-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathHopper.m
609 lines (564 loc) · 25.7 KB
/
Hopper.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
classdef Hopper < handle
properties
rbm
rbm_vis
littleDog
hopper
platforms = struct('start', {}, 'end', {}, 'height', {}, 'geom', {})
regions
hip_in_body = struct('front', struct('x', 0.5, 'z', -0.25), ...
'hind', struct('x', -0.5, 'z', -0.25))
leg_length
v
t_data
r_data
v_data
F_data
r_hip_data
p_data
f_data
f_data_vis
th_data
T_data
k_data
q_data
T_actual
region_indicators
body_region_indicators
qtraj
Ftraj
end
methods
function obj = Hopper(leg_length, hip_in_body)
urdf = fullfile(getDrakePath(), 'solvers', 'test', 'littleBrick.urdf');
particle_urdf = fullfile(getDrakePath(), 'systems', 'plants', 'test', 'PointMass.urdf');
options = struct();
options.floating = true;
options.use_new_kinsol = true;
obj.littleDog = LittleDog();
obj.rbm = RigidBodyManipulator(urdf, options);
obj.rbm = obj.rbm.setBody(2, obj.littleDog.getBody(2));
obj.rbm_vis = obj.rbm;
obj.leg_length = leg_length;
obj.hip_in_body = obj.constructHipInBody();
colormap('lines')
colors = colormap';
colormap('default')
options.collision = false;
leg = RigidBodyCapsule(0.01, leg_length, [0; 0; leg_length/2], [0; 0; 0]);
for j = 1:2
% Add feet
obj.rbm_vis = obj.rbm_vis.addRobotFromURDF(particle_urdf, [], [], options);
body = obj.rbm_vis.body(end);
body.visual_geometry{1} = body.visual_geometry{1}.setColor(colors(:,j));
body.visual_geometry{1}.radius = 0.02;
obj.rbm_vis = obj.rbm_vis.setBody(obj.rbm_vis.getNumBodies(), body);
obj.rbm_vis = obj.rbm_vis.addVisualGeometryToBody(obj.rbm_vis.getNumBodies(), leg);
%Add hips
obj.rbm_vis = obj.rbm_vis.addRobotFromURDF(particle_urdf, [], [], options);
body = obj.rbm_vis.body(end);
body.visual_geometry{1} = body.visual_geometry{1}.setColor(colors(:,j));
body.visual_geometry{1}.radius = 0.03;
obj.rbm_vis = obj.rbm_vis.setBody(obj.rbm_vis.getNumBodies(), body);
end
obj.rbm_vis = obj.rbm_vis.compile();
obj.littleDog = obj.littleDog.removeCollisionGroupsExcept('body');
obj.littleDog = obj.littleDog.compile();
end
function q_data = getQData(obj)
q_data = obj.q_data;
end
function constructVisualizer(obj)
obj.rbm_vis = obj.rbm_vis.compile();
obj.v = HopperVisualizer(obj.rbm_vis.constructVisualizer());
end
function num_positions = getNumPositions(obj)
num_positions = obj.rbm_vis.getNumPositions();
end
function loadResults(obj, data)
m = obj.littleDog.getMass();
obj.t_data = sqrt(obj.leg_length/9.81)*data.t;
obj.r_data = obj.leg_length*data.r;
obj.v_data = obj.leg_length/sqrt(obj.leg_length/9.81)*data.v;
obj.F_data = m*9.81*data.F;
obj.r_hip_data = obj.leg_length*data.r_hip;
obj.p_data = obj.leg_length*data.p;
obj.f_data = data.f;%*obj.littleDog.getMass()*9.81;
obj.f_data_vis = data.f*2*obj.leg_length;%*obj.littleDog.getMass()*9.81;
obj.th_data = data.th;
obj.T_data = data.T*obj.littleDog.getMass()*9.81*obj.leg_length;
obj.k_data = data.k*(obj.littleDog.getMass()*obj.leg_length^2)*sqrt(9.81/obj.leg_length);
obj.region_indicators = round(data.region_indicators);
obj.body_region_indicators = round(data.body_region_indicators);
N = size(obj.r_data, 2);
n_feet = size(obj.p_data, 3);
leg_pitch_data = zeros(1, N, double(n_feet));
for j = 1:double(n_feet)
leg_pitch_data(:,:,j) = -atan2(obj.p_data(1,:,j), -obj.p_data(2,:,j));
end
q_data = zeros(obj.rbm_vis.getNumPositions(), N);
q_data([1,3], :) = obj.r_data;
q_data(5, :) = obj.th_data;
for j = 1:n_feet
q_data(6 + 12*(j-1) + [1,3], :) = obj.r_data + obj.r_hip_data(:,:,j) + obj.p_data(:,:,j);
q_data(6 + 12*(j-1) + 5, :) = leg_pitch_data(:,:,j);
q_data(6 + 12*(j-1) + [7,9], :) = obj.r_data + obj.r_hip_data(:,:,j);
end
obj.q_data = q_data;
t = obj.t_data;
qtraj = PPTrajectory(foh(t, q_data));
obj.qtraj = qtraj.setOutputFrame(obj.rbm_vis.getPositionFrame());
obj.Ftraj = PPTrajectory(zoh(t, reshape(permute(obj.f_data_vis, [1, 3, 2]),[],N)));
obj.T_actual = sum((obj.p_data(1,:,:)+obj.r_hip_data(1,:,:)).*obj.f_data(2,:,:) - (obj.p_data(2,:,:)+obj.r_hip_data(2,:,:)).*obj.f_data(1,:,:),3);
end
function playback(obj, speed)
obj.v.playback_speed = speed;
obj.v.playback(obj.qtraj, obj.Ftraj)
end
function Istar = getDimensionlessMomentOfInertia(obj)
A = obj.littleDog.centroidalMomentumMatrix(obj.littleDog.home());
I = A(2,5);
m = obj.littleDog.getMass();
Istar = I/(m*obj.leg_length^2);
end
function hip_in_body = getHipInBody(obj)
hip_in_body = obj.hip_in_body;
end
function addRegion(obj, region)
if isempty(obj.regions)
obj.regions = region;
else
obj.regions(end+1) = region;
end
end
function addPlatform(obj, platform_start, platform_end, platform_height, platform_left, platform_right)
platform.start = platform_start*obj.leg_length;
platform.end = platform_end*obj.leg_length;
platform.height = platform_height*obj.leg_length;
platform.geom = RigidBodyBox([platform.end-platform.start; platform_left - platform_right; 0.1], [(platform.end+platform.start)/2; 0.5*(platform_left+platform_right); platform.height-0.05], [0; 0; 0]);
obj.platforms(end + 1) = platform;
obj.rbm_vis = obj.rbm_vis.addVisualGeometryToBody(1, platform.geom);
obj.rbm_vis = obj.rbm_vis.compile();
obj.littleDog = obj.littleDog.addGeometryToBody(1, platform.geom);
obj.littleDog = obj.littleDog.compile();
end
function hip_in_body = constructHipInBody(obj)
for name = {'front', 'back'}
str = name{1};
body = obj.littleDog.parseBodyOrFrameID(sprintf('%s_left_hip', str));
Ttree = obj.littleDog.getBody(body).Ttree;
hip_in_body.(str).x = Ttree(1,4)/obj.leg_length;
hip_in_body.(str).z = Ttree(3,4)/obj.leg_length;
end
end
function [sol, prog] = solveCDFKP(obj, seed, options)
if nargin < 3 || isempty(options), options = struct(); end
if nargin < 2, seed = []; end
options = obj.parseOptionsStruct(options);
assert(~isempty(obj.r_data), 'You must load data first!');
robot = obj.littleDog;
nq = robot.getNumPositions();
min_distance = 0.03;
foot = struct('id',[],'in_stance',[]);
foot(1,1).id = robot.findFrameId('front_left_foot_center');
foot(1,2).id = robot.findFrameId('front_right_foot_center');
foot(2,1).id = robot.findFrameId('back_left_foot_center');
foot(2,2).id = robot.findFrameId('back_right_foot_center');
foot_positions = bsxfun(@plus, obj.r_data, obj.r_hip_data + obj.p_data);
for i = 1:2
knee = RigidBodySphere(0.01, [(-1)^i*0.0265; 0; 0], [0; 0; 0]);
for j = 1:2
robot = robot.addCollisionGeometryToBody(robot.getFrame(foot(i,j).id).body_ind, knee);
end
end
robot = robot.compile();
N = size(obj.r_data, 2);
% Load nominal data
xstar = home(robot);
qstar = xstar(1:nq);
q0 = qstar;
qf = qstar;
% Set up time parameters
dt = diff(obj.t_data);
dt_range = [min(dt), max(dt)];
tf_range = [0.5*N*dt_range(1), 3*N*dt_range(2)];
%tf_range = [obj.t_data(end), obj.t_data(end)];
% Compute q_nom
q_nom = zeros(nq, N);
foot_constraints = cell(2,2);
for n = 1:N
for i = 1:2 %front-back
for j = 1:2 %left-right
lb = NaN(3,1);
lb([1,3]) = foot_positions(:, n, i);
ub = lb;
foot_constraints{i, j} = WorldPositionConstraint(robot, foot(i,j).id, zeros(3,1), lb, ub);
end
end
lb = zeros(3, 1);
lb([1,3]) = obj.r_data(:, n);
lb(2) = NaN;
ub = lb;
com_constraint = WorldCoMConstraint(robot, lb, ub);
hip_inds = robot.findPositionIndices('hip_roll');
posture_constraint = PostureConstraint(robot);
posture_constraint = posture_constraint.setJointLimits([2;4;5;6;hip_inds], [0; 0; -pi/8; 0; zeros(size(hip_inds))], [0; 0; pi/8; 0; zeros(size(hip_inds))]);
min_distance_constraint = MinDistanceConstraint(robot, min_distance);
ikoptions = IKoptions(robot);
if n == 1
qseed = qstar;
else
qseed = q_nom(:,n-1);
end
[q_nom(:, n), info, infeasible_constraint] = robot.inverseKin(qseed, qstar, foot_constraints{:}, com_constraint, posture_constraint, ikoptions);
assert(info < 10)
end
%keyboard
% Set up cost variables
%q_nom = bsxfun(@times,qstar,ones(1,N));
%q_nom(5,:) = obj.th_data;
%q_nom([1,3],:) = obj.r_data;
state_cost = Point(getStateFrame(robot),ones(getNumStates(robot),1));
state_cost.base_x = 0;
state_cost.base_y = 0;
state_cost.base_roll = 10;
state_cost.base_yaw = 10;
state_cost.front_left_hip_roll = 5;
state_cost.front_right_hip_roll = 5;
state_cost.back_left_hip_roll = 5;
state_cost.back_right_hip_roll = 5;
state_cost = double(state_cost);
Q = 1e3*diag(state_cost(1:nq));
state_cost(nq+1:nq+6) = 0;
Qv = 1e3*diag(state_cost(nq+1:end));
Q_comddot = diag([1,1,1]);
Q_contact_force = 5*eye(3);
% Set up common friction-cone variables
num_edges = 3;
FC_angles = linspace(0,2*pi,num_edges+1);FC_angles(end) = [];
% Set up contact wrench structure
contact_wrench_struct = struct('active_knot', {}, 'cw', {});
for i = 1:2 % front-back
for j = 1:numel(obj.regions)
mu = double(obj.regions(j).mu);
if mu > 0
idx = find(obj.region_indicators(j, :, i));
end_idx = find(diff(idx) ~= 1);
start_idx = end_idx + 1;
start_idx = [1, start_idx];
end_idx = [end_idx,numel(idx)];
time_index = cell(numel(start_idx),1);
for l = 1:numel(start_idx)
time_index{l} = idx(start_idx(l):(end_idx(l)-1));
end
if ~isempty(idx)
for k = 1:2 % left-right
FC_axis = zeros(3,1);
FC_axis([1,3]) = obj.regions(j).normal;
FC_perp1 = rotx(pi/2)*FC_axis;
FC_perp2 = cross(FC_axis, FC_perp1);
FC_edge = bsxfun(@plus, FC_axis, mu*(bsxfun(@times,cos(FC_angles),FC_perp1) + ...
bsxfun(@times,sin(FC_angles),FC_perp2)));
%contact_wrench_struct(end+1).active_knot = setdiff(idx, _idx);
contact_wrench_struct(end+1).active_knot = idx;
contact_wrench_struct(end).active_knot(end_idx) = [];
%contact_wrench_struct(end+1).active_knot = idx;
%contact_wrench_struct(end).cw = LinearFrictionConeWrench(robot,foot(i,k).id,zeros(3,1),FC_edge);
contact_wrench_struct(end).cw = FrictionConeWrench(robot,foot(i,k).id,zeros(3,1),mu,FC_axis);
end
end
end
end
end
options.time_option = 2;
prog = ComDynamicsFullKinematicsPlanner(robot,N,tf_range,Q_comddot,Qv,Q,q_nom,Q_contact_force,contact_wrench_struct,options);
%prog = prog.setCheckGrad(true);
% Add velocity constraints
max_joint_velocity = 4*pi;
lb = -max_joint_velocity*ones(nq-6, N);
ub = max_joint_velocity*ones(nq-6, N);
prog = prog.addConstraint(BoundingBoxConstraint(lb, ub), prog.v_inds(7:end, :));
% Add collision avoidance
prog = prog.addRigidBodyConstraint(MinDistanceConstraint(robot, min_distance),1:N);
% Add Timestep bounds
h_min = 0.5*dt_range(1); h_max = 3*dt_range(2);
prog = prog.addBoundingBoxConstraint(BoundingBoxConstraint(h_min*ones(N-1,1),h_max*ones(N-1,1)),prog.h_inds(:));
%prog = prog.addConstraint(ConstantConstraint(dt), prog.h_inds(:));
%prog = prog.addCost(QuadraticConstraint(-Inf, Inf, eye(numel(prog.h_inds)), zeros(numel(prog.h_inds),1)), prog.h_inds(:));
% Add symmetry constraints
prog = prog.addConstraint(obj.symmetryConstraint(obj.littleDog, 2:N), prog.q_inds(:,2:end));
%
% Constrain hip roll
posture_constraint = PostureConstraint(robot);
hip_inds = robot.findPositionIndices('hip_roll');
hip_pitch_inds = robot.findPositionIndices('hip_pitch');
knee_inds = robot.findPositionIndices('knee');
joint_inds = [hip_pitch_inds; hip_inds; knee_inds];
knee_lb = 60*pi/180;
hip_pitch_ub = 80*pi/180;
posture_constraint = posture_constraint.setJointLimits(joint_inds, [-hip_pitch_ub*ones(4, 1); zeros(size(hip_inds)); -Inf(2,1); knee_lb*ones(2,1)], [hip_pitch_ub*ones(4, 1); zeros(size(hip_inds)); -knee_lb*ones(2,1); Inf(2,1)]);
prog = prog.addRigidBodyConstraint(posture_constraint, 1:N);
% Add initial conditions
prog = prog.addConstraint(BoundingBoxConstraint(qstar(7:end)-0.1, qstar(7:end)+0.1), prog.q_inds(7:end,1));
prog = prog.addConstraint(ConstantConstraint(zeros(3,1)), prog.H_inds(:,1));
prog = prog.addConstraint(ConstantConstraint(zeros(3,1)), prog.Hdot_inds(:,1));
prog = prog.addConstraint(ConstantConstraint(obj.r_data(1,1)), prog.com_inds(1,1));
prog = prog.addConstraint(ConstantConstraint(zeros(3,1)), prog.comdot_inds(:,1));
%prog = prog.addConstraint(ConstantConstraint(zeros(3,1)), prog.comddot_inds(:,1));
prog = prog.addConstraint(ConstantConstraint(zeros(nq,1)), prog.v_inds(:,1));
% Add final conditions
%prog = prog.addConstraint(ConstantConstraint(qstar(7:end)), prog.q_inds(7:end,N));
prog = prog.addConstraint(ConstantConstraint(zeros(3,1)), prog.H_inds(:,N));
prog = prog.addConstraint(ConstantConstraint(zeros(3,1)), prog.Hdot_inds(:,N));
prog = prog.addConstraint(ConstantConstraint(obj.r_data(1,N)), prog.com_inds(1,N));
prog = prog.addConstraint(ConstantConstraint(zeros(3,1)), prog.comdot_inds(:,N));
%prog = prog.addConstraint(ConstantConstraint(zeros(3,1)), prog.comddot_inds(:,N));
prog = prog.addConstraint(ConstantConstraint(zeros(nq,1)), prog.v_inds(:,N));
% Add constraints on base
tol = 0.5;
%prog = prog.addConstraint(BoundingBoxConstraint(obj.r_data(1,:)-tol, obj.r_data(1,:)+tol), prog.com_inds(1,:));
%prog = prog.addConstraint(BoundingBoxConstraint(obj.r_data(2,:)-tol, obj.r_data(2,:)+tol), prog.com_inds(3,:));
%prog = prog.addConstraint(BoundingBoxConstraint(obj.k_data-tol, obj.k_data+tol), prog.H_inds(2,:));
prog = prog.addConstraint(BoundingBoxConstraint(zeros(N,1), zeros(N,1)), prog.q_inds(2,:));
prog = prog.addConstraint(BoundingBoxConstraint(zeros(N,1), zeros(N,1)), prog.q_inds(4,:));
prog = prog.addConstraint(BoundingBoxConstraint(zeros(N,1), zeros(N,1)), prog.q_inds(6,:));
prog = prog.addConstraint(BoundingBoxConstraint(-pi/8*ones(N,1), pi/8*ones(N,1)), prog.q_inds(5,:));
% Foot region constraints
foot_position_fcn = cell(2);
for i = 1:2
for k = 1:2
foot_position_fcn{i, k} = drakeFunction.kinematic.WorldPosition(robot, foot(i,k).id);
for n = 1:N
tol = 1e-1;
lb = -tol*ones(2,1);
ub = tol*ones(2,1);
if k == 1
lb = [lb(1); 0; lb(2)];
ub = [ub(1); Inf; ub(2)];
else
lb = [lb(1); -Inf; lb(2)];
ub = [ub(1); 0; ub(2)];
end
foot_position = zeros(3, 1);
foot_position([1, 3]) = foot_positions(:, n, i);
xz_error_fcn = drakeFunction.Affine([1, 0, 0; 0, 0, 1], -foot_positions(:, n, i));
%xz_error_fcn = drakeFunction.Affine(eye(3), -foot_position);
norm_squared_fcn = drakeFunction.euclidean.NormSquared(2);
cost = DrakeFunctionConstraint(-Inf, Inf, norm_squared_fcn(xz_error_fcn(foot_position_fcn{i,k})));
%constraint = DrakeFunctionConstraint(lb, ub, xz_error_fcn(foot_position_fcn{i,k}));
cnstr_inds = prog.q_inds(:,n);
prog = prog.addCost(cost, cnstr_inds);
%prog = prog.addConstraint(constraint, cnstr_inds);
end
end
end
for j = 1:numel(obj.regions)
A = [obj.regions(j).A; obj.regions(j).Aeq; -obj.regions(j).Aeq]./obj.leg_length;
A = [A(:, 1), zeros(size(A,1), 1), A(:, 2)];
b = [obj.regions(j).b; obj.regions(j).beq; -obj.regions(j).beq];
region_fcn = drakeFunction.Affine(A, -b);
lb = -Inf(size(b));
ub = zeros(size(b));
% body region constraint
idx = find(obj.body_region_indicators(j, :));
constraint = DrakeFunctionConstraint(lb, ub-0.25, region_fcn);
for time_index = reshape(idx, 1, [])
cnstr_inds = prog.q_inds(1:3,time_index);
prog = prog.addConstraint(constraint,cnstr_inds);
end
% leg region constraints
for i = 1:2 % front-back
idx = find(obj.region_indicators(j, :, i));
if ~isempty(idx)
for k = 1:2 % left-right
fcn = region_fcn(foot_position_fcn{i,k});
% debug = drakeFunction.Debug(fcn.dim_output, @(f,df) display(f(f>0)));
% fcn = compose(debug, fcn);
constraint = DrakeFunctionConstraint(lb, ub, fcn);
for time_index = reshape(idx, 1, [])
cnstr_inds = prog.q_inds(:,time_index);
prog = prog.addConstraint(constraint,cnstr_inds);
end
if obj.regions(j).mu > 0
% stance feet stationary constraints
end_idx = find(diff(idx) ~= 1);
start_idx = end_idx + 1;
start_idx = [1, start_idx];
end_idx = [end_idx,numel(idx)];
time_index = cell(numel(start_idx),1);
for l = 1:numel(start_idx)
time_index{l} = idx(start_idx(l):end_idx(l));
if time_index{l}(1) ~= 1
time_index{l} = [time_index{l}(1)-1, time_index{l}];
end
end
prog = prog.addRigidBodyConstraint(WorldFixedPositionConstraint(robot,foot(i,k).id, zeros(3,1)),time_index);
else
% swing feet collision avoidance
swing_idx = setdiff([idx-1, idx+1], idx);
swing_idx(swing_idx < 1 | swing_idx > N) = [];
for time_index = reshape(swing_idx, 1, [])
cnstr_inds = prog.q_inds(:,time_index);
prog = prog.addConstraint(constraint,cnstr_inds);
end
end
end
end
end
end
% Set up seed
if isempty(seed)
x_seed = zeros(prog.num_vars,1);
q_seed = q_nom;
%q_seed([1, 3], :) = obj.r_data;
%q_seed(5, :) = obj.th_data;
v_seed = gradient(q_seed);
com_seed = zeros(3, N);
com_seed([1, 3], :) = obj.r_data;
comdot_seed = zeros(3, N);
comdot_seed([1, 3], :) = obj.v_data;
comddot_seed = zeros(3, N);
comddot_seed([1, 3], :) = (1/obj.littleDog.getMass())*obj.F_data;
I = obj.getDimensionlessMomentOfInertia()*obj.littleDog.getMass()*obj.leg_length^2;
H_seed = zeros(3, N);
H_seed(2,:) = obj.k_data;
Hdot_seed = gradient(H_seed);
x_seed(prog.H_inds(:)) = reshape(H_seed,[],1);
x_seed(prog.Hdot_inds(:)) = reshape(Hdot_seed,[],1);
x_seed(prog.h_inds) = dt;
x_seed(prog.q_inds(:)) = reshape(q_seed,[],1);
x_seed(prog.v_inds(:)) = reshape(v_seed,[],1);
x_seed(prog.com_inds(:)) = reshape(com_seed,[],1);
x_seed(prog.comdot_inds(:)) = reshape(comdot_seed,[],1);
x_seed(prog.comddot_inds(:)) = reshape(comddot_seed,[],1);
for i = 1:2
for j = 1:2
f_seed = zeros(3, N);
f_seed([1, 3], :) = obj.f_data(:, :, i);
x_seed(prog.lambda_inds{2*(i-1) + j}(:)) = 0.5*f_seed;
%prog = prog.addCost(QuadraticConstraint(-Inf, Inf, eye(numel(f_seed)), -2*f_seed(:)), prog.lambda_inds{2*(i-1) + j}(:));
end
end
else
x_seed = seed.x_sol;
end
% Set up solver options
%prog = prog.setSolver('ipopt');
prog = prog.setSolverOptions('snopt','iterationslimit',1e6);
prog = prog.setSolverOptions('snopt','majoriterationslimit', options.major_iteration_limit);
prog = prog.setSolverOptions('snopt','majorfeasibilitytolerance',1e-5);
prog = prog.setSolverOptions('snopt','majoroptimalitytolerance',6e-4);
prog = prog.setSolverOptions('snopt','superbasicslimit',2000);
prog = prog.setSolverOptions('snopt','linesearchtolerance',0.99);
prog = prog.setSolverOptions('snopt','scaleoption',1);
%prog = prog.setSolverOptions('snopt','sense','Feasible point');
prog = prog.setSolverOptions('snopt','print','snopt.out');
% Solve trajectory optimization
tic
%profile on;
[x_sol,~,~] = prog.solve(x_seed);
%profile off;
toc
% Parse trajectory optimization output
sol.x_sol = x_sol;
sol.q = reshape(x_sol(prog.q_inds(:)),nq,N);
sol.v = reshape(x_sol(prog.v_inds(:)),nq,N);
sol.h = reshape(x_sol(prog.h_inds),1,[]);
sol.t = cumsum([0 sol.h]);
sol.com = reshape(x_sol(prog.com_inds),3,[]);
sol.comdot = reshape(x_sol(prog.comdot_inds),3,[]);
sol.comddot = reshape(x_sol(prog.comddot_inds),3,[]);
sol.H = reshape(x_sol(prog.H_inds),3,[]);
sol.Hdot = reshape(x_sol(prog.Hdot_inds),3,[])*prog.torque_multiplier;
sol.lambda = cell(2,1);
for i = 1:numel(prog.lambda_inds)
sol.lambda{i} = reshape(x_sol(prog.lambda_inds{i}),size(prog.lambda_inds{i},1),[],N);
end
sol.xtraj= PPTrajectory(foh(sol.t,[sol.q;sol.v]));
sol.xtraj= sol.xtraj.setOutputFrame(robot.getStateFrame);
end
function delete(obj)
obj.rbm = [];
obj.rbm_vis = [];
obj.littleDog = [];
obj.hopper = [];
obj.platforms = [];
obj.hip_in_body = [];
obj.leg_length = [];
obj.v = [];
obj.t_data = [];
obj.r_data = [];
obj.r_hip_data = [];
obj.p_data = [];
obj.f_data = [];
obj.th_data = [];
obj.T_data = [];
obj.q_data = [];
obj.T_actual = [];
obj.region_indicators = [];
obj.regions = [];
obj.qtraj = [];
obj.Ftraj = [];
end
end
methods (Static)
function options = parseOptionsStruct(options_in)
options = Hopper.defaultOptionsStruct();
for fieldname_cell = fields(options_in)'
fieldname = fieldname_cell{1};
if isfield(options,fieldname)
options.(fieldname) = options_in.(fieldname);
end
end
end
function options = defaultOptionsStruct()
options.visualize = true;
options.major_iteration_limit = 200;
options.suffix = '';
end
function symmetry_cnstr = symmetryConstraint(robot,t_idx)
N = numel(t_idx);
num_symmetry = 6;
symmetry_matrix = zeros(num_symmetry,robot.getNumPositions);
function mat = addSymmetricPair(mat, rows, joint1_name, joint2_name)
joint1_idx = robot.getBody(robot.findJointId(joint1_name)).position_num;
joint2_idx = robot.getBody(robot.findJointId(joint2_name)).position_num;
mat(rows, [joint1_idx, joint2_idx]) = [-1, 1];
end
function mat = addAntiSymmetricPair(mat, rows, joint1_name, joint2_name)
joint1_idx = robot.getBody(robot.findJointId(joint1_name)).position_num;
joint2_idx = robot.getBody(robot.findJointId(joint2_name)).position_num;
mat(rows, [joint1_idx, joint2_idx]) = [1, 1];
end
symmetry_matrix = addAntiSymmetricPair(symmetry_matrix, 1, ...
'front_left_hip_roll', ...
'front_right_hip_roll');
symmetry_matrix = addSymmetricPair(symmetry_matrix, 2, ...
'front_left_hip_pitch', ...
'front_right_hip_pitch');
symmetry_matrix = addSymmetricPair(symmetry_matrix, 3, ...
'front_left_knee', ...
'front_right_knee');
symmetry_matrix = addAntiSymmetricPair(symmetry_matrix, 4, ...
'back_left_hip_roll', ...
'back_right_hip_roll');
symmetry_matrix = addSymmetricPair(symmetry_matrix, 5, ...
'back_left_hip_pitch', ...
'back_right_hip_pitch');
symmetry_matrix = addSymmetricPair(symmetry_matrix, 6, ...
'back_left_knee', ...
'back_right_knee');
symmetry_cnstr = LinearConstraint(zeros(num_symmetry*N,1),zeros(num_symmetry*N,1),kron(speye(N),symmetry_matrix));
cnstr_name = cell(num_symmetry*N,1);
for i = 1:N
cnstr_name{(i-1)*num_symmetry+1} = sprintf('front_hip_roll_symmetry[%d]',t_idx(i));
cnstr_name{(i-1)*num_symmetry+2} = sprintf('front_hip_pitch_symmetry[%d]',t_idx(i));
cnstr_name{(i-1)*num_symmetry+3} = sprintf('front_knee_symmetry[%d]',t_idx(i));
cnstr_name{(i-1)*num_symmetry+4} = sprintf('back_hip_roll_symmetry[%d]',t_idx(i));
cnstr_name{(i-1)*num_symmetry+5} = sprintf('back_hip_pitch_symmetry[%d]',t_idx(i));
cnstr_name{(i-1)*num_symmetry+6} = sprintf('back_knee_symmetry[%d]',t_idx(i));
end
symmetry_cnstr = symmetry_cnstr.setName(cnstr_name);
end
end
end