Repository navigation
Expand file tree
/
Copy pathfull_stack_av.html
More file actions
481 lines (434 loc) · 30.1 KB
/
Copy pathfull_stack_av.html
File metadata and controls
481 lines (434 loc) · 30.1 KB
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
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>Full-Stack AV | IEKF Localization Design Doc</title>
<meta name="description" content="Design doc for a full-stack autonomous-vehicle localization system: Invariant EKF sensor fusion with learned perception modules." />
<link rel="icon" href="michigan_logo.png" type="image/png" />
<link rel="preconnect" href="https://fonts.googleapis.com" />
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
<link
href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&family=Newsreader:ital,opsz,wght@0,16..72,400;0,16..72,500;0,16..72,600&display=swap"
rel="stylesheet"
/>
<script src="https://cdn.tailwindcss.com"></script>
<script>
tailwind.config = {
theme: {
extend: {
fontFamily: {
sans: ["Inter", "system-ui", "sans-serif"],
serif: ["Newsreader", "Georgia", "serif"],
},
colors: {
accent: {
50: "#eff6ff",
100: "#dbeafe",
500: "#3b82f6",
600: "#2563eb",
700: "#1d4ed8",
800: "#1e40af",
},
},
},
},
};
</script>
<style>
html { scroll-behavior: smooth; }
body { font-family: "Inter", system-ui, sans-serif; }
section[id] { scroll-margin-top: 5rem; }
.tag {
font-size: 11px;
font-weight: 600;
padding: 4px 10px;
border-radius: 9999px;
color: #475569;
background: #f1f5f9;
border: 1px solid #e2e8f0;
}
</style>
</head>
<body class="bg-white text-slate-600 antialiased">
<!-- Top Nav -->
<header class="sticky top-0 z-50 backdrop-blur-md bg-white/85 border-b border-slate-100">
<div class="max-w-6xl mx-auto px-4 py-4 flex items-center justify-between">
<a href="index.html" class="text-sm font-medium text-accent-700 hover:text-accent-800 transition-colors">← Back to Home</a>
<div class="hidden md:flex items-center gap-3 text-xs text-slate-400">
<a href="#overview" class="hover:text-slate-800 transition-colors">Overview</a>
<span class="text-slate-300">·</span>
<a href="#system-map" class="hover:text-slate-800 transition-colors">System Map</a>
<span class="text-slate-300">·</span>
<a href="#deliverables" class="hover:text-slate-800 transition-colors">Deliverables</a>
<span class="text-slate-300">·</span>
<a href="#plan" class="hover:text-slate-800 transition-colors">Step-by-Step</a>
<span class="text-slate-300">·</span>
<a href="#ros2" class="hover:text-slate-800 transition-colors">ROS 2 Graph</a>
<span class="text-slate-300">·</span>
<a href="#evaluation" class="hover:text-slate-800 transition-colors">Evaluation</a>
</div>
</div>
</header>
<!-- Hero -->
<section id="overview" class="max-w-6xl mx-auto px-4 py-12">
<div class="bg-white border border-slate-200 rounded-2xl p-8">
<h1 class="font-serif text-3xl md:text-4xl font-semibold text-slate-900 mb-3">
Full-Stack AV: IEKF Localization + Deep Learning Modules
</h1>
<p class="text-slate-500 leading-relaxed max-w-4xl">
A system design document for an autonomous-vehicle-grade localization stack centered on an
<span class="text-slate-800 font-semibold">Invariant EKF (IEKF)</span> that fuses
<span class="text-slate-800 font-semibold">IMU</span>, <span class="text-slate-800 font-semibold">wheel odometry</span>,
<span class="text-slate-800 font-semibold">GNSS/RTK</span>, <span class="text-slate-800 font-semibold">camera VO/VIO</span>
(stereo + optional monocular), and <span class="text-slate-800 font-semibold">LiDAR scan-to-map</span>.
Deep learning components improve robustness via
<span class="text-slate-800 font-semibold">learned matching</span>,
<span class="text-slate-800 font-semibold">learned uncertainty calibration</span>,
<span class="text-slate-800 font-semibold">slip detection</span>,
<span class="text-slate-800 font-semibold">semantic map corrections</span>,
and <span class="text-slate-800 font-semibold">global relocalization recovery</span>.
</p>
<div class="mt-6 flex flex-wrap gap-2 text-xs font-semibold">
<span class="tag">#IEKF</span>
<span class="tag">#SensorFusion</span>
<span class="tag">#IMU</span>
<span class="tag">#WheelOdom</span>
<span class="tag">#GNSS/RTK</span>
<span class="tag">#StereoVO</span>
<span class="tag">#LiDARToMap</span>
<span class="tag">#DeepLearning</span>
<span class="tag">#ROS2</span>
</div>
<div class="mt-6 grid md:grid-cols-3 gap-4">
<div class="bg-slate-50 border border-slate-200 rounded-xl p-4">
<div class="text-slate-400 text-xs">Primary Output</div>
<div class="font-semibold">Pose + Covariance</div>
<div class="text-slate-500 text-sm mt-1">/odometry/filtered + /tf (map → base_link)</div>
</div>
<div class="bg-slate-50 border border-slate-200 rounded-xl p-4">
<div class="text-slate-400 text-xs">Reliability Output</div>
<div class="font-semibold">Sensor Health + Gating</div>
<div class="text-slate-500 text-sm mt-1">Innovation checks + learned confidence scaling</div>
</div>
<div class="bg-slate-50 border border-slate-200 rounded-xl p-4">
<div class="text-slate-400 text-xs">Recovery Output</div>
<div class="font-semibold">Relocalization</div>
<div class="text-slate-500 text-sm mt-1">Place recognition + re-init scan-to-map</div>
</div>
</div>
</div>
</section>
<!-- System Map -->
<section id="system-map" class="max-w-6xl mx-auto px-4 pb-12">
<h2 class="font-serif text-2xl font-semibold text-slate-900 mb-4">System Map (Localization + ML Modules)</h2>
<div class="grid lg:grid-cols-2 gap-6">
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<p class="text-slate-500 mb-4">
The IEKF runs continuously. Sensors provide updates. Deep learning modules improve the
<span class="text-slate-800 font-semibold">quality</span> and
<span class="text-slate-800 font-semibold">trustworthiness</span> of measurements (and can trigger relocalization).
</p>
<div class="bg-slate-900 border border-slate-800 rounded-xl p-4 overflow-x-auto">
<pre class="text-xs md:text-sm text-slate-200 leading-relaxed">
(RTK Corrections / NTRIP / Base Station)
|
v
+--------------------------------------------------------------+
| IEKF Core |
| State: {R, p, v, bg, ba, (optional: wheel_scale, cam_scale)}|
| Cov: P |
| Outputs: /odometry/filtered, /tf (map->base_link), diag |
+--------------------------^-----------------------------------+
|
IMU propagation |
[IMU] ---> Strapdown integration + covariance propagation
|
+--> (DL/ML Health): detect saturation / vibration / dropouts
Measurement Update Inputs (with ML modules):
--------------------------------------------
[Wheel Encoders + Steering] ---> Vehicle model ---> (Update)
| \
| +--> (ML Slip Detector) -> down-weight wheel updates
v
(ML Slip Classifier: slip/no-slip + slip score)
[GNSS/RTK] ----------------------------------------------> (Update)
|
+--> (ML GNSS reliability) -> multipath suspicion -> inflate R or reject
[Stereo Camera] --> (CNN keypoints+matching: SuperPoint + LightGlue) --> Stereo VO --> (Update)
| |
| +--> (ML Uncertainty Calibrator) -> scale R_vo
|
+--> (Optional CNN stereo depth) -> denser 3D points -> stronger VO constraints
[LiDAR] --> scan-to-map (NDT/ICP) --> pose correction --> (Update)
|
+--> (ML/heuristic degeneracy + uncertainty calibration) -> scale R_lidar
Semantic Map Localization (DL + Geometry):
-----------------------------------------
[Camera] -> (DL lane/sign/pole detection) -> associate to HD semantic map -> absolute pose -> (Update)
Global Recovery (Relocalization):
--------------------------------
[LiDAR or Camera] -> (Place Recognition / Descriptor Retrieval) -> coarse global pose -> re-init scan-to-map + IEKF
</pre>
</div>
</div>
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-3">What Deep Learning Changes</h3>
<ul class="text-slate-500 space-y-2 list-disc list-inside">
<li><span class="text-slate-800 font-semibold">Better visual constraints:</span> learned matching stabilizes VO under blur/lighting/low texture.</li>
<li><span class="text-slate-800 font-semibold">Better filter consistency:</span> learned confidence scales measurement noise to prevent divergence.</li>
<li><span class="text-slate-800 font-semibold">Vehicle realism:</span> slip detection prevents wheel odometry from corrupting the estimate.</li>
<li><span class="text-slate-800 font-semibold">Map-locked accuracy:</span> semantic landmarks give absolute corrections even when GNSS degrades.</li>
<li><span class="text-slate-800 font-semibold">Recovery:</span> relocalization finds the right area in the map after failures.</li>
</ul>
<div class="mt-6">
<h3 class="text-base font-bold text-slate-800 mb-3">Core Outputs</h3>
<ul class="text-slate-500 space-y-2 list-disc list-inside">
<li><span class="text-slate-800 font-semibold">Pose:</span> <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">T_map_base</code>, <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">T_odom_base</code></li>
<li><span class="text-slate-800 font-semibold">Velocity:</span> world/body velocity + yaw-rate</li>
<li><span class="text-slate-800 font-semibold">Biases:</span> IMU gyro bias <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">bg</code>, accel bias <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">ba</code></li>
<li><span class="text-slate-800 font-semibold">Covariance:</span> full state covariance matrix</li>
<li><span class="text-slate-800 font-semibold">Health:</span> per-sensor trust score + rejection reasons</li>
</ul>
</div>
</div>
</div>
</section>
<!-- Deliverables -->
<section id="deliverables" class="max-w-6xl mx-auto px-4 pb-12">
<h2 class="font-serif text-2xl font-semibold text-slate-900 mb-4">Deliverables (Software + ML + Results)</h2>
<div class="grid md:grid-cols-2 gap-6">
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-3">Software Deliverables</h3>
<ul class="text-slate-500 space-y-2 list-disc list-inside">
<li><span class="text-slate-800 font-semibold">ROS 2 IEKF localizer</span> (propagation + measurement updates + tf publishing)</li>
<li><span class="text-slate-800 font-semibold">Sensor adapters</span>: IMU, wheel/steering, GNSS, stereo VO, LiDAR scan-to-map</li>
<li><span class="text-slate-800 font-semibold">Deep matching node</span>: SuperPoint + LightGlue/SuperGlue (stereo correspondences)</li>
<li><span class="text-slate-800 font-semibold">Uncertainty calibration node</span>: learned confidence → adaptive measurement noise scaling</li>
<li><span class="text-slate-800 font-semibold">Slip detection node</span>: classifier → down-weight wheel updates during low traction</li>
<li><span class="text-slate-800 font-semibold">Semantic localization node</span>: lanes/landmarks detection → map alignment → absolute update</li>
<li><span class="text-slate-800 font-semibold">Relocalization node</span>: place recognition → coarse pose init → recovery pipeline</li>
<li><span class="text-slate-800 font-semibold">Evaluation toolkit</span>: ATE/RPE, yaw error, dropout drift curves, innovation stats</li>
<li><span class="text-slate-800 font-semibold">Reproducibility</span>: launch files, config YAMLs, calibration docs, bag datasets</li>
</ul>
</div>
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-3">Results Deliverables</h3>
<ul class="text-slate-500 space-y-2 list-disc list-inside">
<li><span class="text-slate-800 font-semibold">Ablation plots</span>: IMU+wheel → +GNSS → +VO → +LiDAR-to-map → +semantics → +relocalization</li>
<li><span class="text-slate-800 font-semibold">Dropout tests</span>: GNSS denied segments (10–60s), camera blackout, LiDAR degeneracy</li>
<li><span class="text-slate-800 font-semibold">Consistency checks</span>: innovation gating rates, residual histograms, “overconfidence” prevention</li>
<li><span class="text-slate-800 font-semibold">Robustness wins</span>: learned matching + uncertainty scaling reduces failures</li>
<li><span class="text-slate-800 font-semibold">Final report</span>: system architecture + design decisions + quantitative metrics + failure analysis</li>
</ul>
</div>
</div>
</section>
<!-- Step-by-step plan -->
<section id="plan" class="max-w-6xl mx-auto px-4 pb-16">
<h2 class="font-serif text-2xl font-semibold text-slate-900 mb-4">Step-by-Step Build Plan (Full Stack, Very Detailed)</h2>
<div class="space-y-6">
<!-- Phase 0 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 0: Specs, Frames, State, and Success Metrics</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Choose conventions:</span> ENU vs NED, yaw definition, right-handed frames.</li>
<li><span class="text-slate-800 font-semibold">Define frames:</span> <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">map</code>, <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">odom</code>, <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">base_link</code>, sensor frames.</li>
<li><span class="text-slate-800 font-semibold">State vector:</span> SE(3) (R,p,v,bg,ba) or SE(2)+attitude; optionally wheel scale + camera scale.</li>
<li><span class="text-slate-800 font-semibold">Measurement list:</span> wheel speed/steering, GNSS position/vel, VO increment, LiDAR-to-map pose, semantic landmark pose.</li>
<li><span class="text-slate-800 font-semibold">Success metrics:</span> RMS position error, yaw error, drift under GNSS dropout, failure rate, relocalization time.</li>
</ol>
</div>
<!-- Phase 1 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 1: Data Plumbing, Time Sync, and Calibration</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Timestamp discipline:</span> ring buffers per sensor + interpolation to measurement time.</li>
<li><span class="text-slate-800 font-semibold">Extrinsics:</span> calibrate <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">T_base_imu</code>, <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">T_base_cam</code>, <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">T_base_lidar</code>, GNSS lever arm.</li>
<li><span class="text-slate-800 font-semibold">Noise characterization:</span> IMU noise/bias; wheel quantization; GNSS quality flags; VO feature stats.</li>
<li><span class="text-slate-800 font-semibold">Unit checks:</span> rad/s, m/s², meters, seconds, with no hidden conversions.</li>
<li><span class="text-slate-800 font-semibold">tf tree validation:</span> static transforms correct; no frame flips; consistent orientation.</li>
</ol>
</div>
<!-- Phase 2 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 2: IEKF Core (Propagation + Covariance)</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">IMU strapdown:</span> integrate R,v,p with bias-corrected gyro/accel.</li>
<li><span class="text-slate-800 font-semibold">Bias model:</span> random walk for <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">bg</code>, <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">ba</code>.</li>
<li><span class="text-slate-800 font-semibold">Covariance propagation:</span> continuous-time error dynamics → discretize → update P.</li>
<li><span class="text-slate-800 font-semibold">Sanity guards:</span> dt bounds, NaN checks, accel magnitude clamps, bias bounds.</li>
<li><span class="text-slate-800 font-semibold">Baseline output:</span> publish propagation-only odom (no updates) to validate stability.</li>
</ol>
</div>
<!-- Phase 3 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 3: Wheel + Steering Update (With ML Slip Detection)</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Vehicle model:</span> bicycle model (preferred) using steering angle; else velocity-only.</li>
<li><span class="text-slate-800 font-semibold">Wheel measurement:</span> ticks → wheel speed → forward velocity (and yaw-rate if available).</li>
<li><span class="text-slate-800 font-semibold">Slip detector (ML):</span> train classifier using features:
IMU accel spikes, wheel-vs-IMU inconsistency, yaw-rate mismatch, sudden wheel speed changes.</li>
<li><span class="text-slate-800 font-semibold">Fusion rule:</span> when slip score high → inflate wheel measurement noise (or reject update).</li>
<li><span class="text-slate-800 font-semibold">Validation:</span> compare wheel velocity to GNSS velocity (when available), and analyze slip-trigger events.</li>
</ol>
</div>
<!-- Phase 4 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 4: GNSS/RTK Update (With Reliability ML)</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Measurement model:</span> GNSS position/velocity in map frame; include lever arm from base.</li>
<li><span class="text-slate-800 font-semibold">Quality gating:</span> filter by fix type (RTK fixed/float), HDOP/VDOP thresholds, innovation gating.</li>
<li><span class="text-slate-800 font-semibold">GNSS reliability ML:</span> predict multipath risk using quality flags + innovation patterns → scale GNSS noise.</li>
<li><span class="text-slate-800 font-semibold">Dropout testing:</span> remove GNSS for 10–60 seconds and measure drift growth curves.</li>
</ol>
</div>
<!-- Phase 5 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 5: Stereo VO/VIO (With CNN Matching + Uncertainty Calibration)</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Start baseline:</span> classic stereo VO with feature tracking + RANSAC + triangulation.</li>
<li><span class="text-slate-800 font-semibold">Upgrade front-end (CNN):</span> SuperPoint keypoints + LightGlue/SuperGlue matching.</li>
<li><span class="text-slate-800 font-semibold">Pose extraction:</span> estimate relative motion Δpose from stereo constraints; compute inlier stats.</li>
<li><span class="text-slate-800 font-semibold">Uncertainty calibration (ML):</span> model predicts VO confidence from inliers, reprojection error,
track length, blur/brightness metrics → scale <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">R_vo</code>.</li>
<li><span class="text-slate-800 font-semibold">Failure detection:</span> low inliers / high reprojection error → reject update.</li>
<li><span class="text-slate-800 font-semibold">(Optional)</span> CNN stereo depth for denser points in low-texture scenes.</li>
</ol>
</div>
<!-- Phase 6 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 6: LiDAR Scan-to-Map (With Degeneracy + Confidence Scaling)</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Map:</span> create/load a pointcloud map (PCD) of your environment.</li>
<li><span class="text-slate-800 font-semibold">Localization:</span> NDT/ICP scan-to-map matching yields pose correction.</li>
<li><span class="text-slate-800 font-semibold">Degeneracy checks:</span> low overlap, high condition number, planar scene → reduce trust.</li>
<li><span class="text-slate-800 font-semibold">Uncertainty scaling:</span> use match fitness + degeneracy to scale <code class="text-accent-700 bg-accent-50 px-1 py-0.5 rounded">R_lidar</code>.</li>
<li><span class="text-slate-800 font-semibold">Initialization:</span> GNSS/IEKF prior initializes scan-to-map to avoid wrong minima.</li>
</ol>
</div>
<!-- Phase 7 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 7: Semantic Map Localization (DL + Geometry)</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Semantic perception:</span> lane segmentation + pole/sign/landmark detection (CNN).</li>
<li><span class="text-slate-800 font-semibold">Association:</span> match detected semantics to a semantic HD map (landmark database).</li>
<li><span class="text-slate-800 font-semibold">Pose correction:</span> solve pose from landmark geometry (PnP / alignment) and update IEKF.</li>
<li><span class="text-slate-800 font-semibold">Confidence:</span> use detection confidence + association score to scale measurement noise.</li>
</ol>
</div>
<!-- Phase 8 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 8: Global Relocalization Recovery (Place Recognition)</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Trigger conditions:</span> large residual spikes, repeated update rejections, scan-to-map failure, pose jump.</li>
<li><span class="text-slate-800 font-semibold">Place recognition:</span> compute global descriptor (LiDAR scan descriptor or image descriptor) and retrieve best map region.</li>
<li><span class="text-slate-800 font-semibold">Coarse init:</span> use retrieved region to seed scan-to-map and reset/realign IEKF to map.</li>
<li><span class="text-slate-800 font-semibold">Verification:</span> accept recovery only if subsequent scan-to-map fitness is strong.</li>
</ol>
</div>
<!-- Phase 9 -->
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-2">Phase 9: Evaluation, Ablations, and “Production” Hardening</h3>
<ol class="text-slate-500 space-y-2 list-decimal list-inside">
<li><span class="text-slate-800 font-semibold">Ablations:</span> turn modules on/off and compare metrics.</li>
<li><span class="text-slate-800 font-semibold">Latency audits:</span> measure delay for VO/LiDAR/DL nodes; compensate by buffering and applying correct timestamps.</li>
<li><span class="text-slate-800 font-semibold">Consistency:</span> track innovation distributions, gating rates, covariance growth during dropouts.</li>
<li><span class="text-slate-800 font-semibold">Failure analysis:</span> multipath, slip, motion blur, textureless scenes, degeneracy corridors.</li>
<li><span class="text-slate-800 font-semibold">Finalize:</span> stable configs + README + bag files + plotted report.</li>
</ol>
</div>
</div>
</section>
<!-- ROS 2 Node Graph -->
<section id="ros2" class="max-w-6xl mx-auto px-4 pb-16">
<h2 class="font-serif text-2xl font-semibold text-slate-900 mb-4">ROS 2 Node Graph (Full Stack)</h2>
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<div class="bg-slate-900 border border-slate-800 rounded-xl p-4 overflow-x-auto">
<pre class="text-xs md:text-sm text-slate-200 leading-relaxed">
Sensors:
/imu_driver -> /imu/data
/wheel_driver -> /wheel/encoders
/steering_driver -> /vehicle/steering
/gnss_driver -> /gnss/fix, /gnss/vel, /gnss/status
/stereo_cam -> /stereo/left/image, /stereo/right/image
/lidar_driver -> /points_raw
Deep Learning Modules:
/dl_matching : (left/right images) -> /vo/matches, /vo/quality
/dl_slip_detector : (imu + wheel + steering) -> /wheel/slip_score
/dl_uncertainty : (vo/lidar/gnss stats) -> /trust/scale_vo, /trust/scale_lidar, /trust/scale_gnss
/dl_semantics : (camera) -> /sem/lanes, /sem/landmarks
/relocalizer : (lidar/cam) -> /recovery/coarse_pose
Classical Geometry:
/stereo_vo : (matches + stereo) -> /vo/odom, /vo/diag
/lidar_scan_match : (points + map) -> /lidar/pose, /lidar/fitness
/semantic_localize : (semantics + map) -> /sem/pose, /sem/score
Estimator:
/iekf_localizer : fuses everything -> /odometry/filtered, /tf, /diagnostics, /health
Offline:
/mapping_pipeline -> map.pcd, semantic_landmarks.json, descriptor_db.bin
</pre>
</div>
<p class="text-slate-500 mt-4">
Key design rule: <span class="text-slate-800 font-semibold">IEKF owns the state</span>.
DL nodes produce either (1) better measurements, or (2) better measurement trust (adaptive noise),
or (3) recovery proposals (relocalization seed).
</p>
</div>
</section>
<!-- Evaluation -->
<section id="evaluation" class="max-w-6xl mx-auto px-4 pb-16">
<h2 class="font-serif text-2xl font-semibold text-slate-900 mb-4">Evaluation Plan (Metrics + Tests + Ablations)</h2>
<div class="grid md:grid-cols-2 gap-6">
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-3">Metrics</h3>
<ul class="text-slate-500 space-y-2 list-disc list-inside">
<li><span class="text-slate-800 font-semibold">ATE/RPE:</span> absolute/relative trajectory error (position + yaw)</li>
<li><span class="text-slate-800 font-semibold">Yaw error:</span> especially critical for lane-level planning</li>
<li><span class="text-slate-800 font-semibold">Drift under dropout:</span> growth rate during GNSS denial</li>
<li><span class="text-slate-800 font-semibold">Update rejection rate:</span> how often gating rejects sensors</li>
<li><span class="text-slate-800 font-semibold">Recovery time:</span> time from failure → relocalized</li>
<li><span class="text-slate-800 font-semibold">Consistency:</span> innovation histograms + “overconfidence” events</li>
</ul>
</div>
<div class="bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-3">Ablations (Required Plots)</h3>
<ul class="text-slate-500 space-y-2 list-disc list-inside">
<li>IMU + wheel only (dead reckoning baseline)</li>
<li>+ GNSS/RTK</li>
<li>+ Stereo VO (classic) vs + Stereo VO (CNN matching)</li>
<li>+ LiDAR scan-to-map</li>
<li>+ Learned uncertainty scaling (VO/LiDAR/GNSS)</li>
<li>+ Slip detection enabled</li>
<li>+ Semantic map localization enabled</li>
<li>+ Relocalization enabled (full robustness)</li>
</ul>
</div>
</div>
<div class="mt-6 bg-white border border-slate-200 rounded-2xl p-6">
<h3 class="text-base font-bold text-slate-800 mb-3">Test Scenarios</h3>
<div class="grid md:grid-cols-3 gap-4 text-slate-500">
<div class="bg-slate-50 border border-slate-200 rounded-xl p-4">
<div class="font-semibold text-slate-800">GNSS Multipath</div>
<div class="text-sm mt-1">Downtown / tree cover → reliability ML inflates GNSS noise.</div>
</div>
<div class="bg-slate-50 border border-slate-200 rounded-xl p-4">
<div class="font-semibold text-slate-800">Wheel Slip</div>
<div class="text-sm mt-1">Hard accel / wet surface → slip detector down-weights wheel updates.</div>
</div>
<div class="bg-slate-50 border border-slate-200 rounded-xl p-4">
<div class="font-semibold text-slate-800">Low Texture / Blur</div>
<div class="text-sm mt-1">Stereo VO baseline fails → CNN matching stabilizes correspondences.</div>
</div>
</div>
</div>
</section>
<!-- Footer -->
<footer class="border-t border-slate-100 py-10">
<div class="max-w-6xl mx-auto px-4 text-slate-400 text-sm flex flex-col md:flex-row gap-2 md:items-center md:justify-between">
<div>© Ashrith Edukulla · Full-Stack AV Localization (IEKF + Deep Learning)</div>
<div class="text-slate-400">Built with Tailwind CSS</div>
</div>
</footer>
</body>
</html>