaPos = obj.pos - obj.lastPos;
% Collision geometry edges
for jj = 1:size(obj.collisionGeometry.lines, 2)
% Update plotting
@@ -29,21 +25,9 @@ function updatePlots(obj)
end
end
% Communications geometry edges
if obj.plotCommsGeometry
for jj = 1:size(obj.commsGeometry.lines, 2)
for ii = 1:size(obj.collisionGeometry.lines(:, jj), 1)
obj.collisionGeometry.lines(ii, jj).XData = obj.collisionGeometry.lines(ii, jj).XData + deltaPos(1);
obj.collisionGeometry.lines(ii, jj).YData = obj.collisionGeometry.lines(ii, jj).YData + deltaPos(2);
obj.collisionGeometry.lines(ii, jj).ZData = obj.collisionGeometry.lines(ii, jj).ZData + deltaPos(3);
end
end
end
% Update FOV geometry surfaces
for jj = 1:size(obj.fovGeometry.surface, 2)
% Update each plot
% obj.fovGeometry = obj.fovGeometry.plot(obj.spatialPlotIndices)
obj.fovGeometry.surface(jj).XData = obj.fovGeometry.surface(jj).XData + deltaPos(1);
obj.fovGeometry.surface(jj).YData = obj.fovGeometry.surface(jj).YData + deltaPos(2);
obj.fovGeometry.surface(jj).ZData = obj.fovGeometry.surface(jj).ZData + deltaPos(3);

View File

@@ -1,154 +0,0 @@
function [obj] = constrainMotion(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
if size(obj.agents, 1) < 2
nAAPairs = 0;
else
nAAPairs = nchoosek(size(obj.agents, 1), 2); % unique agent/agent pairs
end
agents = [obj.agents{:}];
v = reshape(([agents.pos] - [agents.lastPos])./obj.timestep, 3, size(obj.agents, 1))';
if all(isnan(v), 'all') || all(v == zeros(size(obj.agents, 1), 3), 'all')
% Agents are not attempting to move, so there is no motion to be
% constrained
return;
end
% Initialize QP based on number of agents and obstacles
nAOPairs = size(obj.agents, 1) * size(obj.obstacles, 1); % unique agent/obstacle pairs
nADPairs = size(obj.agents, 1) * 5; % agents x (4 walls + 1 ceiling)
nLNAPairs = sum(obj.constraintAdjacencyMatrix, 'all') - size(obj.agents, 1);
total = nAAPairs + nAOPairs + nADPairs + nLNAPairs;
kk = 1;
A = zeros(total, 3 * size(obj.agents, 1));
b = zeros(total, 1);
% Set up collision avoidance constraints
h = NaN(size(obj.agents, 1));
h(logical(eye(size(obj.agents, 1)))) = 0; % self value is 0
for ii = 1:(size(obj.agents, 1) - 1)
for jj = (ii + 1):size(obj.agents, 1)
h(ii, jj) = norm(agents(ii).pos - agents(jj).pos)^2 - (agents(ii).collisionGeometry.radius + agents(jj).collisionGeometry.radius)^2;
h(jj, ii) = h(ii, jj);
A(kk, (3 * ii - 2):(3 * ii)) = -2 * (agents(ii).pos - agents(jj).pos);
A(kk, (3 * jj - 2):(3 * jj)) = -A(kk, (3 * ii - 2):(3 * ii));
b(kk) = obj.barrierGain * h(ii, jj)^obj.barrierExponent;
kk = kk + 1;
end
end
hObs = NaN(size(obj.agents, 1), size(obj.obstacles, 1));
% Set up obstacle avoidance constraints
for ii = 1:size(obj.agents, 1)
for jj = 1:size(obj.obstacles, 1)
% find closest position to agent on/in obstacle
cPos = obj.obstacles{jj}.closestToPoint(agents(ii).pos);
hObs(ii, jj) = dot(agents(ii).pos - cPos, agents(ii).pos - cPos) - agents(ii).collisionGeometry.radius^2;
A(kk, (3 * ii - 2):(3 * ii)) = -2 * (agents(ii).pos - cPos);
b(kk) = obj.barrierGain * hObs(ii, jj)^obj.barrierExponent;
kk = kk + 1;
end
end
% Set up domain constraints (walls and ceiling only)
% Floor constraint is implicit with an obstacle corresponding to the
% minimum allowed altitude, but I included it anyways
for ii = 1:size(obj.agents, 1)
% X minimum
h_xMin = (agents(ii).pos(1) - obj.domain.minCorner(1)) - agents(ii).collisionGeometry.radius;
A(kk, (3 * ii - 2):(3 * ii)) = [-1, 0, 0];
b(kk) = obj.barrierGain * h_xMin^obj.barrierExponent;
kk = kk + 1;
% X maximum
h_xMax = (obj.domain.maxCorner(1) - agents(ii).pos(1)) - agents(ii).collisionGeometry.radius;
A(kk, (3 * ii - 2):(3 * ii)) = [1, 0, 0];
b(kk) = obj.barrierGain * h_xMax^obj.barrierExponent;
kk = kk + 1;
% Y minimum
h_yMin = (agents(ii).pos(2) - obj.domain.minCorner(2)) - agents(ii).collisionGeometry.radius;
A(kk, (3 * ii - 2):(3 * ii)) = [0, -1, 0];
b(kk) = obj.barrierGain * h_yMin^obj.barrierExponent;
kk = kk + 1;
% Y maximum
h_yMax = (obj.domain.maxCorner(2) - agents(ii).pos(2)) - agents(ii).collisionGeometry.radius;
A(kk, (3 * ii - 2):(3 * ii)) = [0, 1, 0];
b(kk) = obj.barrierGain * h_yMax^obj.barrierExponent;
kk = kk + 1;
% Z minimum
h_zMin = (agents(ii).pos(3) - obj.domain.minCorner(3)) - agents(ii).collisionGeometry.radius;
A(kk, (3 * ii - 2):(3 * ii)) = [0, 0, -1];
b(kk) = obj.barrierGain * h_zMin^obj.barrierExponent;
kk = kk + 1;
% Z maximum
h_zMax = (obj.domain.maxCorner(2) - agents(ii).pos(2)) - agents(ii).collisionGeometry.radius;
A(kk, (3 * ii - 2):(3 * ii)) = [0, 0, 1];
b(kk) = obj.barrierGain * h_zMax^obj.barrierExponent;
kk = kk + 1;
end
% Save off h function values (ignoring network constraints which may evolve in time)
obj.h(:, obj.timestepIndex) = [h(triu(true(size(obj.agents, 1)), 1)); reshape(hObs, [], 1); h_xMin; h_xMax; h_yMin; h_yMax; h_zMin; h_zMax;];
% Add communication network constraints
hComms = NaN(size(obj.agents, 1));
hComms(logical(eye(size(obj.agents, 1)))) = 0;
for ii = 1:(size(obj.agents, 1) - 1)
for jj = (ii + 1):size(obj.agents, 1)
if obj.constraintAdjacencyMatrix(ii, jj)
hComms(ii, jj) = min([obj.agents{ii}.commsGeometry.radius, obj.agents{jj}.commsGeometry.radius])^2 - norm(agents(ii).pos - agents(jj).pos)^2;
A(kk, (3 * ii - 2):(3 * ii)) = 2 * (agents(ii).pos - agents(jj).pos);
A(kk, (3 * jj - 2):(3 * jj)) = -A(kk, (3 * ii - 2):(3 * ii));
b(kk) = obj.barrierGain * hComms(ii, jj)^obj.barrierExponent;
kk = kk + 1;
end
end
end
% Solve QP program generated earlier
vhat = reshape(v', 3 * size(obj.agents, 1), 1);
H = 2 * eye(3 * size(obj.agents, 1));
f = -2 * vhat;
% Update solution based on constraints
assert(size(A,2) == size(H,1))
assert(size(A,1) == size(b,1))
assert(size(H,1) == length(f))
opt = optimoptions('quadprog', 'Display', 'off');
[vNew, ~, exitflag, m] = quadprog(sparse(H), double(f), A, b, [],[], [], [], [], opt);
assert(exitflag == 1, sprintf('quadprog failure... %s%s', newline, m.message));
vNew = reshape(vNew, 3, size(obj.agents, 1))';
if exitflag <= 0
warning("QP failed, continuing with unconstrained solution...")
vNew = v;
end
% Update the "next position" that was previously set by unconstrained
% GA using the constrained solution produced here
for ii = 1:size(vNew, 1)
obj.agents{ii}.pos = obj.agents{ii}.lastPos + vNew(ii, :) * obj.timestep;
end
% Here we run this at the simulation level, but in reality there is no
% parent level, so this would be run independently on each agent.
% Running at the simulation level is just meant to simplify the
% simulation
end

View File

@@ -1,97 +1,49 @@
function [obj] = initialize(obj, domain, agents, barrierGain, barrierExponent, minAlt, timestep, maxIter, obstacles, makePlots, makeVideo)
function obj = initialize(obj, domain, objective, agents, timestep, partitoningFreq, maxIter, obstacles)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
domain (1, 1) {mustBeGeometry};
objective (1, 1) {mustBeA(objective, 'sensingObjective')};
agents (:, 1) cell;
barrierGain (1, 1) double = 100;
barrierExponent (1, 1) double = 3;
minAlt (1, 1) double = 1;
timestep (:, 1) double = 0.05;
partitoningFreq (:, 1) double = 0.25
maxIter (:, 1) double = 1000;
obstacles (:, 1) cell {mustBeGeometry} = cell(0, 1);
makePlots(1, 1) logical = true;
makeVideo (1, 1) logical = true;
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
% enable/disable plotting and video writer
obj.makePlots = makePlots;
if ~obj.makePlots
if makeVideo
warning("makeVideo set to true, but makePlots set to false. Setting makeVideo to false.");
makeVideo = false;
end
end
obj.makeVideo = makeVideo;
% Generate artifact(s) name
obj.artifactName = strcat(string(datetime('now'), 'yyyy_MM_dd_HH_mm_ss'));
% Define simulation time parameters
obj.timestep = timestep;
obj.timestepIndex = 0;
obj.maxIter = maxIter - 1;
% Define domain
obj.domain = domain;
obj.partitioningFreq = partitoningFreq;
% Add geometries representing obstacles within the domain
obj.obstacles = obstacles;
% Add an additional obstacle spanning the domain's footprint to
% represent the minimum allowable altitude
if minAlt > 0
obj.obstacles{end + 1, 1} = rectangularPrism;
obj.obstacles{end, 1} = obj.obstacles{end, 1}.initialize([obj.domain.minCorner; obj.domain.maxCorner(1:2), minAlt], "OBSTACLE", "Minimum Altitude Domain Constraint");
end
% Define objective
obj.objective = objective;
% Define agents
obj.agents = agents;
obj.constraintAdjacencyMatrix = logical(eye(size(agents, 1)));
% Set labels for agents and collision geometries in cases where they
% were not provieded at the time of their initialization
for ii = 1:size(obj.agents, 1)
% Agent
if isempty(char(obj.agents{ii}.label))
obj.agents{ii}.label = sprintf("Agent %d", ii);
end
% Collision geometry
if isempty(char(obj.agents{ii}.collisionGeometry.label))
obj.agents{ii}.collisionGeometry.label = sprintf("Agent %d Collision Geometry", ii);
end
end
% Set CBF parameters
obj.barrierGain = barrierGain;
obj.barrierExponent = barrierExponent;
% Compute adjacency matrix and lesser neighbors
% Compute adjacency matrix
obj = obj.updateAdjacency();
obj = obj.lesserNeighbor();
% Set up times to iterate over
obj.times = linspace(0, obj.timestep * obj.maxIter, obj.maxIter+1)';
obj.partitioningTimes = obj.times(obj.partitioningFreq:obj.partitioningFreq:size(obj.times, 1));
% Prepare performance data store (at t = 0, all have 0 performance)
obj.fPerf = figure;
obj.perf = [zeros(size(obj.agents, 1) + 1, 1), NaN(size(obj.agents, 1) + 1, size(obj.partitioningTimes, 1) - 1)];
% Prepare h function data store
obj.h = NaN(size(obj.agents, 1) * (size(obj.agents, 1) - 1) / 2 + size(obj.agents, 1) * size(obj.obstacles, 1) + 6, size(obj.times, 1));
% Create initial partitioning
obj.partitioning = obj.agents{1}.partition(obj.agents, obj.domain.objective);
% Initialize variable that will store agent positions for trail plots
obj.posHist = NaN(size(obj.agents, 1), obj.maxIter + 1, 3);
obj.posHist(1:size(obj.agents, 1), 1, 1:3) = reshape(cell2mat(cellfun(@(x) x.pos, obj.agents, 'UniformOutput', false)), size(obj.agents, 1), 1, 3);
obj = obj.partition();
% Set up plots showing initialized state
obj = obj.plot();
% Run validations
obj.validate();
end

View File

@@ -1,76 +0,0 @@
function obj = lesserNeighbor(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
% initialize solution with self-connections only
constraintAdjacencyMatrix = logical(eye(size(obj.agents, 1)));
for ii = 1:size(obj.agents, 1)
% Find lesser neighbors of each agent
% Lesser neighbors of ii are jj < ii in range of ii
lesserNeighbors = [];
for jj = 1:(ii - 1)
if obj.adjacency(ii, jj)
lesserNeighbors = [lesserNeighbors, jj];
end
end
obj.agents{ii}.lesserNeighbors = lesserNeighbors;
% Early exit for isolated agents
if isempty(obj.agents{ii}.lesserNeighbors)
continue
end
% Focus on subgraph defined by lesser neighbors
subgraphAdjacency = obj.adjacency(obj.agents{ii}.lesserNeighbors, obj.agents{ii}.lesserNeighbors);
% Find connected components in each agent's subgraph
% TODO: rewrite this using matlab "conncomp" function?
visited = false(size(subgraphAdjacency, 1), 1);
components = {};
for jj = 1:size(subgraphAdjacency, 1)
if ~visited(jj)
reachable = bfs(subgraphAdjacency, jj);
visited(reachable) = true;
components{end+1} = obj.agents{ii}.lesserNeighbors(reachable);
end
end
% Connect to the greatest index in each connected component in the
% lesser neighborhood of this agent
for jj = 1:size(components, 2)
constraintAdjacencyMatrix(ii, max(components{jj})) = true;
constraintAdjacencyMatrix(max(components{jj}), ii) = true;
end
end
obj.constraintAdjacencyMatrix = constraintAdjacencyMatrix | constraintAdjacencyMatrix';
end
function cComp = bfs(subgraphAdjacency, startIdx)
n = size(subgraphAdjacency, 1);
visited = false(1, n);
queue = startIdx;
cComp = startIdx;
visited(startIdx) = true;
while ~isempty(queue)
current = queue(1);
queue(1) = [];
% Find all neighbors of current node in the subgraph
neighbors = find(subgraphAdjacency(current, :));
for neighbor = neighbors
if ~visited(neighbor)
visited(neighbor) = true;
cComp = [cComp, neighbor];
queue = [queue, neighbor];
end
end
end
cComp = sort(cComp);
end

View File

@@ -4,73 +4,53 @@ classdef miSim
% Simulation parameters
properties (SetAccess = private, GetAccess = public)
timestep = NaN; % delta time interval for simulation iterations
timestepIndex = NaN; % index of the current timestep (useful for time-indexed arrays)
partitioningFreq = NaN; % number of simulation timesteps at which the partitioning routine is re-run
maxIter = NaN; % maximum number of simulation iterations
domain = rectangularPrism;
objective = sensingObjective;
obstacles = cell(0, 1); % geometries that define obstacles within the domain
agents = cell(0, 1); % agents that move within the domain
adjacency = NaN; % Adjacency matrix representing communications network graph
constraintAdjacencyMatrix = NaN; % Adjacency matrix representing desired lesser neighbor connections
sensorPerformanceMinimum = 1e-6; % minimum sensor performance to allow assignment of a point in the domain to a partition
partitioning = NaN;
perf; % sensor performance timeseries array
performance = 0; % simulation performance timeseries vector
barrierGain = 100; % CBF gain parameter
barrierExponent = 3; % CBF exponent parameter
artifactName = "";
fPerf; % performance plot figure
performance = NaN; % current cumulative sensor performance
end
properties (Access = private)
% Sim
t = NaN; % current sim time
perf; % sensor performance timeseries array
times;
partitioningTimes;
% Plot objects
makePlots = true; % enable/disable simulation plotting (performance implications)
makeVideo = true; % enable/disable VideoWriter (performance implications)
f; % main plotting tiled layout figure
f = firstPlotSetup(); % main plotting tiled layout figure
connectionsPlot; % objects for lines connecting agents in spatial plots
graphPlot; % objects for abstract network graph plot
partitionPlot; % objects for partition plot
performancePlot; % objects for sensor performance plot
posHist; % data for trail plot
trailPlot; % objects for agent trail plot
fPerf; % performance plot figure
performancePlot; % objects for sensor performance plot
% Indicies for various plot types in the main tiled layout figure
spatialPlotIndices = [6, 4, 3, 2];
objectivePlotIndices = [6, 4];
networkGraphIndex = 5;
partitionGraphIndex = 1;
% CBF plotting
h; % h function values
hf; % h function plotting figure
caPlot; % objects for collision avoidance h function plot
obsPlot; % objects for obstacle h function plot
domPlot; % objects for domain h function plot
end
methods (Access = public)
[obj] = initialize(obj, domain, agents, barrierGain, barrierExponent, minAlt, timestep, maxIter, obstacles, makePlots, makeVideo);
[obj] = initialize(obj, domain, objective, agents, timestep, partitoningFreq, maxIter, obstacles);
[obj] = run(obj);
[obj] = lesserNeighbor(obj);
[obj] = constrainMotion(obj);
[obj] = partition(obj);
[obj] = updateAdjacency(obj);
[obj] = plot(obj);
[obj] = plotConnections(obj);
[obj] = plotPartitions(obj);
[obj] = plotGraph(obj);
[obj] = plotTrails(obj);
[obj] = plotH(obj);
[obj] = updatePlots(obj);
validate(obj);
teardown(obj);
[obj] = updatePlots(obj, updatePartitions);
end
methods (Access = private)
[v] = setupVideoWriter(obj);
end
end
end

36
@miSim/partition.m Normal file
View File

@@ -0,0 +1,36 @@
function obj = partition(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
% Assess sensing performance of each agent at each sample point
% in the domain
agentPerformances = cellfun(@(x) reshape(x.sensorModel.sensorPerformance(x.pos, x.pan, x.tilt, [obj.objective.X(:), obj.objective.Y(:), zeros(size(obj.objective.X(:)))]), size(obj.objective.X)), obj.agents, 'UniformOutput', false);
agentPerformances{end + 1} = obj.sensorPerformanceMinimum * ones(size(agentPerformances{end})); % add additional layer to represent the threshold that has to be cleared for assignment to any partiton
agentPerformances = cat(3, agentPerformances{:});
% Get highest performance value at each point
[~, idx] = max(agentPerformances, [], 3);
% Collect agent indices in the same way as performance
agentInds = cellfun(@(x) x.index * ones(size(obj.objective.X)), obj.agents, 'UniformOutput', false);
agentInds{end + 1} = zeros(size(agentInds{end})); % index for no assignment
agentInds = cat(3, agentInds{:});
% Get highest performing agent's index
[m,n,~] = size(agentInds);
[jj,kk] = ndgrid(1:m, 1:n);
obj.partitioning = agentInds(sub2ind(size(agentInds), jj, kk, idx));
% Get individual agent sensor performance
nowIdx = [0; obj.partitioningTimes] == obj.t;
for ii = 1:size(obj.agents, 1)
obj.perf(ii, nowIdx) = sum(agentPerformances(sub2ind(size(agentInds), jj, kk, ii)), 'all');
end
% Current total performance
obj.perf(end, nowIdx) = sum(obj.perf(1:(end - 1), nowIdx));
end

View File

@@ -5,11 +5,6 @@ function obj = plot(obj)
arguments (Output)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
% fast exit when plotting is disabled
if ~obj.makePlots
return;
end
% Plot domain
[obj.domain, obj.f] = obj.domain.plotWireframe(obj.spatialPlotIndices);
@@ -22,7 +17,7 @@ function obj = plot(obj)
% Plot objective gradient
obj.f = obj.domain.objective.plot(obj.objectivePlotIndices, obj.f);
% Plot agents and their collision/communications geometries
% Plot agents and their collision geometries
for ii = 1:size(obj.agents, 1)
[obj.agents{ii}, obj.f] = obj.agents{ii}.plot(obj.spatialPlotIndices, obj.f);
end
@@ -36,9 +31,6 @@ function obj = plot(obj)
% Plot domain partitioning
obj = obj.plotPartitions();
% Plot agent trails
obj = obj.plotTrails();
% Enforce plot limits
for ii = 1:size(obj.spatialPlotIndices, 2)
xlim(obj.f.Children(1).Children(obj.spatialPlotIndices(ii)), [obj.domain.minCorner(1), obj.domain.maxCorner(1)]);
@@ -48,10 +40,4 @@ function obj = plot(obj)
% Plot performance
obj = obj.plotPerformance();
% Plot h functions
obj = obj.plotH();
% Switch back to primary figure
figure(obj.f);
end

View File

@@ -9,9 +9,9 @@ function obj = plotConnections(obj)
% Iterate over lower triangle off-diagonal region of the
% adjacency matrix to plot communications links between agents
X = []; Y = []; Z = [];
for ii = 2:size(obj.constraintAdjacencyMatrix, 1)
for ii = 2:size(obj.adjacency, 1)
for jj = 1:(ii - 1)
if obj.constraintAdjacencyMatrix(ii, jj)
if obj.adjacency(ii, jj)
X = [X; obj.agents{ii}.pos(1), obj.agents{jj}.pos(1)];
Y = [Y; obj.agents{ii}.pos(2), obj.agents{jj}.pos(2)];
Z = [Z; obj.agents{ii}.pos(3), obj.agents{jj}.pos(3)];
@@ -32,8 +32,10 @@ function obj = plotConnections(obj)
end
% Copy to other plots
for ii = 2:size(obj.spatialPlotIndices, 2)
o = [o, copyobj(o(:, 1), obj.f.Children(1).Children(obj.spatialPlotIndices(ii)))];
if size(obj.spatialPlotIndices, 2) > 1
for ii = 2:size(obj.spatialPlotIndices, 2)
o = [o, copyobj(o(:, 1), obj.f.Children(1).Children(obj.spatialPlotIndices(ii)))];
end
end
obj.connectionsPlot = o;

View File

@@ -7,7 +7,7 @@ function obj = plotGraph(obj)
end
% Form graph from adjacency matrix
G = graph(obj.constraintAdjacencyMatrix, 'omitselfloops');
G = graph(obj.adjacency, 'omitselfloops');
% Plot graph object
if isnan(obj.networkGraphIndex)

View File

@@ -1,61 +0,0 @@
function obj = plotH(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
obj.hf = figure;
tiledlayout(obj.hf, 4, 1, "TileSpacing", "tight", "Padding", "compact");
nexttile(obj.hf.Children(1));
axes(obj.hf.Children(1).Children(1));
grid(obj.hf.Children(1).Children(1), "on");
xlabel(obj.hf.Children(1).Children(1), "Time (s)"); % ylabel(obj.hf.Children(1).Children(1), "");
title(obj.hf.Children(1).Children(1), "Collision Avoidance");
hold(obj.hf.Children(1).Children(1), "on");
obj.caPlot = plot(obj.h(1:(size(obj.agents, 1) * (size(obj.agents, 1) - 1) / 2), :)');
legendStrings = [];
for ii = 2:size(obj.agents, 1)
for jj = 1:(ii - 1)
legendStrings = [legendStrings; sprintf("A%d A%d", jj, ii)];
end
end
legend(obj.hf.Children(1).Children(1), legendStrings, 'Location', 'bestoutside');
hold(obj.hf.Children(1).Children(2), "off");
nexttile(obj.hf.Children(1));
axes(obj.hf.Children(1).Children(1));
grid(obj.hf.Children(1).Children(1), "on");
xlabel(obj.hf.Children(1).Children(1), "Time (s)"); % ylabel(obj.hf.Children(1).Children(2), "");
title(obj.hf.Children(1).Children(1), "Obstacles");
hold(obj.hf.Children(1).Children(1), "on");
obj.obsPlot = plot(obj.h((1 + (size(obj.agents, 1) * (size(obj.agents, 1) - 1) / 2)):(((size(obj.agents, 1) * (size(obj.agents, 1) - 1) / 2)) + size(obj.agents, 1) * size(obj.obstacles, 1)), :)');
legendStrings = [];
for ii = 1:size(obj.obstacles, 1)
for jj = 1:size(obj.agents, 1)
legendStrings = [legendStrings; sprintf("A%d O%d", jj, ii)];
end
end
legend(obj.hf.Children(1).Children(1), legendStrings, 'Location', 'bestoutside');
hold(obj.hf.Children(1).Children(2), "off");
nexttile(obj.hf.Children(1));
axes(obj.hf.Children(1).Children(1));
grid(obj.hf.Children(1).Children(1), "on");
xlabel(obj.hf.Children(1).Children(1), "Time (s)"); % ylabel(obj.hf.Children(1).Children(1), "");
title(obj.hf.Children(1).Children(1), "Domain");
hold(obj.hf.Children(1).Children(1), "on");
obj.domPlot = plot(obj.h((1 + (((size(obj.agents, 1) * (size(obj.agents, 1) - 1) / 2)) + size(obj.agents, 1) * size(obj.obstacles, 1))):size(obj.h, 1), 1:end)');
legend(obj.hf.Children(1).Children(1), ["X Min"; "X Max"; "Y Min"; "Y Max"; "Z Min"; "Z Max";], 'Location', 'bestoutside');
hold(obj.hf.Children(1).Children(2), "off");
nexttile(obj.hf.Children(1));
axes(obj.hf.Children(1).Children(1));
grid(obj.hf.Children(1).Children(1), "on");
xlabel(obj.hf.Children(1).Children(1), "Time (s)"); % ylabel(obj.hf.Children(1).Children(1), "");
title(obj.hf.Children(1).Children(1), "Communications");
% skipped this for now because it is very complicated
end

View File

@@ -6,13 +6,6 @@ function obj = plotPerformance(obj)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
% fast exit when plotting is disabled
if ~obj.makePlots
return;
end
obj.fPerf = figure;
axes(obj.fPerf);
title(obj.fPerf.Children(1), "Sensor Performance");
xlabel(obj.fPerf.Children(1), 'Time (s)');
@@ -22,24 +15,14 @@ function obj = plotPerformance(obj)
% Plot current cumulative performance
hold(obj.fPerf.Children(1), 'on');
o = plot(obj.fPerf.Children(1), obj.perf(end, :));
warning('off', 'MATLAB:gui:array:InvalidArrayShape'); % suppress this warning to avoid polluting output
o.XData = NaN(1, obj.maxIter); % correct time will be set at runtime
o.YData = [0, NaN(1, obj.maxIter - 1)];
hold(obj.fPerf.Children(1), 'off');
% Plot current agent performance
for ii = 1:(size(obj.perf, 1) - 1)
hold(obj.fPerf.Children(1), 'on');
o = [o; plot(obj.fPerf.Children(1), obj.perf(ii, :))];
o(end).XData = NaN(1, obj.maxIter); % correct time will be set at runtime
o(end).YData = [0, NaN(1, obj.maxIter - 1)];
hold(obj.fPerf.Children(1), 'off');
end
% Add legend
agentStrings = string(cellfun(@(x) x.label, obj.agents, 'UniformOutput', false));
agentStrings = ["Total"; agentStrings];
legend(obj.fPerf.Children(1), agentStrings, 'Location', 'northwest');
obj.performancePlot = o;
end

View File

@@ -1,30 +0,0 @@
function obj = plotTrails(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')}
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'miSim')}
end
% fast exit when plotting is disabled
if ~obj.makePlots
return;
end
% Plot full range of position history on each spatial plot axes
o = [];
for ii = 1:(size(obj.posHist, 1))
hold(obj.f.Children(1).Children(obj.spatialPlotIndices(1)), 'on');
o = [o; plot3(obj.f.Children(1).Children(obj.spatialPlotIndices(1)), obj.posHist(ii, 1:obj.maxIter, 1), obj.posHist(ii, 1:obj.maxIter, 2), obj.posHist(ii, 1:obj.maxIter, 3), 'Color', 'k', 'LineWidth', 1)];
hold(obj.f.Children(1).Children(obj.spatialPlotIndices(1)), 'off');
end
% Copy to other plots
for ii = 2:size(obj.spatialPlotIndices, 2)
o = [o, copyobj(o(:, 1), obj.f.Children(1).Children(obj.spatialPlotIndices(ii)))];
end
% Add legend?
obj.trailPlot = o;
end

View File

@@ -7,60 +7,37 @@ function [obj] = run(obj)
end
% Start video writer
if obj.makeVideo
v = obj.setupVideoWriter();
v.open();
end
v = obj.setupVideoWriter();
v.open();
for ii = 1:size(obj.times, 1)
% Display current sim time
obj.t = obj.times(ii);
obj.timestepIndex = ii;
fprintf("Sim Time: %4.2f (%d/%d)\n", obj.t, ii, obj.maxIter + 1);
% Before moving
% Validate current simulation configuration
obj.validate();
% Update partitioning before moving (this one is strictly for
% plotting purposes, the real partitioning is done by the agents)
obj.partitioning = obj.agents{1}.partition(obj.agents, obj.domain.objective);
% Determine desired communications links
obj = obj.lesserNeighbor();
% Moving
% Iterate over agents to simulate their unconstrained motion
for jj = 1:size(obj.agents, 1)
obj.agents{jj} = obj.agents{jj}.run(obj.domain, obj.partitioning, obj.timestepIndex, jj, obj.agents);
% Check if it's time for new partitions
updatePartitions = false;
if ismember(obj.t, obj.partitioningTimes)
updatePartitions = true;
obj = obj.partition();
end
% Adjust motion determined by unconstrained gradient ascent using
% CBF constraints solved by QP
obj = constrainMotion(obj);
% After moving
% Update agent position history array
obj.posHist(1:size(obj.agents, 1), obj.timestepIndex + 1, 1:3) = reshape(cell2mat(cellfun(@(x) x.pos, obj.agents, 'UniformOutput', false)), size(obj.agents, 1), 1, 3);
% Update total performance
obj.performance = [obj.performance, sum(cellfun(@(x) x.performance(obj.timestepIndex+1), obj.agents))];
% Iterate over agents to simulate their motion
for jj = 1:size(obj.agents, 1)
obj.agents{jj} = obj.agents{jj}.run(obj.objective, obj.domain, obj.partitioning);
end
% Update adjacency matrix
obj = obj.updateAdjacency();
% Update plots
obj = obj.updatePlots();
obj = obj.updatePlots(updatePartitions);
% Write frame in to video
if obj.makeVideo
I = getframe(obj.f);
v.writeVideo(I);
end
I = getframe(obj.f);
v.writeVideo(I);
end
if obj.makeVideo
% Close video file
v.close();
end
end
% Close video file
v.close();
end

View File

@@ -7,9 +7,9 @@ function v = setupVideoWriter(obj)
end
if ispc || ismac
v = VideoWriter(fullfile(matlab.project.rootProject().RootFolder, 'sandbox', strcat(obj.artifactName, "_miSimHist")), 'MPEG-4');
v = VideoWriter(fullfile('sandbox', strcat(string(datetime('now'), 'yyyy_MM_dd_HH_mm_ss'), '_miSimHist')), 'MPEG-4');
elseif isunix
v = VideoWriter(fullfile(matlab.project.rootProject().RootFolder, 'sandbox', strcat(obj.artifactName, "_miSimHist")), 'Motion JPEG AVI');
v = VideoWriter(fullfile('.', strcat(string(datetime('now'), 'yyyy_MM_dd_HH_mm_ss'), '_miSimHist')), 'Motion JPEG AVI');
end
v.FrameRate = 1 / obj.timestep;

View File

@@ -1,13 +0,0 @@
function teardown(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
arguments (Output)
end
% Close plots
close(obj.hf);
close(obj.fPerf);
close(obj.f);
end

View File

@@ -7,18 +7,26 @@ function obj = updateAdjacency(obj)
end
% Initialize assuming only self-connections
A = true(size(obj.agents, 1));
A = logical(eye(size(obj.agents, 1)));
% Check lower triangle off-diagonal connections
for ii = 2:size(A, 1)
for jj = 1:(ii - 1)
% Check that agents are not out of range
if norm(obj.agents{ii}.pos - obj.agents{jj}.pos) > min([obj.agents{ii}.commsGeometry.radius, obj.agents{jj}.commsGeometry.radius])
A(ii, jj) = false; % comm range violation
continue;
if norm(obj.agents{ii}.pos - obj.agents{jj}.pos) <= min([obj.agents{ii}.comRange, obj.agents{jj}.comRange])
% Make sure that obstacles don't obstruct the line
% of sight, breaking the connection
for kk = 1:size(obj.obstacles, 1)
if ~obj.obstacles{kk}.containsLine(obj.agents{ii}.pos, obj.agents{jj}.pos)
A(ii, jj) = true;
end
end
% need extra handling for cases with no obstacles
if isempty(obj.obstacles)
A(ii, jj) = true;
end
end
end
end
obj.adjacency = A & A';
obj.adjacency = A | A';
end

View File

@@ -1,23 +1,19 @@
function [obj] = updatePlots(obj)
function [obj] = updatePlots(obj, updatePartitions)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
updatePartitions (1, 1) logical = false;
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
% Fast exit when plotting is disabled
if ~obj.makePlots
return;
end
% Update agent positions, collision/communication geometries
% Update agent positions, collision geometries
for ii = 1:size(obj.agents, 1)
obj.agents{ii}.updatePlots();
end
% The remaining updates might should all be possible to do in a clever
% way that moves existing lines instead of clearing and
% The remaining updates might be possible to do in a clever way
% that moves existing lines instead of clearing and
% re-plotting, which is much better for performance boost
% Update agent connections plot
@@ -29,8 +25,10 @@ function [obj] = updatePlots(obj)
obj = obj.plotGraph();
% Update partitioning plot
delete(obj.partitionPlot);
obj = obj.plotPartitions();
if updatePartitions
delete(obj.partitionPlot);
obj = obj.plotPartitions();
end
% reset plot limits to fit domain
for ii = 1:size(obj.spatialPlotIndices, 2)
@@ -38,36 +36,17 @@ function [obj] = updatePlots(obj)
ylim(obj.f.Children(1).Children(obj.spatialPlotIndices(ii)), [obj.domain.minCorner(2), obj.domain.maxCorner(2)]);
zlim(obj.f.Children(1).Children(obj.spatialPlotIndices(ii)), [obj.domain.minCorner(3), obj.domain.maxCorner(3)]);
end
% Update agent trails
for jj = 1:size(obj.spatialPlotIndices, 2)
for ii = 1:size(obj.agents, 1)
obj.trailPlot((jj - 1) * size(obj.agents, 1) + ii).XData(obj.timestepIndex) = obj.posHist(ii, obj.timestepIndex, 1);
obj.trailPlot((jj - 1) * size(obj.agents, 1) + ii).YData(obj.timestepIndex) = obj.posHist(ii, obj.timestepIndex, 2);
obj.trailPlot((jj - 1) * size(obj.agents, 1) + ii).ZData(obj.timestepIndex) = obj.posHist(ii, obj.timestepIndex, 3);
end
end
drawnow;
% Update performance plot
% Re-normalize performance plot
normalizingFactor = 1/max(obj.performance);
obj.performancePlot(1).YData(1:(length(obj.performance) + 1)) = [obj.performance, 0] * normalizingFactor;
obj.performancePlot(1).XData([obj.timestepIndex, obj.timestepIndex + 1]) = [obj.t, obj.t + obj.timestep];
for ii = 1:(size(obj.agents, 1))
obj.performancePlot(ii + 1).YData(1:(length(obj.performance) + 1)) = [obj.agents{ii}.performance(1:length(obj.performance)), 0] * normalizingFactor;
obj.performancePlot(ii + 1).XData([obj.timestepIndex, obj.timestepIndex + 1]) = [obj.t, obj.t + obj.timestep];
if updatePartitions
nowIdx = [0; obj.partitioningTimes] == obj.t;
% set(obj.performancePlot(1), 'YData', obj.perf(end, 1:find(nowIdx)));
obj.performancePlot(1).YData(nowIdx) = obj.perf(end, nowIdx);
for ii = 2:size(obj.performancePlot, 1)
obj.performancePlot(ii).YData(nowIdx) = obj.perf(ii, nowIdx);
end
drawnow;
end
% Update h function plots
for ii = 1:size(obj.caPlot, 1)
obj.caPlot(ii).YData(obj.timestepIndex) = obj.h(ii, obj.timestepIndex);
end
for ii = 1:size(obj.obsPlot, 1)
obj.obsPlot(ii).YData(obj.timestepIndex) = obj.h(ii + size(obj.caPlot, 1), obj.timestepIndex);
end
for ii = 1:size(obj.domPlot, 1)
obj.domPlot(ii).YData(obj.timestepIndex) = obj.h(ii + size(obj.caPlot, 1) + size(obj.obsPlot, 1), obj.timestepIndex);
end
end

View File

@@ -1,27 +0,0 @@
function validate(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
arguments (Output)
end
%% Communications Network Validators
if max(conncomp(graph(obj.adjacency))) ~= 1
warning("Network is not connected");
end
if any(obj.adjacency - obj.constraintAdjacencyMatrix < 0, 'all')
warning("Eliminated network connections that were necessary");
end
%% Obstacle Validators
AO_collisions = cellfun(@(a) cellfun(@(o) o.contains(a.pos), obj.obstacles), obj.agents, 'UniformOutput', false);
AO_collisions = vertcat(AO_collisions{:});
if any(AO_collisions)
[idx, idy] = find(AO_collisions);
for ii = 1:size(idx, 1)
error("Agent(s) %d colliding with obstacle(s) %d", idx(ii), idy(ii));
end
end
end

View File

@@ -1,25 +0,0 @@
function writeParams(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'miSim')};
end
arguments (Output)
end
% Collect agent parameters
collisionRadii = cellfun(@(x) x.collisionGeometry.radius, obj.agents);
alphaDist = cellfun(@(x) x.sensorModel.alphaDist, obj.agents);
betaDist = cellfun(@(x) x.sensorModel.betaDist, obj.agents);
alphaTilt = cellfun(@(x) x.sensorModel.alphaTilt, obj.agents);
betaTilt = cellfun(@(x) x.sensorModel.alphaDist, obj.agents);
comRange = cellfun(@(x) x.commsGeometry.radius, obj.agents);
% Combine with simulation parameters
params = struct('timestep', obj.timestep, 'maxIter', obj.maxIter, 'minAlt', obj.obstacles{end}.maxCorner(3), 'discretizationStep', obj.domain.objective.discretizationStep, ...
'collisionRadius', collisionRadii, 'alphaDist', alphaDist, 'betaDist', betaDist, ...
'alphaTilt', alphaTilt, 'betaTilt', betaTilt, 'comRange', comRange);
% Save all parameters to output file
paramsFile = strcat(obj.artifactName, "_miSimParams");
paramsFile = fullfile(matlab.project.rootProject().RootFolder, 'sandbox', paramsFile);
save(paramsFile, "-struct", "params");
end

View File

@@ -1,20 +1,15 @@
function obj = initialize(obj, objectiveFunction, domain, discretizationStep, protectedRange, sensorPerformanceMinimum)
function obj = initialize(obj, objectiveFunction, domain, discretizationStep, protectedRange)
arguments (Input)
obj (1,1) {mustBeA(obj, 'sensingObjective')};
objectiveFunction (1, 1) {mustBeA(objectiveFunction, 'function_handle')};
domain (1, 1) {mustBeGeometry};
discretizationStep (1, 1) double = 1;
protectedRange (1, 1) double = 1;
sensorPerformanceMinimum (1, 1) double = 1e-6;
end
arguments (Output)
obj (1,1) {mustBeA(obj, 'sensingObjective')};
end
obj.discretizationStep = discretizationStep;
obj.sensorPerformanceMinimum = sensorPerformanceMinimum;
obj.groundAlt = domain.minCorner(3);
obj.protectedRange = protectedRange;
@@ -24,8 +19,8 @@ function obj = initialize(obj, objectiveFunction, domain, discretizationStep, pr
yMin = min(domain.footprint(:, 2));
yMax = max(domain.footprint(:, 2));
xGrid = unique([xMin:obj.discretizationStep:xMax, xMax]);
yGrid = unique([yMin:obj.discretizationStep:yMax, yMax]);
xGrid = unique([xMin:discretizationStep:xMax, xMax]);
yGrid = unique([yMin:discretizationStep:yMax, yMax]);
% Store grid points for plotting later
[obj.X, obj.Y] = meshgrid(xGrid, yGrid);
@@ -33,14 +28,10 @@ function obj = initialize(obj, objectiveFunction, domain, discretizationStep, pr
% Evaluate function over grid points
obj.objectiveFunction = objectiveFunction;
obj.values = reshape(obj.objectiveFunction(obj.X, obj.Y), size(obj.X));
% Normalize
obj.values = obj.values ./ max(obj.values, [], "all");
% store ground position
idx = obj.values == 1;
idx = obj.values == max(obj.values, [], "all");
obj.groundPos = [obj.X(idx), obj.Y(idx)];
obj.groundPos = obj.groundPos(1, 1:2); % for safety, in case 2 points are maximal (somehow)
assert(domain.distance([obj.groundPos, domain.center(3)]) > protectedRange, "Domain is crowding the sensing objective")
end

View File

@@ -14,12 +14,13 @@ function obj = initializeRandomMvnpdf(obj, domain, discretizationStep, protected
while domain.distance(mu) < protectedRange
mu = domain.random();
end
mu = mu(1:2);
% Set random distribution parameters
sig = [2 + rand * 2, 1; 1, 2 + rand * 2];
% Set up random bivariate normal distribution function
objectiveFunction = objectiveFunctionWrapper(mu(1:2), sig);
objectiveFunction = @(x, y) mvnpdf([x(:), y(:)], mu, sig);
% Regular initialization
obj = obj.initialize(objectiveFunction, domain, discretizationStep, protectedRange);

View File

@@ -2,19 +2,18 @@ classdef sensingObjective
% Sensing objective definition parent class
properties (SetAccess = private, GetAccess = public)
label = "";
groundAlt = NaN;
groundPos = [NaN, NaN];
discretizationStep = NaN;
objectiveFunction = @(x, y) NaN; % define objective functions over a grid in this manner
groundAlt = 0;
groundPos = [0, 0];
discretizationStep = 1;
objectiveFunction = @(x, y) 0; % define objective functions over a grid in this manner
X = [];
Y = [];
values = [];
protectedRange = NaN; % keep obstacles from crowding objective
sensorPerformanceMinimum = NaN; % minimum sensor performance to allow assignment of a point in the domain to a partition
protectedRange = 1; % keep obstacles from crowding objective
end
methods (Access = public)
[obj] = initialize(obj, objectiveFunction, domain, discretizationStep, protectedRange, sensorPerformanceMinimum);
[obj] = initialize(obj, objectiveFunction, domain, discretizationStep, protectedRange);
[obj] = initializeRandomMvnpdf(obj, domain, protectedRange, discretizationStep, protectedRange);
[f ] = plot(obj, ind, f);
end

View File

@@ -17,6 +17,6 @@ classdef cone
methods (Access = public)
[obj ] = initialize(obj, center, radius, height, tag, label);
[obj, f] = plot(obj, ind, f, maxAlt);
[obj, f] = plot(obj, ind, f);
end
end

View File

@@ -1,9 +1,8 @@
function [obj, f] = plot(obj, ind, f, maxAlt)
function [obj, f] = plot(obj, ind, f)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'cone')};
ind (1, :) double = NaN;
f (1, 1) {mustBeA(f, 'matlab.ui.Figure')} = figure;
maxAlt (1, 1) = 10;
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'cone')};
@@ -13,18 +12,16 @@ function [obj, f] = plot(obj, ind, f, maxAlt)
% Create axes if they don't already exist
f = firstPlotSetup(f);
scalingFactor = (maxAlt / obj.height);
% Plot cone
[X, Y, Z] = cylinder([scalingFactor * obj.radius, 0], obj.n);
[X, Y, Z] = cylinder([obj.radius, 0], obj.n);
% Scale to match height
Z = Z * maxAlt;
Z = Z * obj.height;
% Move to center location
X = X + obj.center(1);
Y = Y + obj.center(2);
Z = Z + obj.center(3) - maxAlt;
Z = Z + obj.center(3);
% Plot
if isnan(ind)

View File

@@ -1,19 +0,0 @@
function cPos = closestToPoint(obj, pos)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'rectangularPrism')};
pos (:, 3) double;
end
arguments (Output)
cPos (:, 3) double;
end
cPos = NaN(1, 3);
for ii = 1:3
if pos(ii) < obj.minCorner(ii)
cPos(ii) = obj.minCorner(ii);
elseif pos(ii) > obj.maxCorner(ii)
cPos(ii) = obj.maxCorner(ii);
else
cPos(ii) = pos(ii);
end
end
end

View File

@@ -9,38 +9,33 @@ function c = containsLine(obj, pos1, pos2)
end
d = pos2 - pos1;
% endpoint contained (trivial case)
if obj.contains(pos1) || obj.contains(pos2)
c = true;
% edge case where the line is parallel to the geometry
if abs(d) < 1e-12
% check if it happens to start or end inside or outside of
% the geometry
if obj.contains(pos1) || obj.contains(pos2)
c = true;
else
c = false;
end
return;
end
% parameterize the line segment to check for an intersection
tMin = 0;
tMax = 1;
tmin = -inf;
tmax = inf;
% Standard case
for ii = 1:3
% line is parallel to geometry
if abs(d(ii)) < 1e-12
if pos1(ii) < obj.minCorner(ii) || pos1(ii) > obj.maxCorner(ii)
c = false;
return;
end
else
t1 = (obj.minCorner(ii) - pos1(ii)) / d(ii);
t2 = (obj.maxCorner(ii) - pos1(ii)) / d(ii);
tLow = min(t1, t2);
tHigh = max(t1, t2);
tMin = max(tMin, tLow);
tMax = min(tMax, tHigh);
if tMin > tMax
c = false;
return;
end
t1 = (obj.minCorner(ii) - pos1(ii)) / d(ii);
t2 = (obj.maxCorner(ii) - pos2(ii)) / d(ii);
tmin = max(tmin, min(t1, t2));
tmax = min(tmax, max(t1, t2));
if tmin > tmax
c = false;
return;
end
end
c = true;
end
c = (tmax >= 0) && (tmin <= 1);
end

View File

@@ -4,7 +4,7 @@ function d = distance(obj, pos)
pos (:, 3) double;
end
arguments (Output)
d (:, 1) double;
d (:, 1) double
end
if obj.contains(pos)
% Queried point is inside geometry

View File

@@ -1,42 +0,0 @@
function g = distanceGradient(obj, pos)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'rectangularPrism')};
pos (:, 3) double;
end
arguments (Output)
g (:, 3) double
end
% find nearest point on surface to query position
q = min(max(pos, obj.minCorner), obj.maxCorner);
% Find distance and direction between pos and q
v = pos - q;
vNorm = norm(v);
% position is outside geometry
if vNorm > 0
% gradient is normalized vector from q to p
g = v / vNorm;
return;
end
% position is on or in geometry
% find distances to each face in each dimension
distances = [pos(1) - obj.minCorner(1), obj.maxCorner(1) - pos(1), pos(2) - obj.minCorner(2), obj.maxCorner(2) - pos(2), pos(3) - obj.minCorner(3), obj.maxCorner(3) - pos(3)];
[~, idx] = min(distances);
% I think there needs to be additional handling here for the
% edge/corner cases, where there are ways to balance or resolve ties
% when two faces are equidistant to the query position
assert(sum(idx) == idx, "Implement edge case handling");
% select gradient that brings us quickest to the nearest face
g = [ 1, 0, 0; ...
-1, 0, 0; ...
0, 1, 0; ...
0, -1, 0; ...
0, 0, 1; ...
0, 0, -1;];
g = g(idx, :);
end

View File

@@ -24,10 +24,6 @@ function obj = initialize(obj, bounds, tag, label, objectiveFunction, discretiza
% Compute center
obj.center = obj.minCorner + obj.dimensions ./ 2;
% Compute a (fake) radius
% fully contains the rectangular prism from the center
obj.radius = (1/2) * sqrt(sum(obj.dimensions.^2));
% Compute vertices
obj.vertices = [obj.minCorner;
obj.maxCorner(1), obj.minCorner(2:3);

View File

@@ -1,12 +1,11 @@
function [obj] = initializeRandom(obj, tag, label, minDimension, maxDimension, domain, minAlt)
function [obj] = initializeRandom(obj, tag, label, minDimension, maxDimension, domain)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'rectangularPrism')};
tag (1, 1) REGION_TYPE = REGION_TYPE.INVALID;
label (1, 1) string = "";
minDimension (1, 1) double = 10;
maxDimension (1, 1) double = 20;
maxDimension (1, 1) double= 20;
domain (1, 1) {mustBeGeometry} = rectangularPrism;
minAlt (1, 1) double = 1;
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'rectangularPrism')};
@@ -28,7 +27,7 @@ function [obj] = initializeRandom(obj, tag, label, minDimension, maxDimension, d
while ~domain.contains(candidateMaxCorner) || all(domain.objective.groundPos + domain.objective.protectedRange >= candidateMinCorner(1:2), 2) && all(domain.objective.groundPos - domain.objective.protectedRange <= candidateMaxCorner(1:2), 2)
if ii == 0 || ii > 10
candidateMinCorner = domain.random();
candidateMinCorner(3) = minAlt; % bind to floor (plus minimum altitude constraint)
candidateMinCorner(3) = 0; % bind to floor
ii = 1;
end

View File

@@ -3,6 +3,7 @@ classdef rectangularPrism
properties (SetAccess = private, GetAccess = public)
% Meta
tag = REGION_TYPE.INVALID;
label = "";
% Spatial
minCorner = NaN(1, 3);
@@ -10,7 +11,6 @@ classdef rectangularPrism
dimensions = NaN(1, 3);
center = NaN;
footprint = NaN(4, 2);
radius = NaN; % fake radius
% Graph
vertices = NaN(8, 3);
@@ -22,7 +22,6 @@ classdef rectangularPrism
lines;
end
properties (SetAccess = public, GetAccess = public)
label = "";
% Sensing objective (for DOMAIN region type only)
objective;
end
@@ -32,9 +31,7 @@ classdef rectangularPrism
[obj ] = initializeRandom(obj, tag, label, minDimension, maxDimension, domain);
[r ] = random(obj);
[c ] = contains(obj, pos);
[cPos ] = closestToPoint(obj, pos);
[d ] = distance(obj, pos);
[g ] = distanceGradient(obj, pos);
[c ] = containsLine(obj, pos1, pos2);
[obj, f] = plotWireframe(obj, ind, f);
end

View File

@@ -1,10 +0,0 @@
function c = contains(obj, pos)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'spherical')};
pos (:, 3) double;
end
arguments (Output)
c (:, 1) logical
end
c = norm(obj.center - pos) <= obj.radius;
end

View File

@@ -1,28 +0,0 @@
function c = containsLine(obj, pos1, pos2)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'spherical')};
pos1 (1, 3) double;
pos2 (1, 3) double;
end
arguments (Output)
c (1, 1) logical
end
d = pos2 - pos1;
f = pos1 - obj.center;
a = dot(d, d);
b = 2 * dot(f, d);
c = dot(f, f) - obj.radius^2;
disc = b^2 - 4*a*c;
if disc < 0
c = false;
return;
end
t = [(-b - sqrt(disc)) / (2 * a), (-b + sqrt(disc)) / (2 * a)];
c = (t(1) >= 0 && t(1) <= 1) || (t(2) >= 0 && t(2) <= 1);
end

View File

@@ -1,37 +0,0 @@
function obj = initialize(obj, center, radius, tag, label)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'spherical')};
center (1, 3) double;
radius (1, 1) double;
tag (1, 1) REGION_TYPE = REGION_TYPE.INVALID;
label (1, 1) string = "";
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'spherical')};
end
obj.tag = tag;
obj.label = label;
% Define geometry
obj.center = center;
obj.radius = radius;
obj.diameter = 2 * obj.radius;
% fake vertices in a cross pattern
obj.vertices = [obj.center + [obj.radius, 0, 0]; ...
obj.center - [obj.radius, 0, 0]; ...
obj.center + [0, obj.radius, 0]; ...
obj.center - [0, obj.radius, 0]; ...
obj.center + [0, 0, obj.radius]; ...
obj.center - [0, 0, obj.radius]];
% fake edges in two perpendicular rings
obj.edges = [1, 3; ...
3, 2; ...
2, 4; ...
4, 1; ...
1, 5; ...
5, 2; ...
2, 6; ...
6, 1];
end

View File

@@ -1,43 +0,0 @@
function [obj, f] = plotWireframe(obj, ind, f)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'spherical')};
ind (1, :) double = NaN;
f (1, 1) {mustBeA(f, 'matlab.ui.Figure')} = figure;
end
arguments (Output)
obj (1, 1) {mustBeA(obj, 'spherical')};
f (1, 1) {mustBeA(f, 'matlab.ui.Figure')};
end
% Create axes if they don't already exist
f = firstPlotSetup(f);
% Create plotting inputs
[X, Y, Z] = sphere(8);
% Scale
X = X * obj.radius;
Y = Y * obj.radius;
Z = Z * obj.radius;
% Shift
X = X + obj.center(1);
Y = Y + obj.center(2);
Z = Z + obj.center(3);
% Plot the boundaries of the geometry into 3D view
if isnan(ind)
o = plot3(f.CurrentAxes, X, Y, Z, '-', 'Color', obj.tag.color, 'LineWidth', 2);
else
hold(f.Children(1).Children(ind(1)), "on");
o = plot3(f.Children(1).Children(ind(1)), X, Y, Z, '-', 'Color', obj.tag.color, 'LineWidth', 1);
hold(f.Children(1).Children(ind(1)), "off");
end
% Copy to other requested tiles
if numel(ind) > 1
for ii = 2:size(ind, 2)
o = [o, copyobj(o(:, 1), f.Children(1).Children(ind(ii)))];
end
end
obj.lines = o;
end

View File

@@ -1,15 +0,0 @@
function r = random(obj)
arguments (Input)
obj (1, 1) {mustBeA(obj, 'spherical')};
end
arguments (Output)
r (1, 3) double
end
y = (rand - 0.5) * 2; % uniform draw on [-1, 1]
R = sqrt(1 - y^2);
lon = (rand - 0.5) * 2 * pi; % uniform draw on [-pi, pi]
s = [R * sin(lon), y, R * cos(lon)]; % random point on surface
r = s * rand^(1/3); % scaled to random normalized radius [0, 1]
r = obj.center + obj.radius * r;
end

View File

@@ -1,33 +0,0 @@
classdef spherical
% Rectangular prism geometry
properties (SetAccess = private, GetAccess = public)
% Spatial
center = NaN;
radius = NaN;
diameter = NaN;
vertices; % fake vertices
edges; % fake edges
% Plotting
lines;
end
properties (SetAccess = public, GetAccess = public)
% Meta
tag = REGION_TYPE.INVALID;
label = "";
% Sensing objective (for DOMAIN region type only)
objective;
end
methods (Access = public)
[obj ] = initialize(obj, center, radius, tag, label);
[r ] = random(obj);
[c ] = contains(obj, pos);
[d ] = distance(obj, pos);
[g ] = distanceGradient(obj, pos);
[c ] = containsLine(obj, pos1, pos2);
[obj, f] = plotWireframe(obj, ind, f);
end
end

View File

@@ -9,7 +9,6 @@ classdef REGION_TYPE
OBSTACLE (2, [255, 127, 127]); % obstacle region
COLLISION (3, [255, 255, 128]); % collision avoidance region
FOV (4, [255, 165, 0]); % field of view region
COMMS (5, [0, 255, 0]); % comunications region
end
methods
function obj = REGION_TYPE(id, color)

View File

@@ -0,0 +1,26 @@
function nextPos = gradientAscent(sensedValues, sensedPositions, pos, rate)
arguments (Input)
sensedValues (:, 1) double;
sensedPositions (:, 3) double;
pos (1, 3) double;
rate (1, 1) double = 0.1;
end
arguments (Output)
nextPos(1, 3) double;
end
% As a default, maintain current position
if size(sensedValues, 1) == 0 && size(sensedPositions, 1) == 0
nextPos = pos;
return;
end
% Select next position by maximum sensed value
nextPos = sensedPositions(sensedValues == max(sensedValues), :);
nextPos = [nextPos(1, 1:2), pos(3)]; % just in case two get selected, simply pick one
% rate-limit motion
v = nextPos - pos;
nextPos = pos + (v / norm(v, 2)) * rate;
end

View File

@@ -0,0 +1,2 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="fixedCardinalSensor.m" type="File"/>

View File

@@ -1,2 +1,2 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="plotH.m" type="File"/>
<Info location="sense.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="objectiveFunctionWrapper.m" type="File"/>

View File

@@ -0,0 +1,2 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info Ref="sensingModels" Type="Relative"/>

View File

@@ -0,0 +1,2 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="420d04e4-3880-4a45-8609-11cb30d87302" type="Reference"/>

View File

@@ -0,0 +1,2 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info Ref="guidanceModels" Type="Relative"/>

View File

@@ -0,0 +1,2 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="1d8d2b42-2863-4985-9cf2-980917971eba" type="Reference"/>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="test"/>
</Category>
</Info>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="parametricTestSuite.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="containsLine.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="spherical.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="contains.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="plotWireframe.m" type="File"/>

View File

@@ -1,2 +1,2 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="random.m" type="File"/>
<Info location="sense.m" type="File"/>

View File

@@ -0,0 +1,2 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="sensorPerformance.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="@spherical" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="writeParams.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="plotTrails.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="lesserNeighbor.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="teardown.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="constrainMotion.m" type="File"/>

View File

@@ -1,2 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info location="validate.m" type="File"/>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

View File

@@ -1,6 +0,0 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

View File

@@ -1,2 +1,6 @@
<?xml version="1.0" encoding="UTF-8"?>
<Info/>
<Info>
<Category UUID="FileClassCategory">
<Label UUID="design"/>
</Category>
</Info>

Some files were not shown because too many files have changed in this diff Show More