Robot Calibration Service
We Make
Industrial Robots
Accurate.
Industrial robots are built for repeatability — not accuracy. The difference between where a robot thinks it is and where it actually is can be 2 to 5 millimetres or more, straight out of the box. For precision applications, that's unacceptable.
The Problem & The Fix
Precision Metrology.
Controller-Level Results.
MINOTOR Labs offers a kinematic calibration service that closes the accuracy gap. We measure your robot's true behaviour using precision metrology — laser trackers and photogrammetric systems — then identify the real geometric parameters of your specific machine.
Every robot is slightly different due to manufacturing tolerances, assembly deviations, and wear. Our software builds a corrected kinematic model unique to your robot — not the nominal model that shipped from the factory.
The output is loaded directly into the robot controller as a precision file. No external compensation layers. No runtime corrections. No additional hardware. The robot simply operates from a better model of itself.
◈
Sub-millimeter absolute positioning
Positional accuracy after calibration that far exceeds factory specifications.
◈
No hardware modifications
Pure model correction — the calibration lives inside the robot controller.
◈
Proprietary software platform
Developed entirely in-house — purpose-built for this problem, not a generic tool.
◈
On-site service
We bring the measurement equipment to your facility. Minimal downtime.
Our Process
Four Stages.
One Precise Robot.
01
Measure
We deploy precision metrology equipment — laser trackers and photogrammetric systems — on-site at your facility. The robot is moved through a carefully designed set of configurations to capture its true end-effector pose at each position.
02
Analyse
Our software analyses the quality of the measurement dataset — checking joint configuration coverage, identifying underexcited axes, detecting redundant poses, and confirming the data is sufficient to reliably identify all kinematic parameters.
03
Optimise
A dual-stage calibration pipeline processes the data. The first stage establishes the robot's real kinematic geometry — link lengths, offsets, and angular deviations. The second stage refines the tool centre point using probe-sphere contact measurements.
04
Deploy
The corrected kinematic model is loaded directly into your KUKA robot controller as a
.prec file. No external software. No middleware. No application-level corrections. Your robot now operates from a refined absolute model.
Use Cases
Built for Any Application
Where Accuracy Matters.
⬡
Milling
Precision material removal where positional drift compounds across a toolpath.
⬡
Drilling
Hole placement accuracy in aerospace, automotive, and structural components.
⬡
Trimming
Consistent edge and contour trimming on complex composite or sheet-metal parts.
⬡
Automated Inspection
Sensor or probe positioning that demands repeatable absolute coordinates.
⬡
Additive Manufacturing
Deposition accuracy across large workspaces where nominal models fall short.
⬡
Assembly
High-tolerance part mating and fastening in automotive and electronics production.
Good
This tab compares the calibrated result against the nominal root on the same final-stage filtered point set. Positive percentages mean the error became smaller.
Nominal mean
0.478
mm before calibration
Calibrated mean
0.193
mm after calibration
Mean improvement
+59.7%
position error reduction
Max improvement
+61.1%
worst-case reduction
| Baseline | Before photogrammetry_full (after photogrammetry_init) |
| Final | Calibrated result |
| Stage | photogrammetry_full |
| Compared samples | 78 |
| Rotation weight | 0.0000 |
| Metric | Nominal root | Calibrated | Delta | Improvement |
|---|---|---|---|---|
| Mean position error | 0.4782 mm | 0.1929 mm | 0.2853 mm | +59.7% |
| Max position error | 1.1318 mm | 0.4401 mm | 0.6916 mm | +61.1% |
| Std position error | 0.2022 mm | 0.0863 mm | 0.1159 mm | +57.3% |
| Run ID | 20260505_164735_fb4832 |
| Timestamp | 2026-05-05T16:47:58 |
| Scope | Final (photogrammetry_full) |
| Method | Jacobian SVD correlation |
| Bad |r| threshold | 0.900 |
B (77/100) Good - minor correlations, acceptable for calibration
| Link | Parameter | link1/offset_rx | link1/offset_ry | link1/offset_rz | link2/offset_rx | link2/offset_ry | link2/offset_rz | link3/offset_x | link3/offset_rx | link3/offset_ry | link3/offset_rz | link4/offset_rx | link4/offset_ry | link4/offset_rz | link5/offset_x | link5/offset_z | link5/offset_ry | link5/offset_rz | link6/offset_x | link6/offset_z | link6/offset_rx | link6/offset_ry | link6/offset_rz |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| link1 | link1/offset_rx | 1.000 | -0.221 | -0.018 | -0.068 | -0.040 | 0.486 | 0.224 | 0.031 | -0.004 | -0.075 | 0.121 | 0.085 | 0.132 | 0.268 | -0.016 | 0.093 | 0.018 | -0.131 | -0.025 | 0.007 | 0.032 | 0.259 |
| link1 | link1/offset_ry | -0.221 | 1.000 | 0.039 | -0.288 | 0.034 | -0.526 | -0.061 | -0.027 | 0.046 | -0.078 | 0.026 | 0.081 | 0.249 | 0.089 | -0.097 | 0.129 | 0.081 | 0.261 | -0.008 | 0.093 | 0.105 | 0.203 |
| link1 | link1/offset_rz | -0.018 | 0.039 | 1.000 | 0.671 | 0.992 | 0.310 | 0.374 | -0.234 | -0.875 | -0.439 | 0.765 | 0.052 | -0.040 | -0.066 | -0.484 | -0.011 | 0.120 | -0.054 | -0.039 | 0.047 | 0.114 | 0.076 |
| link2 | link2/offset_rx | -0.068 | -0.288 | 0.671 | 1.000 | 0.710 | 0.058 | 0.218 | 0.370 | -0.782 | -0.160 | 0.248 | -0.112 | -0.103 | 0.059 | -0.037 | -0.091 | -0.003 | -0.226 | 0.053 | -0.085 | -0.007 | 0.050 |
| link2 | link2/offset_ry | -0.040 | 0.034 | 0.992 | 0.710 | 1.000 | 0.265 | 0.378 | -0.128 | -0.899 | -0.410 | 0.730 | 0.039 | -0.050 | -0.058 | -0.440 | -0.003 | 0.135 | -0.046 | -0.046 | 0.059 | 0.130 | 0.064 |
| link2 | link2/offset_rz | 0.486 | -0.526 | 0.310 | 0.058 | 0.265 | 1.000 | 0.239 | -0.477 | -0.189 | -0.456 | 0.509 | 0.075 | -0.097 | -0.025 | -0.640 | -0.002 | -0.032 | -0.069 | 0.020 | -0.024 | -0.036 | -0.015 |
| link3 | link3/offset_x | 0.224 | -0.061 | 0.374 | 0.218 | 0.378 | 0.239 | 1.000 | -0.057 | -0.467 | -0.634 | 0.416 | -0.131 | 0.094 | -0.012 | -0.401 | 0.038 | 0.180 | 0.078 | -0.062 | 0.085 | 0.174 | 0.154 |
| link3 | link3/offset_rx | 0.031 | -0.027 | -0.234 | 0.370 | -0.128 | -0.477 | -0.057 | 1.000 | -0.087 | 0.352 | -0.576 | -0.104 | 0.006 | 0.187 | 0.567 | -0.013 | 0.047 | -0.100 | -0.010 | -0.019 | 0.053 | 0.034 |
| link3 | link3/offset_ry | -0.004 | 0.046 | -0.875 | -0.782 | -0.899 | -0.189 | -0.467 | -0.087 | 1.000 | 0.358 | -0.605 | 0.030 | -0.077 | 0.026 | 0.344 | 0.021 | -0.147 | 0.060 | 0.039 | -0.057 | -0.143 | -0.030 |
| link3 | link3/offset_rz | -0.075 | -0.078 | -0.439 | -0.160 | -0.410 | -0.456 | -0.634 | 0.352 | 0.358 | 1.000 | -0.553 | 0.462 | -0.298 | -0.276 | 0.731 | -0.034 | 0.064 | 0.058 | -0.194 | 0.159 | 0.037 | -0.094 |
| link4 | link4/offset_rx | 0.121 | 0.026 | 0.765 | 0.248 | 0.730 | 0.509 | 0.416 | -0.576 | -0.605 | -0.553 | 1.000 | 0.014 | 0.053 | -0.003 | -0.721 | 0.190 | 0.087 | 0.049 | -0.019 | 0.085 | 0.100 | 0.082 |
| link4 | link4/offset_ry | 0.085 | 0.081 | 0.052 | -0.112 | 0.039 | 0.075 | -0.131 | -0.104 | 0.030 | 0.462 | 0.014 | 1.000 | -0.177 | -0.414 | -0.023 | -0.080 | 0.176 | 0.196 | -0.247 | 0.230 | 0.116 | 0.354 |
| link4 | link4/offset_rz | 0.132 | 0.249 | -0.040 | -0.103 | -0.050 | -0.097 | 0.094 | 0.006 | -0.077 | -0.298 | 0.053 | -0.177 | 1.000 | 0.250 | -0.019 | 0.146 | -0.159 | -0.056 | 0.144 | -0.100 | -0.123 | 0.444 |
| link5 | link5/offset_x | 0.268 | 0.089 | -0.066 | 0.059 | -0.058 | -0.025 | -0.012 | 0.187 | 0.026 | -0.276 | -0.003 | -0.414 | 0.250 | 1.000 | 0.015 | 0.041 | 0.103 | -0.239 | 0.263 | -0.280 | 0.159 | 0.041 |
| link5 | link5/offset_z | -0.016 | -0.097 | -0.484 | -0.037 | -0.440 | -0.640 | -0.401 | 0.567 | 0.344 | 0.731 | -0.721 | -0.023 | -0.019 | 0.015 | 1.000 | -0.003 | -0.055 | -0.100 | -0.008 | -0.032 | -0.054 | -0.021 |
| link5 | link5/offset_ry | 0.093 | 0.129 | -0.011 | -0.091 | -0.003 | -0.002 | 0.038 | -0.013 | 0.021 | -0.034 | 0.190 | -0.080 | 0.146 | 0.041 | -0.003 | 1.000 | -0.049 | 0.085 | 0.115 | 0.105 | 0.002 | 0.125 |
| link5 | link5/offset_rz | 0.018 | 0.081 | 0.120 | -0.003 | 0.135 | -0.032 | 0.180 | 0.047 | -0.147 | 0.064 | 0.087 | 0.176 | -0.159 | 0.103 | -0.055 | -0.049 | 1.000 | 0.552 | -0.530 | 0.541 | 0.990 | -0.166 |
| link6 | link6/offset_x | -0.131 | 0.261 | -0.054 | -0.226 | -0.046 | -0.069 | 0.078 | -0.100 | 0.060 | 0.058 | 0.049 | 0.196 | -0.056 | -0.239 | -0.100 | 0.085 | 0.552 | 1.000 | -0.686 | 0.807 | 0.575 | -0.014 |
| link6 | link6/offset_z | -0.025 | -0.008 | -0.039 | 0.053 | -0.046 | 0.020 | -0.062 | -0.010 | 0.039 | -0.194 | -0.019 | -0.247 | 0.144 | 0.263 | -0.008 | 0.115 | -0.530 | -0.686 | 1.000 | -0.925 | -0.527 | -0.029 |
| link6 | link6/offset_rx | 0.007 | 0.093 | 0.047 | -0.085 | 0.059 | -0.024 | 0.085 | -0.019 | -0.057 | 0.159 | 0.085 | 0.230 | -0.100 | -0.280 | -0.032 | 0.105 | 0.541 | 0.807 | -0.925 | 1.000 | 0.548 | -0.021 |
| link6 | link6/offset_ry | 0.032 | 0.105 | 0.114 | -0.007 | 0.130 | -0.036 | 0.174 | 0.053 | -0.143 | 0.037 | 0.100 | 0.116 | -0.123 | 0.159 | -0.054 | 0.002 | 0.990 | 0.575 | -0.527 | 0.548 | 1.000 | -0.134 |
| link6 | link6/offset_rz | 0.259 | 0.203 | 0.076 | 0.050 | 0.064 | -0.015 | 0.154 | 0.034 | -0.030 | -0.094 | 0.082 | 0.354 | 0.444 | 0.041 | -0.021 | 0.125 | -0.166 | -0.014 | -0.029 | -0.021 | -0.134 | 1.000 |
Jacobian SVD correlation Analytics
| Total parameters | 22 |
| Unique parameter pairs | 231 |
| Bad pairs (|r| ≥ 0.900) | 3 |
| Warning pairs (0.750 ≤ |r| < 0.900) | 5 |
| Average |r| (off-diagonal) | 0.193 |
| Max |r| pair | link1/offset_rz ↔ link2/offset_ry (0.992) |
Correlation validation (SVD covariance paths)
| Max abs diff | 2.2204e-16 |
| Mean abs diff | 1.7360e-17 |
| Off-diag max abs diff | 2.2204e-16 |
| Off-diag mean abs diff | 1.8186e-17 |
| Run ID | 20260505_164735_fb4832 |
| Timestamp | 2026-05-05T16:47:58 |
| Scope | photogrammetry_init |
| Method | Jacobian SVD correlation (photogrammetry_init) |
| Bad |r| threshold | 0.900 |
A (92/100) Excellent - parameters are well-identifiable, dataset is sufficient
| Link | Parameter | base/x | base/y | base/z | base/rx | base/ry | base/rz | camera/x | camera/y | camera/z | camera/rx | camera/ry | camera/rz |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| base | base/x | 1.000 | 0.219 | -0.635 | -0.129 | -0.700 | -0.334 | -0.640 | -0.103 | -0.634 | 0.000 | 0.000 | 0.000 |
| base | base/y | 0.219 | 1.000 | 0.403 | 0.749 | -0.018 | -0.586 | -0.071 | 0.483 | -0.071 | 0.000 | 0.000 | 0.000 |
| base | base/z | -0.635 | 0.403 | 1.000 | 0.615 | 0.791 | 0.052 | 0.128 | 0.334 | 0.145 | 0.000 | 0.000 | 0.000 |
| base | base/rx | -0.129 | 0.749 | 0.615 | 1.000 | 0.066 | 0.051 | 0.041 | 0.529 | 0.011 | 0.000 | 0.000 | 0.000 |
| base | base/ry | -0.700 | -0.018 | 0.791 | 0.066 | 1.000 | 0.030 | 0.068 | 0.065 | 0.109 | 0.000 | 0.000 | 0.000 |
| base | base/rz | -0.334 | -0.586 | 0.052 | 0.051 | 0.030 | 1.000 | -0.038 | -0.119 | -0.056 | 0.000 | 0.000 | 0.000 |
| camera | camera/x | -0.640 | -0.071 | 0.128 | 0.041 | 0.068 | -0.038 | 1.000 | 0.114 | 0.862 | 0.000 | 0.000 | 0.000 |
| camera | camera/y | -0.103 | 0.483 | 0.334 | 0.529 | 0.065 | -0.119 | 0.114 | 1.000 | 0.098 | 0.000 | 0.000 | 0.000 |
| camera | camera/z | -0.634 | -0.071 | 0.145 | 0.011 | 0.109 | -0.056 | 0.862 | 0.098 | 1.000 | 0.000 | 0.000 | 0.000 |
| camera | camera/rx | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 | 0.000 | 0.000 |
| camera | camera/ry | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 | 0.000 |
| camera | camera/rz | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 |
Jacobian SVD correlation (photogrammetry_init) Analytics
| Total parameters | 12 |
| Unique parameter pairs | 66 |
| Bad pairs (|r| ≥ 0.900) | 0 |
| Warning pairs (0.750 ≤ |r| < 0.900) | 2 |
| Average |r| (off-diagonal) | 0.153 |
| Max |r| pair | camera/x ↔ camera/z (0.862) |
| Run ID | 20260505_164735_fb4832 |
| Timestamp | 2026-05-05T16:47:58 |
| Scope | photogrammetry_full |
| Method | Jacobian SVD correlation (photogrammetry_full) |
| Bad |r| threshold | 0.900 |
B (77/100) Good - minor correlations, acceptable for calibration
| Link | Parameter | link1/offset_rx | link1/offset_ry | link1/offset_rz | link2/offset_rx | link2/offset_ry | link2/offset_rz | link3/offset_x | link3/offset_rx | link3/offset_ry | link3/offset_rz | link4/offset_rx | link4/offset_ry | link4/offset_rz | link5/offset_x | link5/offset_z | link5/offset_ry | link5/offset_rz | link6/offset_x | link6/offset_z | link6/offset_rx | link6/offset_ry | link6/offset_rz |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| link1 | link1/offset_rx | 1.000 | -0.221 | -0.018 | -0.068 | -0.040 | 0.486 | 0.224 | 0.031 | -0.004 | -0.075 | 0.121 | 0.085 | 0.132 | 0.268 | -0.016 | 0.093 | 0.018 | -0.131 | -0.025 | 0.007 | 0.032 | 0.259 |
| link1 | link1/offset_ry | -0.221 | 1.000 | 0.039 | -0.288 | 0.034 | -0.526 | -0.061 | -0.027 | 0.046 | -0.078 | 0.026 | 0.081 | 0.249 | 0.089 | -0.097 | 0.129 | 0.081 | 0.261 | -0.008 | 0.093 | 0.105 | 0.203 |
| link1 | link1/offset_rz | -0.018 | 0.039 | 1.000 | 0.671 | 0.992 | 0.310 | 0.374 | -0.234 | -0.875 | -0.439 | 0.765 | 0.052 | -0.040 | -0.066 | -0.484 | -0.011 | 0.120 | -0.054 | -0.039 | 0.047 | 0.114 | 0.076 |
| link2 | link2/offset_rx | -0.068 | -0.288 | 0.671 | 1.000 | 0.710 | 0.058 | 0.218 | 0.370 | -0.782 | -0.160 | 0.248 | -0.112 | -0.103 | 0.059 | -0.037 | -0.091 | -0.003 | -0.226 | 0.053 | -0.085 | -0.007 | 0.050 |
| link2 | link2/offset_ry | -0.040 | 0.034 | 0.992 | 0.710 | 1.000 | 0.265 | 0.378 | -0.128 | -0.899 | -0.410 | 0.730 | 0.039 | -0.050 | -0.058 | -0.440 | -0.003 | 0.135 | -0.046 | -0.046 | 0.059 | 0.130 | 0.064 |
| link2 | link2/offset_rz | 0.486 | -0.526 | 0.310 | 0.058 | 0.265 | 1.000 | 0.239 | -0.477 | -0.189 | -0.456 | 0.509 | 0.075 | -0.097 | -0.025 | -0.640 | -0.002 | -0.032 | -0.069 | 0.020 | -0.024 | -0.036 | -0.015 |
| link3 | link3/offset_x | 0.224 | -0.061 | 0.374 | 0.218 | 0.378 | 0.239 | 1.000 | -0.057 | -0.467 | -0.634 | 0.416 | -0.131 | 0.094 | -0.012 | -0.401 | 0.038 | 0.180 | 0.078 | -0.062 | 0.085 | 0.174 | 0.154 |
| link3 | link3/offset_rx | 0.031 | -0.027 | -0.234 | 0.370 | -0.128 | -0.477 | -0.057 | 1.000 | -0.087 | 0.352 | -0.576 | -0.104 | 0.006 | 0.187 | 0.567 | -0.013 | 0.047 | -0.100 | -0.010 | -0.019 | 0.053 | 0.034 |
| link3 | link3/offset_ry | -0.004 | 0.046 | -0.875 | -0.782 | -0.899 | -0.189 | -0.467 | -0.087 | 1.000 | 0.358 | -0.605 | 0.030 | -0.077 | 0.026 | 0.344 | 0.021 | -0.147 | 0.060 | 0.039 | -0.057 | -0.143 | -0.030 |
| link3 | link3/offset_rz | -0.075 | -0.078 | -0.439 | -0.160 | -0.410 | -0.456 | -0.634 | 0.352 | 0.358 | 1.000 | -0.553 | 0.462 | -0.298 | -0.276 | 0.731 | -0.034 | 0.064 | 0.058 | -0.194 | 0.159 | 0.037 | -0.094 |
| link4 | link4/offset_rx | 0.121 | 0.026 | 0.765 | 0.248 | 0.730 | 0.509 | 0.416 | -0.576 | -0.605 | -0.553 | 1.000 | 0.014 | 0.053 | -0.003 | -0.721 | 0.190 | 0.087 | 0.049 | -0.019 | 0.085 | 0.100 | 0.082 |
| link4 | link4/offset_ry | 0.085 | 0.081 | 0.052 | -0.112 | 0.039 | 0.075 | -0.131 | -0.104 | 0.030 | 0.462 | 0.014 | 1.000 | -0.177 | -0.414 | -0.023 | -0.080 | 0.176 | 0.196 | -0.247 | 0.230 | 0.116 | 0.354 |
| link4 | link4/offset_rz | 0.132 | 0.249 | -0.040 | -0.103 | -0.050 | -0.097 | 0.094 | 0.006 | -0.077 | -0.298 | 0.053 | -0.177 | 1.000 | 0.250 | -0.019 | 0.146 | -0.159 | -0.056 | 0.144 | -0.100 | -0.123 | 0.444 |
| link5 | link5/offset_x | 0.268 | 0.089 | -0.066 | 0.059 | -0.058 | -0.025 | -0.012 | 0.187 | 0.026 | -0.276 | -0.003 | -0.414 | 0.250 | 1.000 | 0.015 | 0.041 | 0.103 | -0.239 | 0.263 | -0.280 | 0.159 | 0.041 |
| link5 | link5/offset_z | -0.016 | -0.097 | -0.484 | -0.037 | -0.440 | -0.640 | -0.401 | 0.567 | 0.344 | 0.731 | -0.721 | -0.023 | -0.019 | 0.015 | 1.000 | -0.003 | -0.055 | -0.100 | -0.008 | -0.032 | -0.054 | -0.021 |
| link5 | link5/offset_ry | 0.093 | 0.129 | -0.011 | -0.091 | -0.003 | -0.002 | 0.038 | -0.013 | 0.021 | -0.034 | 0.190 | -0.080 | 0.146 | 0.041 | -0.003 | 1.000 | -0.049 | 0.085 | 0.115 | 0.105 | 0.002 | 0.125 |
| link5 | link5/offset_rz | 0.018 | 0.081 | 0.120 | -0.003 | 0.135 | -0.032 | 0.180 | 0.047 | -0.147 | 0.064 | 0.087 | 0.176 | -0.159 | 0.103 | -0.055 | -0.049 | 1.000 | 0.552 | -0.530 | 0.541 | 0.990 | -0.166 |
| link6 | link6/offset_x | -0.131 | 0.261 | -0.054 | -0.226 | -0.046 | -0.069 | 0.078 | -0.100 | 0.060 | 0.058 | 0.049 | 0.196 | -0.056 | -0.239 | -0.100 | 0.085 | 0.552 | 1.000 | -0.686 | 0.807 | 0.575 | -0.014 |
| link6 | link6/offset_z | -0.025 | -0.008 | -0.039 | 0.053 | -0.046 | 0.020 | -0.062 | -0.010 | 0.039 | -0.194 | -0.019 | -0.247 | 0.144 | 0.263 | -0.008 | 0.115 | -0.530 | -0.686 | 1.000 | -0.925 | -0.527 | -0.029 |
| link6 | link6/offset_rx | 0.007 | 0.093 | 0.047 | -0.085 | 0.059 | -0.024 | 0.085 | -0.019 | -0.057 | 0.159 | 0.085 | 0.230 | -0.100 | -0.280 | -0.032 | 0.105 | 0.541 | 0.807 | -0.925 | 1.000 | 0.548 | -0.021 |
| link6 | link6/offset_ry | 0.032 | 0.105 | 0.114 | -0.007 | 0.130 | -0.036 | 0.174 | 0.053 | -0.143 | 0.037 | 0.100 | 0.116 | -0.123 | 0.159 | -0.054 | 0.002 | 0.990 | 0.575 | -0.527 | 0.548 | 1.000 | -0.134 |
| link6 | link6/offset_rz | 0.259 | 0.203 | 0.076 | 0.050 | 0.064 | -0.015 | 0.154 | 0.034 | -0.030 | -0.094 | 0.082 | 0.354 | 0.444 | 0.041 | -0.021 | 0.125 | -0.166 | -0.014 | -0.029 | -0.021 | -0.134 | 1.000 |
Jacobian SVD correlation (photogrammetry_full) Analytics
| Total parameters | 22 |
| Unique parameter pairs | 231 |
| Bad pairs (|r| ≥ 0.900) | 3 |
| Warning pairs (0.750 ≤ |r| < 0.900) | 5 |
| Average |r| (off-diagonal) | 0.193 |
| Max |r| pair | link1/offset_rz ↔ link2/offset_ry (0.992) |
📖 How to interpret this tab
| Metric | How to understand | Good sign | Warning sign | What to do |
|---|---|---|---|---|
| Condition number (J) | Numerical stability of parameter estimation. Large value means tiny data noise can strongly change parameters. | Low to moderate condition number. | Very high condition number. | Add orthogonal poses that excite different parameter directions, not repeats. |
| VIF / VIF status | Collinearity indicator for each parameter. High VIF means a parameter is explained by others. | Most parameters in OK range. | Many WARNING/CRITICAL parameters. | Collect poses where correlated parameters affect measurements differently. |
| Max |corr| / worst pair | Strong linear dependency between parameter sensitivities in Jacobian. | Low-to-moderate correlations for most parameters. | Correlations near threshold for many pairs. | Design measurements that change one member of the pair while keeping the other nearly fixed. |
| Uncertainty (σ) | Estimated standard deviation of parameter (from Jacobian-based uncertainty). | Small and balanced values. | Very large values for specific parameters. | Target those parameters with dedicated pose families. |
B77 / 100
Good - minor correlations, acceptable for calibration
Health score weighting details
Config: C:\NeoSetDesign\nsd_monocular_calibration_app_py\nsd-monocular-calibration-app\health_score_config.yaml
Overall = correlation*0.90 + points_coverage*0.10
Correlation = median|r|*0.40 + bad_pairs*0.30 + vif*0.20 + rank*0.05 + cond*0.05
Correlation scoring = robust median|r|, near-bad penalty=6.0, VIF grace=5.0, cond grace=1e2.0
Points coverage thresholds ratio(samples/params): low=3.0, mid=6.0, good=10.0, full=12.0
Overall = correlation*0.90 + points_coverage*0.10
Correlation = median|r|*0.40 + bad_pairs*0.30 + vif*0.20 + rank*0.05 + cond*0.05
Correlation scoring = robust median|r|, near-bad penalty=6.0, VIF grace=5.0, cond grace=1e2.0
Points coverage thresholds ratio(samples/params): low=3.0, mid=6.0, good=10.0, full=12.0
High VIF values are normal and expected when optimizing many parameters (20+ DOF). Your Health Score already accounts for this. High VIF does NOT mean the calibration is unreliable — it means those parameters are statistically correlated. What matters: can your measurement data distinguish them? The Health Score answers this question.
Global Metrics
| Jacobian size | 468 observations × 22 parameters |
| Condition number (J) | 16422.76 |
| Log₁₀(condition) | 4.22 |
| Max singular value | 2.6029e+04 |
| Min singular value | 1.5849e+00 |
| Singular value ratio (min/max) | 6.089112e-05 |
| Max VIF | 3289.33 |
| Mean VIF | 318.54 |
| Parameters with VIF > 1000 | 2 |
| Parameters with VIF > 100 | 5 |
Per-Parameter Analysis
| Link | Parameter | VIF | VIF status | Uncertainty (σ) | Max |corr| | Worst correlated pair | Identifiability |
|---|---|---|---|---|---|---|---|
| link1 | link1/offset_rx | 25.65 | OK | 3.6718e-04 | 0.4856 | link2/offset_rz | GOOD |
| link1 | link1/offset_ry | 38.67 | OK | 3.7929e-04 | 0.5262 | link2/offset_rz | GOOD |
| link1 | link1/offset_rz | 3289.33 | CRITICAL | 4.0774e-03 | 0.9916 | link2/offset_ry | CRITICAL: VIF critical; high corr |
| link2 | link2/offset_rx | 52.58 | OK | 7.7865e-04 | 0.7817 | link3/offset_ry | GOOD |
| link2 | link2/offset_ry | 2798.46 | CRITICAL | 4.7097e-03 | 0.9916 | link1/offset_rz | CRITICAL: VIF critical; high corr |
| link2 | link2/offset_rz | 73.30 | OK | 6.4514e-04 | 0.6402 | link5/offset_z | GOOD |
| link3 | link3/offset_x | 18.26 | OK | 4.8388e-01 | 0.6336 | link3/offset_rz | GOOD |
| link3 | link3/offset_rx | 149.15 | WARNING | 1.0551e-03 | 0.5761 | link4/offset_rx | NOTE: VIF elevated |
| link3 | link3/offset_ry | 27.10 | OK | 8.0322e-04 | 0.8993 | link2/offset_ry | GOOD |
| link3 | link3/offset_rz | 54.60 | OK | 6.2612e-04 | 0.7312 | link5/offset_z | GOOD |
| link4 | link4/offset_rx | 38.86 | OK | 1.0721e-03 | 0.7647 | link1/offset_rz | GOOD |
| link4 | link4/offset_ry | 14.14 | OK | 7.4900e-04 | 0.4615 | link3/offset_rz | GOOD |
| link4 | link4/offset_rz | 4.52 | OK | 4.0113e-04 | 0.4440 | link6/offset_rz | GOOD |
| link5 | link5/offset_x | 3.79 | OK | 2.2031e-01 | 0.4141 | link4/offset_ry | GOOD |
| link5 | link5/offset_z | 24.80 | OK | 5.6383e-01 | 0.7312 | link3/offset_rz | GOOD |
| link5 | link5/offset_ry | 3.07 | OK | 4.7871e-04 | 0.1901 | link4/offset_rx | GOOD |
| link5 | link5/offset_rz | 164.04 | WARNING | 2.1359e-03 | 0.9898 | link6/offset_ry | WARNING: VIF elevated; high corr |
| link6 | link6/offset_x | 6.95 | OK | 2.9845e-01 | 0.8073 | link6/offset_rx | GOOD |
| link6 | link6/offset_z | 22.81 | OK | 5.4073e-01 | 0.9253 | link6/offset_rx | WARNING: high corr |
| link6 | link6/offset_rx | 26.40 | OK | 1.2688e-03 | 0.9253 | link6/offset_z | WARNING: high corr |
| link6 | link6/offset_ry | 166.44 | WARNING | 3.2402e-03 | 0.9898 | link5/offset_rz | WARNING: VIF elevated; high corr |
| link6 | link6/offset_rz | 4.97 | OK | 9.9608e-04 | 0.4440 | link4/offset_rz | GOOD |
| Condition number | 1.64e+04 |
| QR effective rank | 22 / 22 |
| Observable | 17 |
| Borderline | 5 |
| Unobservable | 0 |
Method: per-parameter observability via SVD projection + QR column pivoting for identifiability ranking
| # | Link | Parameter | Sensitivity (||J col||) | Observability Index | QR Identifiable | Classification | VIF | Recommendation |
|---|---|---|---|---|---|---|---|---|
| 1 | link3 | link3/offset_x | 8.8318 | 0.0005 | YES | Borderline | 18.26 | Consider regularisation or additional poses |
| 2 | link5 | link5/offset_x | 8.8318 | 0.0005 | YES | Borderline | 3.79 | Consider regularisation or additional poses |
| 3 | link6 | link6/offset_x | 8.8318 | 0.0005 | YES | Borderline | 6.95 | Consider regularisation or additional poses |
| 4 | link6 | link6/offset_z | 8.8318 | 0.0005 | YES | Borderline | 22.81 | Consider regularisation or additional poses |
| 5 | link5 | link5/offset_z | 8.8318 | 0.0005 | YES | Borderline | 24.80 | Consider regularisation or additional poses |
| 6 | link6 | link6/offset_rz | 2238.1583 | 0.1365 | YES | Observable | 4.97 | |
| 7 | link5 | link5/offset_ry | 3659.7336 | 0.2232 | YES | Observable | 3.07 | |
| 8 | link6 | link6/offset_ry | 3981.5408 | 0.2428 | YES | Observable | 166.44 | |
| 9 | link6 | link6/offset_rx | 4049.9659 | 0.2470 | YES | Observable | 26.40 | |
| 10 | link4 | link4/offset_ry | 5019.8035 | 0.3062 | YES | Observable | 14.14 | |
| 11 | link4 | link4/offset_rz | 5300.2646 | 0.3233 | YES | Observable | 4.52 | |
| 12 | link4 | link4/offset_rx | 5814.4170 | 0.3546 | YES | Observable | 38.86 | |
| 13 | link5 | link5/offset_rz | 5996.5206 | 0.3657 | YES | Observable | 164.04 | |
| 14 | link3 | link3/offset_ry | 6481.1763 | 0.3953 | YES | Observable | 27.10 | |
| 15 | link2 | link2/offset_rx | 9312.6586 | 0.5680 | YES | Observable | 52.58 | |
| 16 | link2 | link2/offset_ry | 11232.2993 | 0.6851 | YES | Observable | 2798.46 | Very high VIF — strongly correlated with other params |
| 17 | link3 | link3/offset_rx | 11574.4279 | 0.7060 | YES | Observable | 149.15 | |
| 18 | link3 | link3/offset_rz | 11800.9876 | 0.7198 | YES | Observable | 54.60 | |
| 19 | link2 | link2/offset_rz | 13270.4860 | 0.8094 | YES | Observable | 73.30 | |
| 20 | link1 | link1/offset_rx | 13793.3456 | 0.8413 | YES | Observable | 25.65 | |
| 21 | link1 | link1/offset_rz | 14065.8960 | 0.8579 | YES | Observable | 3289.33 | Very high VIF — strongly correlated with other params |
| 22 | link1 | link1/offset_ry | 16395.4675 | 1.0000 | YES | Observable | 38.67 |
Recommendations
• 5 parameter(s) are borderline — fix the weakest ones or add more informative measurement poses.
📖 How to interpret this tab
| Metric | How to understand | Good sign | Warning sign | What to do |
|---|---|---|---|---|
| Axis span / bins / coverage | Shows whether measurements really cover the working range instead of repeating the same zone. | Most axes use all bins and coverage is close to 100%. | Some axes use few bins or have narrow span. | Add measurements at the missing ends of those axes; avoid adding near existing clusters. |
| Required Actions (Prioritized) | Action list sorted by impact: P0 is blocking identifiability, P3 is optimization/cleanup. | Only P2/P3 items remain. | Any P0/P1 item is present. | Execute P0 first, then P1; only then spend time on P2/P3. |
| Data dir | C:\NeoSetDesign\nsd_monocular_calibration_app_py\projects\test11_smallarea\data\photogrammetry |
| Positions source | MeasuredPositions.csv (standard) |
| Joints source | Joints.csv |
| Discovered formats | standard=True, camera_report=False |
| Joint samples | 78 |
| Positions samples | 78 |
| Matched IDs | 78 |
| Missing in positions | none |
| Missing in joints | none |
Joint-space Coverage (A1..A6)
| Axis | Min | Max | Span | Mean | Std | P05 | P95 |
|---|---|---|---|---|---|---|---|
| A1 | 16.937 | 76.987 | 60.050 | 49.717 | 13.599 | 30.463 | 70.869 |
| A2 | -133.658 | -58.671 | 74.987 | -95.972 | 14.713 | -121.532 | -71.667 |
| A3 | 64.221 | 149.897 | 85.675 | 103.980 | 18.763 | 73.282 | 132.361 |
| A4 | -115.918 | 133.689 | 249.607 | 11.178 | 57.989 | -75.697 | 115.945 |
| A5 | -114.162 | 111.502 | 225.664 | -21.250 | 76.256 | -111.850 | 95.254 |
| A6 | -151.158 | 59.998 | 211.156 | -56.456 | 52.774 | -129.759 | 28.750 |
Position Coverage (X, Y, Z)
| Axis | Min | Max | Span | Mean | Std | P05 | P95 |
|---|---|---|---|---|---|---|---|
| X | -3929.418 | -3003.390 | 926.029 | -3481.374 | 275.151 | -3864.725 | -3048.439 |
| Y | -1578.496 | -343.257 | 1235.239 | -830.142 | 266.445 | -1370.256 | -416.886 |
| Z | -345.327 | 829.496 | 1174.823 | 232.258 | 302.970 | -247.066 | 733.440 |
Orientation Coverage (A, B, C / Az, El, Roll)
| Axis | Min | Max | Span | Mean | Std | P05 | P95 |
|---|---|---|---|---|---|---|---|
| A | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| B | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| C | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
Additional Geometry Metrics
| Joint NN (p50) | 31.450 |
| Camera NN (p50) | 115.011 |
| Radial std | 137.768 |
| View azimuth span | -176.09 .. 176.26 |
| View azimuth width | 352.35 |
| View elevation span | -80.29 .. 74.60 |
| View elevation width | 154.89 |
| Near-X alignment fraction | 0.13 |
Recommendations
• Elevation diversity is limited; add captures from lower and higher viewpoints.
• Camera distance diversity is narrow; add both near and far camera positions.
Required Actions (Prioritized)
| Priority | Metric | Current | Target | Required Action | Why |
|---|---|---|---|---|---|
| P1 | Elevation std | 0.00 deg | >= 15 deg | Add captures from lower and higher viewpoints | Low elevation diversity increases coupling between camera and robot parameters |
| P1 | Radial distance std | 137.77 | >= 200 | Add both near and far camera positions | Narrow depth variation weakens separation of translational parameters |
Coverage of robot joint configurations with focus on wrist axes A4/A5/A6
Joint Range Summary
| Axis | Min | Max | Span | Mean | Std |
|---|---|---|---|---|---|
| A1 | 16.937 | 76.987 | 60.050 | 49.717 | 13.599 |
| A2 | -133.658 | -58.671 | 74.987 | -95.972 | 14.713 |
| A3 | 64.221 | 149.897 | 85.675 | 103.980 | 18.763 |
| A4 | -115.918 | 133.689 | 249.607 | 11.178 | 57.989 |
| A5 | -114.162 | 111.502 | 225.664 | -21.250 | 76.256 |
| A6 | -151.158 | 59.998 | 211.156 | -56.456 | 52.774 |
Coarse Arm Configurations (A1/A2/A3 signs)
| Configuration | Count | Share % | Interpretation |
|---|---|---|---|
| A1+ A2- A3+ | 78 | 100.0% | Dominant family |
Wrist Sign Combinations (A4/A5/A6)
| Sign combo | Count | Share % | Status |
|---|---|---|---|
| +++ | 0 | 0.0% | MISSING |
| ++- | 5 | 6.4% | RARE |
| +-+ | 10 | 12.8% | OK |
| +-- | 26 | 33.3% | OK |
| -++ | 1 | 1.3% | RARE |
| -+- | 27 | 34.6% | OK |
| --+ | 2 | 2.6% | RARE |
| --- | 7 | 9.0% | OK |
Sparse Wrist Zones (12-bin histogram)
| Axis | Sparse bin range | Count | Sample IDs |
|---|---|---|---|
| A4 | [-115.9, -95.1) | 3 | 31, 70, 72 |
| A4 | [-95.1, -74.3) | 2 | 28, 52 |
| A4 | [-74.3, -53.5) | 3 | 16, 33, 44 |
| A4 | [92.1, 112.9) | 2 | 30, 55 |
| A5 | [-95.4, -76.6) | 3 | 2, 42, 58 |
| A5 | [-57.7, -38.9) | 3 | 9, 26, 62 |
| A6 | [-151.2, -133.6) | 3 | 21, 49, 71 |
| A6 | [42.4, 60.0] | 1 | 68 |
Wrist Pair Coverage (6x6 occupancy)
| Pair | Occupied cells | Occupancy % | Sparse cells (≤2 points) |
|---|---|---|---|
| A4/A5 | 21/36 | 58.3% | 10 |
| A4/A6 | 23/36 | 63.9% | 7 |
| A5/A6 | 23/36 | 63.9% | 9 |
Orientation Summary (Az/El/Roll)
| Axis | Min | Max | Span | Std |
|---|---|---|---|---|
| Az | 0.000 | 0.000 | 0.000 | 0.000 |
| El | 0.000 | 0.000 | 0.000 | 0.000 |
| Roll | 0.000 | 0.000 | 0.000 | 0.000 |
| Samples used (kept) | 78 |
| Filtered samples excluded | 0 |
| Matched samples (raw) | 78 |
| Features | A1, A2, A3, A4, A5, A6, Xc, Yc, Zc, Az, El, Roll |
| Source | MeasuredPositions.csv (standard) |
| Feature | A1 | A2 | A3 | A4 | A5 | A6 | Xc | Yc | Zc | Az | El | Roll |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| A1 | 1.000 | -0.005 | -0.063 | -0.003 | 0.237 | -0.104 | -0.695 | -0.453 | -0.132 | 0.000 | 0.000 | 0.000 |
| A2 | -0.005 | 1.000 | -0.404 | 0.176 | -0.283 | 0.195 | 0.596 | -0.276 | -0.039 | 0.000 | 0.000 | 0.000 |
| A3 | -0.063 | -0.404 | 1.000 | -0.114 | -0.159 | -0.004 | -0.336 | 0.225 | -0.443 | 0.000 | 0.000 | 0.000 |
| A4 | -0.003 | 0.176 | -0.114 | 1.000 | -0.685 | 0.384 | 0.091 | 0.412 | 0.240 | 0.000 | 0.000 | 0.000 |
| A5 | 0.237 | -0.283 | -0.159 | -0.685 | 1.000 | -0.689 | -0.335 | -0.345 | -0.394 | 0.000 | 0.000 | 0.000 |
| A6 | -0.104 | 0.195 | -0.004 | 0.384 | -0.689 | 1.000 | 0.220 | 0.282 | 0.239 | 0.000 | 0.000 | 0.000 |
| Xc | -0.695 | 0.596 | -0.336 | 0.091 | -0.335 | 0.220 | 1.000 | 0.096 | 0.211 | 0.000 | 0.000 | 0.000 |
| Yc | -0.453 | -0.276 | 0.225 | 0.412 | -0.345 | 0.282 | 0.096 | 1.000 | 0.091 | 0.000 | 0.000 | 0.000 |
| Zc | -0.132 | -0.039 | -0.443 | 0.240 | -0.394 | 0.239 | 0.211 | 0.091 | 1.000 | 0.000 | 0.000 | 0.000 |
| Az | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 | 0.000 | 0.000 |
| El | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 | 0.000 |
| Roll | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 |
Top Correlated Measurement Pairs
| Feature A | Feature B | Pearson r | |r| |
|---|---|---|---|
| A1 | Xc | -0.695 | 0.695 |
| A5 | A6 | -0.689 | 0.689 |
| A4 | A5 | -0.685 | 0.685 |
| A2 | Xc | 0.596 | 0.596 |
| A1 | Yc | -0.453 | 0.453 |
| A3 | Zc | -0.443 | 0.443 |
| A4 | Yc | 0.412 | 0.412 |
| A2 | A3 | -0.404 | 0.404 |
| A5 | Zc | -0.394 | 0.394 |
| A4 | A6 | 0.384 | 0.384 |
| A5 | Yc | -0.345 | 0.345 |
| A3 | Xc | -0.336 | 0.336 |
How strongly each camera measurement axis is co-directed with each robot joint axis.
| Source | MeasuredPositions.csv (standard) |
| Matched samples (kept) | 78 |
| Filtered excluded | 0 |
Cross-Correlation Matrix
| Meas \ Robot | A1 | A2 | A3 | A4 | A5 | A6 |
|---|---|---|---|---|---|---|
| Xc | -0.695 | 0.596 | -0.336 | 0.091 | -0.335 | 0.220 |
| Yc | -0.453 | -0.276 | 0.225 | 0.412 | -0.345 | 0.282 |
| Zc | -0.132 | -0.039 | -0.443 | 0.240 | -0.394 | 0.239 |
| Az | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| El | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| Roll | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
Per-Axis Alignment Summary
| Measurement Axis | Dominant Robot Axis | Correlation | |r| | HHI | Eff. DoF | Verdict |
|---|---|---|---|---|---|---|
| Xc | A1 | -0.695 | 0.695 | 0.217 | 4.6 | MODERATE |
| Yc | A1 | -0.453 | 0.453 | 0.176 | 5.7 | WEAK |
| Zc | A3 | -0.443 | 0.443 | 0.219 | 4.6 | WEAK |
| Az | A1 | 0.000 | 0.000 | 0.000 | 1000000000000.0 | DECOUPLED |
| El | A1 | 0.000 | 0.000 | 0.000 | 1000000000000.0 | DECOUPLED |
| Roll | A1 | 0.000 | 0.000 | 0.000 | 1000000000000.0 | DECOUPLED |
Legend:LOCKED |r| ≥ 0.90 — almost entirely driven by one joint;
STRONG |r| ≥ 0.70 — strong coupling;
MODERATE |r| ≥ 0.50; WEAK |r| ≥ 0.30; DECOUPLED |r|
< 0.30 (best diversity). HHI = concentration index (0.167=uniform, 1.0=single axis). Eff. DoF = 1/HHI (6=ideal, 1=worst).
| Run ID | 20260505_164735_fb4832 |
| Timestamp | 2026-05-05T16:47:58 |
| Scope | Final (photogrammetry_full) |
| Source | Sample idx | Sample ID | Filtered | Filter score | Final residual | Position residual | Orientation residual | tx | ty | tz | rx | ry | rz |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| photogrammetry | 0 | 1 | NO | — | 0.248033 | 0.248033 | 0.000000 | 0.072233 | -0.168439 | 0.167127 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 1 | 2 | NO | — | 0.119376 | 0.119376 | 0.000000 | 0.103890 | 0.003742 | -0.058682 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 2 | 3 | NO | — | 0.198451 | 0.198451 | 0.000000 | 0.024561 | -0.194589 | 0.030243 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 3 | 4 | NO | — | 0.106314 | 0.106314 | 0.000000 | 0.034254 | -0.085179 | -0.053609 | -0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 4 | 5 | NO | — | 0.108011 | 0.108011 | 0.000000 | -0.007145 | 0.011811 | 0.107126 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 5 | 6 | NO | — | 0.078764 | 0.078764 | 0.000000 | 0.035909 | -0.054028 | 0.044669 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 6 | 7 | NO | — | 0.259460 | 0.259460 | 0.000000 | 0.123284 | -0.195888 | -0.117254 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 7 | 8 | NO | — | 0.167045 | 0.167045 | 0.000000 | -0.012905 | 0.165170 | -0.021357 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 8 | 9 | NO | — | 0.180036 | 0.180036 | 0.000000 | -0.080442 | 0.037031 | 0.156751 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 9 | 10 | NO | — | 0.092361 | 0.092361 | 0.000000 | -0.084670 | -0.020876 | -0.030426 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 10 | 11 | NO | — | 0.172148 | 0.172148 | 0.000000 | -0.036485 | -0.066858 | -0.154382 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 11 | 12 | NO | — | 0.155519 | 0.155519 | 0.000000 | -0.061176 | -0.140236 | -0.027884 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 12 | 13 | NO | — | 0.258371 | 0.258371 | 0.000000 | -0.099601 | -0.077511 | 0.225448 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 13 | 14 | NO | — | 0.227546 | 0.227546 | 0.000000 | 0.137619 | -0.156396 | -0.091535 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 14 | 15 | NO | — | 0.249098 | 0.249098 | 0.000000 | -0.013799 | 0.159848 | 0.190547 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 15 | 16 | NO | — | 0.228571 | 0.228571 | 0.000000 | -0.117118 | -0.156403 | -0.118602 | 0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 16 | 17 | NO | — | 0.234488 | 0.234488 | 0.000000 | -0.087780 | -0.196051 | -0.094039 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 17 | 18 | NO | — | 0.131152 | 0.131152 | 0.000000 | -0.039972 | -0.112903 | -0.053441 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 18 | 19 | NO | — | 0.174234 | 0.174234 | 0.000000 | -0.096048 | 0.144219 | -0.018254 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 19 | 20 | NO | — | 0.088564 | 0.088564 | 0.000000 | -0.068126 | -0.048860 | -0.028552 | 0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 20 | 21 | NO | — | 0.208362 | 0.208362 | 0.000000 | -0.070474 | 0.145959 | -0.130935 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 21 | 22 | NO | — | 0.079826 | 0.079826 | 0.000000 | -0.020598 | 0.039316 | 0.066349 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 22 | 23 | NO | — | 0.279016 | 0.279016 | 0.000000 | -0.076998 | 0.257344 | 0.075465 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 23 | 24 | NO | — | 0.073968 | 0.073968 | 0.000000 | -0.008413 | 0.043360 | 0.059333 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 24 | 25 | NO | — | 0.401803 | 0.401803 | 0.000000 | 0.337647 | 0.110291 | 0.187819 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 25 | 26 | NO | — | 0.194686 | 0.194686 | 0.000000 | -0.033762 | 0.042738 | 0.186913 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 26 | 27 | NO | — | 0.152697 | 0.152697 | 0.000000 | 0.067778 | 0.116125 | 0.072372 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 27 | 28 | NO | — | 0.115295 | 0.115295 | 0.000000 | 0.036289 | -0.027294 | 0.105976 | -0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 28 | 29 | NO | — | 0.272520 | 0.272520 | 0.000000 | -0.053203 | 0.262977 | -0.047747 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 29 | 30 | NO | — | 0.094745 | 0.094745 | 0.000000 | -0.027888 | 0.011083 | 0.089867 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 30 | 31 | NO | — | 0.197779 | 0.197779 | 0.000000 | -0.027779 | 0.185251 | 0.063456 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 31 | 32 | NO | — | 0.253963 | 0.253963 | 0.000000 | 0.087075 | 0.186306 | 0.149014 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 32 | 33 | NO | — | 0.189139 | 0.189139 | 0.000000 | 0.057109 | -0.179580 | 0.016223 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 33 | 34 | NO | — | 0.166906 | 0.166906 | 0.000000 | 0.045689 | 0.121330 | -0.105115 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 34 | 35 | NO | — | 0.219871 | 0.219871 | 0.000000 | 0.062206 | -0.031424 | -0.208534 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 35 | 36 | NO | — | 0.324345 | 0.324345 | 0.000000 | -0.104388 | -0.229406 | 0.204145 | -0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 36 | 37 | NO | — | 0.166764 | 0.166764 | 0.000000 | -0.136458 | -0.026712 | -0.092066 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 37 | 38 | NO | — | 0.257893 | 0.257893 | 0.000000 | -0.004918 | -0.010968 | -0.257613 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 38 | 39 | NO | — | 0.350976 | 0.350976 | 0.000000 | 0.168989 | -0.233084 | 0.200746 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 39 | 40 | NO | — | 0.199855 | 0.199855 | 0.000000 | 0.044939 | 0.023889 | 0.193266 | -0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 40 | 41 | NO | — | 0.187924 | 0.187924 | 0.000000 | -0.168203 | 0.069640 | -0.046617 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 41 | 42 | NO | — | 0.311604 | 0.311604 | 0.000000 | 0.058403 | -0.238634 | 0.191677 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 42 | 43 | NO | — | 0.187598 | 0.187598 | 0.000000 | -0.066475 | 0.172390 | -0.032491 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 43 | 44 | NO | — | 0.231809 | 0.231809 | 0.000000 | 0.157871 | -0.157990 | 0.062059 | 0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 44 | 45 | NO | — | 0.066803 | 0.066803 | 0.000000 | -0.065639 | -0.004845 | -0.011431 | -0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 45 | 46 | NO | — | 0.166012 | 0.166012 | 0.000000 | 0.134917 | 0.096669 | 0.003532 | -0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 46 | 47 | NO | — | 0.267567 | 0.267567 | 0.000000 | -0.006263 | -0.218059 | -0.154930 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 47 | 48 | NO | — | 0.263504 | 0.263504 | 0.000000 | 0.099872 | -0.079779 | -0.230423 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 48 | 49 | NO | — | 0.150579 | 0.150579 | 0.000000 | -0.067437 | 0.095549 | 0.094851 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 49 | 50 | NO | — | 0.254584 | 0.254584 | 0.000000 | -0.012925 | -0.123124 | 0.222456 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 50 | 51 | NO | — | 0.284076 | 0.284076 | 0.000000 | -0.162992 | 0.231807 | 0.019951 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 51 | 52 | NO | — | 0.220343 | 0.220343 | 0.000000 | -0.001767 | -0.151096 | 0.160368 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 52 | 53 | NO | — | 0.029701 | 0.029701 | 0.000000 | 0.005492 | -0.023069 | 0.017884 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 53 | 54 | NO | — | 0.103080 | 0.103080 | 0.000000 | 0.034442 | -0.066907 | -0.070447 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 54 | 55 | NO | — | 0.146957 | 0.146957 | 0.000000 | -0.049467 | 0.109094 | 0.085134 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 55 | 56 | NO | — | 0.158453 | 0.158453 | 0.000000 | -0.073961 | 0.124383 | -0.064545 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 56 | 57 | NO | — | 0.052434 | 0.052434 | 0.000000 | 0.043446 | -0.000151 | 0.029356 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 57 | 58 | NO | — | 0.073886 | 0.073886 | 0.000000 | 0.072830 | -0.011479 | 0.004803 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 58 | 59 | NO | — | 0.049058 | 0.049058 | 0.000000 | 0.020642 | -0.026656 | -0.035638 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 59 | 60 | NO | — | 0.049762 | 0.049762 | 0.000000 | 0.016593 | 0.020814 | 0.042044 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 60 | 61 | NO | — | 0.155924 | 0.155924 | 0.000000 | -0.094827 | 0.114112 | -0.047944 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 61 | 62 | NO | — | 0.282592 | 0.282592 | 0.000000 | -0.097034 | 0.263395 | -0.032652 | -0.000000 | -0.000000 | 0.000000 |
| photogrammetry | 62 | 63 | NO | — | 0.063448 | 0.063448 | 0.000000 | 0.003125 | 0.055444 | -0.030691 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 63 | 64 | NO | — | 0.185189 | 0.185189 | 0.000000 | -0.021735 | -0.178415 | -0.044617 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 64 | 65 | NO | — | 0.255452 | 0.255452 | 0.000000 | -0.149960 | 0.145523 | 0.146938 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 65 | 66 | NO | — | 0.294472 | 0.294472 | 0.000000 | -0.110921 | 0.256436 | -0.093008 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 66 | 67 | NO | — | 0.239096 | 0.239096 | 0.000000 | 0.188251 | -0.144667 | -0.028284 | -0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 67 | 68 | NO | — | 0.219018 | 0.219018 | 0.000000 | -0.092846 | 0.169979 | -0.102253 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 68 | 69 | NO | — | 0.408491 | 0.408491 | 0.000000 | 0.329343 | -0.171630 | 0.170121 | -0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 69 | 70 | NO | — | 0.168061 | 0.168061 | 0.000000 | -0.045331 | -0.123718 | -0.104324 | -0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 70 | 71 | NO | — | 0.236668 | 0.236668 | 0.000000 | -0.189319 | -0.141048 | 0.016595 | 0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 71 | 72 | NO | — | 0.080450 | 0.080450 | 0.000000 | -0.013160 | -0.048404 | -0.062897 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 72 | 73 | NO | — | 0.440136 | 0.440136 | 0.000000 | 0.186522 | 0.213061 | -0.336948 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 73 | 74 | NO | — | 0.239211 | 0.239211 | 0.000000 | -0.068566 | -0.056330 | 0.222143 | -0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 74 | 75 | NO | — | 0.219513 | 0.219513 | 0.000000 | -0.139882 | 0.168889 | 0.009771 | 0.000000 | 0.000000 | 0.000000 |
| photogrammetry | 75 | 76 | NO | — | 0.224222 | 0.224222 | 0.000000 | 0.140334 | -0.169803 | 0.041818 | -0.000000 | 0.000000 | -0.000000 |
| photogrammetry | 76 | 77 | NO | — | 0.258522 | 0.258522 | 0.000000 | 0.116912 | 0.230265 | -0.011981 | -0.000000 | -0.000000 | -0.000000 |
| photogrammetry | 77 | 78 | NO | — | 0.144051 | 0.144051 | 0.000000 | 0.020570 | 0.092890 | -0.108162 | -0.000000 | -0.000000 | -0.000000 |
Analysis
| Total points | 78 |
| Filtered points | 0 |
| Filtered ratio | 0.0% |
| Kept points | 78 |
| All points max residual | 0.440136 |
| Kept points max residual | 0.440136 |
| Kept points mean residual | 0.192926 |
Top 10 Residual Points
| Rank | Source | Sample idx | Sample ID | Filtered | Final residual |
|---|---|---|---|---|---|
| 1 | photogrammetry | 72 | 73 | NO | 0.440136 |
| 2 | photogrammetry | 68 | 69 | NO | 0.408491 |
| 3 | photogrammetry | 24 | 25 | NO | 0.401803 |
| 4 | photogrammetry | 38 | 39 | NO | 0.350976 |
| 5 | photogrammetry | 35 | 36 | NO | 0.324345 |
| 6 | photogrammetry | 41 | 42 | NO | 0.311604 |
| 7 | photogrammetry | 65 | 66 | NO | 0.294472 |
| 8 | photogrammetry | 50 | 51 | NO | 0.284076 |
| 9 | photogrammetry | 61 | 62 | NO | 0.282592 |
| 10 | photogrammetry | 22 | 23 | NO | 0.279016 |
📖 How to interpret this tab
| Metric | How to understand | Good sign | Warning sign | What to do |
|---|---|---|---|---|
| Effective FIM rank | How many independent parameter directions are actually observed by your dataset. | Rank close to total parameter count. | Rank much smaller than parameter count. | Add fundamentally different geometries; replacing duplicates is more useful than adding volume. |
| FIM eigenvalue spectrum | Information strength per direction. Bottom tail reveals weak or unobservable combinations. | Tail has no WEAK/DEFICIENT/NEGLIGIBLE zones. | Long weak tail and many deficient directions. | Use weakest-direction hints to build poses that separate listed parameter groups. |
| Weakest parameter directions | Dominant parameters in poorly observed eigenvectors; not currently separable. | No near-zero directions or only mild weak ones. | Top weak directions are near zero. | Create measurements where listed parameters respond with clearly different signatures. |
| Parameter uncertainty (from FIM inverse) | Direct precision score per parameter after considering all couplings. | No CRITICAL statuses. | CRITICAL/HIGH statuses persist. | Add targeted poses for those exact parameters before broad random additions. |
| Leverage uniformity | Whether information is spread across samples or concentrated in a few influential points. | Uniformity is moderate/high. | Uniformity is low; only a few samples dominate. | Replace low-value repetitive samples with novel poses in under-excited regions. |
| Per-sample leverage (HIGH/REDUNDANT) | HIGH samples carry unique information. REDUNDANT samples are likely replaceable. | Small controlled set of HIGH + very few REDUNDANT. | Many REDUNDANT samples. | Keep HIGH samples, replace REDUNDANT first when planning additional measurements. |
| Inseparable parameter groups | Parameters moving together in null-space; current data cannot distinguish them. | No CRITICAL groups. | CRITICAL group exists. | Plan pose families that flip sign/magnitude differently for group members. |
| Redundancy clusters | Near-identical Jacobian signatures across samples; little new information. | Few/no clusters with large replaceable tails. | Many clusters and many replaceable IDs. | Do substitution: remove cluster duplicates, add orthogonal poses instead. |
| Marginal information gain ordering | Greedy ranking of sample usefulness (early = informative, late = weak). | Information gain decays slowly. | Long low-value tail appears early. | Retain top-ranked samples and replace tail with designs guided by weak directions. |
| Diagnostic recommendations (P0..P3) | Consolidated action plan generated from all metrics. | No P0/P1 entries. | P0/P1 entries remain. | Treat table as execution order: close P0, re-run, then proceed to lower priorities. |
Information-theoretic analysis of dataset quality beyond correlation
Information Budget
| Error terms (rows) | 468 |
| Parameters (columns) | 22 |
| Measurement/param ratio | 21.3 |
| FIM condition number | 2.70e+08 |
| Effective FIM rank | 17/22 |
| Information utilization | 77.3% |
| D-efficiency | 1.5055e+04 |
| A-optimality (trace FIM⁻¹) | 9.8211e-01 |
| E-optimality (min eigenvalue) | 2.5119e+00 |
FIM Eigenvalue Spectrum
Eigenvalues of J^T*J sorted descending. Small values = information-starved directions.
| # | Eigenvalue | Ratio to max | Cumulative % | Status |
|---|---|---|---|---|
| 1 | 6.7749e+08 | 1.0000e+00 | 44.14% | OK |
| 2 | 6.1975e+08 | 9.1478e-01 | 84.52% | OK |
| 3 | 8.9830e+07 | 1.3259e-01 | 90.37% | OK |
| ... | (6 OK eigenvalues omitted) | |||
| 10 | 6.5443e+06 | 9.6596e-03 | 99.31% | FAIR |
| 11 | 3.2371e+06 | 4.7781e-03 | 99.52% | FAIR |
| 12 | 2.7666e+06 | 4.0837e-03 | 99.70% | FAIR |
| 13 | 2.5028e+06 | 3.6942e-03 | 99.87% | FAIR |
| 14 | 9.9957e+05 | 1.4754e-03 | 99.93% | FAIR |
| 15 | 8.9063e+05 | 1.3146e-03 | 99.99% | FAIR |
| 16 | 1.3233e+05 | 1.9532e-04 | 100.00% | WEAK |
| 17 | 3.4847e+04 | 5.1435e-05 | 100.00% | WEAK |
| 18 | 2.5701e+01 | 3.7935e-08 | 100.00% | DEFICIENT |
| 19 | 2.2277e+01 | 3.2881e-08 | 100.00% | DEFICIENT |
| 20 | 6.3161e+00 | 9.3229e-09 | 100.00% | DEFICIENT |
| 21 | 2.9253e+00 | 4.3179e-09 | 100.00% | DEFICIENT |
| 22 | 2.5119e+00 | 3.7077e-09 | 100.00% | DEFICIENT |
Weakest Parameter Directions
Bottom eigenvectors of FIM — parameter combinations with least information
| Rank | λ value | λ/λ_max | Dominant Parameters | Weights | Assessment |
|---|---|---|---|---|---|
| 1 | 2.5119e+00 | 3.71e-09 | link5/offset_z
link3/offset_x link6/offset_z link6/offset_x | +0.783
-0.556 +0.234 -0.148 | DEFICIENT — severe information deficit, add targeted measurements |
| 2 | 2.9253e+00 | 4.32e-09 | link6/offset_z
link6/offset_x link5/offset_z link5/offset_x | -0.878
+0.369 +0.267 -0.111 | DEFICIENT — severe information deficit, add targeted measurements |
| 3 | 6.3161e+00 | 9.32e-09 | link3/offset_x
link5/offset_z | -0.825
-0.558 | DEFICIENT — severe information deficit, add targeted measurements |
| 4 | 2.2277e+01 | 3.29e-08 | link5/offset_x
link6/offset_x link6/offset_z | +0.949
-0.222 -0.221 | DEFICIENT — severe information deficit, add targeted measurements |
| 5 | 2.5701e+01 | 3.79e-08 | link6/offset_x
link6/offset_z link5/offset_x | +0.887
+0.353 +0.291 | DEFICIENT — severe information deficit, add targeted measurements |
Parameter Uncertainty (from FIM inverse)
Diagonal of (J^T*J)^{-1} — variance of each parameter estimate
| Parameter | Variance | Std Dev | Relative | Status |
|---|---|---|---|---|
| link1/offset_rx | 1.3482e-07 | 3.6718e-04 | 0.0000 | LOW |
| link1/offset_ry | 1.4386e-07 | 3.7929e-04 | 0.0000 | LOW |
| link1/offset_rz | 1.6625e-05 | 4.0774e-03 | 0.0001 | LOW |
| link2/offset_rx | 6.0630e-07 | 7.7865e-04 | 0.0000 | LOW |
| link2/offset_ry | 2.2181e-05 | 4.7097e-03 | 0.0001 | LOW |
| link2/offset_rz | 4.1620e-07 | 6.4514e-04 | 0.0000 | LOW |
| link3/offset_x | 2.3414e-01 | 4.8388e-01 | 0.7365 | HIGH |
| link3/offset_rx | 1.1133e-06 | 1.0551e-03 | 0.0000 | LOW |
| link3/offset_ry | 6.4516e-07 | 8.0322e-04 | 0.0000 | LOW |
| link3/offset_rz | 3.9203e-07 | 6.2612e-04 | 0.0000 | LOW |
| link4/offset_rx | 1.1494e-06 | 1.0721e-03 | 0.0000 | LOW |
| link4/offset_ry | 5.6100e-07 | 7.4900e-04 | 0.0000 | LOW |
| link4/offset_rz | 1.6090e-07 | 4.0113e-04 | 0.0000 | LOW |
| link5/offset_x | 4.8534e-02 | 2.2031e-01 | 0.1527 | MODERATE |
| link5/offset_z | 3.1791e-01 | 5.6383e-01 | 1.0000 | CRITICAL |
| link5/offset_ry | 2.2916e-07 | 4.7871e-04 | 0.0000 | LOW |
| link5/offset_rz | 4.5620e-06 | 2.1359e-03 | 0.0000 | LOW |
| link6/offset_x | 8.9071e-02 | 2.9845e-01 | 0.2802 | MODERATE |
| link6/offset_z | 2.9239e-01 | 5.4073e-01 | 0.9197 | CRITICAL |
| link6/offset_rx | 1.6098e-06 | 1.2688e-03 | 0.0000 | LOW |
| link6/offset_ry | 1.0499e-05 | 3.2402e-03 | 0.0000 | LOW |
| link6/offset_rz | 9.9217e-07 | 9.9608e-04 | 0.0000 | LOW |
Sample Leverage Analysis
Hat matrix diagonal h_ii — measures influence of each error term
| Total error terms | 468 |
| Expected leverage (p/M) | 0.047009 |
| Mean leverage | 0.047009 |
| Std leverage | 0.053130 |
| Min leverage | 0.000000 |
| Max leverage | 0.219174 |
| Coefficient of variation | 1.130 |
| Leverage uniformity | 0.214 |
| High-leverage terms (h > 0.0940) | 92 |
| Low-leverage terms (h < 0.0235) | 234 |
Per-Sample Leverage
| # | Sample ID | Leverage | Verdict |
|---|---|---|---|
| 1 | 53 | 0.136094 | REDUNDANT |
| 2 | 20 | 0.124825 | REDUNDANT |
| (76 NORMAL samples omitted) |
Inseparable Parameter Groups
Parameters linked in null-space — cannot be separated with current data
| Group # | Size | Parameters | Null dims | Severity |
|---|---|---|---|---|
| 1 | 5 | link3/offset_x
link5/offset_z link6/offset_z link6/offset_x link5/offset_x | 5 | CRITICAL |
Redundancy Clusters
Groups of samples with cosine similarity ≥ 0.95 — near-identical information
| Cluster # | Size | Avg similarity | Keep (sample ID) | Replaceable (sample IDs) |
|---|---|---|---|---|
| 1 | 16 | 0.9456 | 1 | 4, 8, 10, 19, 20, 21, 23, 36, 43, 45, 52, 54, 56, 61, 71 |
| 2 | 11 | 0.9625 | 2 | 3, 6, 11, 39, 42, 53, 57, 62, 64, 78 |
| 3 | 8 | 0.9484 | 5 | 13, 14, 24, 26, 33, 50, 74 |
| 4 | 6 | 0.9635 | 9 | 12, 17, 25, 35, 47 |
| 5 | 6 | 0.9561 | 27 | 29, 55, 58, 59, 60 |
| 6 | 4 | 0.9511 | 37 | 38, 41, 68 |
| 7 | 3 | 0.9531 | 15 | 32, 77 |
| 8 | 3 | 0.9371 | 34 | 65, 67 |
| 9 | 2 | 0.9579 | 7 | 22 |
| 10 | 2 | 0.9609 | 18 | 30 |
| 11 | 2 | 0.9628 | 28 | 44 |
| 12 | 2 | 0.9555 | 31 | 72 |
| 13 | 2 | 0.9823 | 40 | 49 |
| 14 | 2 | 0.9666 | 48 | 69 |
| 15 | 2 | 0.9758 | 51 | 75 |
Marginal Information Gain Ordering
Greedy forward selection: each step picks the sample maximising log-det(FIM)
| Step | Sample ID | log-det(FIM) | Cum. rank |
|---|---|---|---|
| 1 | 29 | -5119.6778 | 3 |
| 2 | 46 | -4315.9191 | 6 |
| 3 | 35 | -2872.3530 | 9 |
| 4 | 3 | -2093.7653 | 12 |
| 5 | 26 | -660.3226 | 15 |
| 6 | 10 | 99.1942 | 17 |
| ... | (67 samples omitted) | ||
| 74 | 12 | 278.3068 | 17 |
| 75 | 1 | 278.4721 | 17 |
| 76 | 36 | 278.6238 | 17 |
| 77 | 53 | 278.7634 | 17 |
| 78 | 20 | 278.8910 | 17 |
Diagnostic Recommendations
| Priority | Metric | Current | Target | Required Action | Why |
|---|---|---|---|---|---|
| P0 | Effective FIM rank | 17/22 | 22/22 | 5 parameter direction(s) have negligible information — add fundamentally different measurement poses | Rank-deficient FIM means some parameters (or combinations) cannot be determined at all |
| P0 | Weak direction #1 | λ/λ_max = 3.71e-09 | λ/λ_max > 0.001 | Add poses that vary [link5/offset_z + link3/offset_x + link6/offset_z] independently | DEFICIENT — severe information deficit, add targeted measurements |
| P0 | Weak direction #2 | λ/λ_max = 4.32e-09 | λ/λ_max > 0.001 | Add poses that vary [link6/offset_z + link6/offset_x + link5/offset_z] independently | DEFICIENT — severe information deficit, add targeted measurements |
| P0 | Weak direction #3 | λ/λ_max = 9.32e-09 | λ/λ_max > 0.001 | Add poses that vary [link3/offset_x + link5/offset_z] independently | DEFICIENT — severe information deficit, add targeted measurements |
| P0 | Weak direction #4 | λ/λ_max = 3.29e-08 | λ/λ_max > 0.001 | Add poses that vary [link5/offset_x + link6/offset_x + link6/offset_z] independently | DEFICIENT — severe information deficit, add targeted measurements |
| P0 | Weak direction #5 | λ/λ_max = 3.79e-08 | λ/λ_max > 0.001 | Add poses that vary [link6/offset_x + link6/offset_z + link5/offset_x] independently | DEFICIENT — severe information deficit, add targeted measurements |
| P1 | Coverage Az | entropy=0.00, coverage=10%, gap=0.0deg | entropy ≥ 0.70, coverage ≥ 60% | Add samples near Az=0.0deg (largest gap = 0% of range) | Clustered sampling leaves coverage holes that reduce parameter sensitivity |
| P1 | Coverage El | entropy=0.00, coverage=10%, gap=0.0deg | entropy ≥ 0.70, coverage ≥ 60% | Add samples near El=0.0deg (largest gap = 0% of range) | Clustered sampling leaves coverage holes that reduce parameter sensitivity |
| P1 | Coverage Roll | entropy=0.00, coverage=10%, gap=0.0deg | entropy ≥ 0.70, coverage ≥ 60% | Add samples near Roll=0.0deg (largest gap = 0% of range) | Clustered sampling leaves coverage holes that reduce parameter sensitivity |
| P2 | Leverage uniformity | 0.21 (CV=1.13) | > 0.30 | Replace redundant poses (low leverage) with novel configurations | Non-uniform leverage means some measurements dominate while others are wasted |
| P0 | Parameter uncertainty | 2 CRITICAL parameters: link5/offset_z, link6/offset_z | No CRITICAL parameters | These parameters have extremely high variance — need dedicated excitation | High FIM-inverse diagonal means the parameter value is poorly constrained by data |
| P0 | Inseparable group (5 params) | link3/offset_x, link5/offset_z, link6/offset_z, link6/offset_x... | Each parameter independently identifiable | Add measurements where these parameters produce DIFFERENT Jacobian signatures | 5 null-space dimension(s) — these parameters move together in current data |
| P2 | Redundancy clusters | 8 cluster(s), 49 replaceable samples | No clusters with ≥ 3 near-identical samples | Replace redundant samples with novel pose configurations | Near-identical Jacobian rows provide no additional information |
| P3 | Marginal information tail | Last 70 samples added < 1% of total information | Each sample contributes meaningfully | These samples could be replaced with data that fills coverage gaps | Diminishing returns indicate saturation of current measurement geometry |
Dataset Quality: WARNING
• Weak sign-configuration coverage (57/64 patterns missing)
• Low workspace cube coverage (29%)
• Low workspace cube coverage (29%)
Summary
| Samples | 78 |
| Axes (DOF) | 6 |
| Duplicate ratio | 0.0% |
| Unique poses | 78 / 78 |
| NN joint median | 31.45 deg |
| NN joint p95 | 50.86 deg |
| Workspace cells | 63 / 216 |
| Workspace coverage | 29.2% |
| NN workspace median | 114.91 mm |
| A5 near-singularity | 1.3% |
Joint Axis Statistics (degrees)
| Axis | Min | Max | Span | Std |
|---|---|---|---|---|
| A1 | 16.94 | 76.99 | 60.05 | 13.60 |
| A2 | -133.66 | -58.67 | 74.99 | 14.71 |
| A3 | 64.22 | 149.90 | 85.68 | 18.76 |
| A4 | -115.92 | 133.69 | 249.61 | 57.99 |
| A5 | -114.16 | 111.50 | 225.66 | 76.26 |
| A6 | -151.16 | 60.00 | 211.16 | 52.77 |
Axis Histogram Imbalance
| Axis | Min bin | Max bin | CV |
|---|---|---|---|
| A1 | 1 | 19 | 0.51 |
| A2 | 1 | 29 | 0.84 |
| A3 | 2 | 19 | 0.59 |
| A4 | 3 | 17 | 0.50 |
| A5 | 5 | 27 | 0.69 |
| A6 | 4 | 14 | 0.32 |
Sign Configurations (Observed)
| Pattern | Count | Share | Status |
|---|---|---|---|
| +-+++- | 5 | 6.4% | OBSERVED |
| +-++-+ | 10 | 12.8% | OBSERVED |
| +-++-- | 26 | 33.3% | OBSERVED |
| +-+-++ | 1 | 1.3% | OBSERVED |
| +-+-+- | 27 | 34.6% | OBSERVED |
| +-+--+ | 2 | 2.6% | OBSERVED |
| +-+--- | 7 | 9.0% | OBSERVED |
Wrist Pair Occupancy
| Pair | Occupied / Total | Coverage | Sparse (≤2) |
|---|---|---|---|
| A4 / A5 | 21 / 36 | 58.3% | 10 |
| A4 / A6 | 23 / 36 | 63.9% | 7 |
| A5 / A6 | 23 / 36 | 63.9% | 9 |
Joint-Angle Correlations (sampling)
| Pair | Corr | |Corr| |
|---|---|---|
| A5 / A6 | -0.689 | 0.689 |
| A4 / A5 | -0.685 | 0.685 |
| A2 / A3 | -0.404 | 0.404 |
| A4 / A6 | 0.384 | 0.384 |
| A2 / A5 | -0.283 | 0.283 |
| A1 / A5 | 0.237 | 0.237 |
| A2 / A6 | 0.195 | 0.195 |
| A2 / A4 | 0.176 | 0.176 |
| A3 / A5 | -0.159 | 0.159 |
| A3 / A4 | -0.114 | 0.114 |
Correlation Matrix
| A1 | A2 | A3 | A4 | A5 | A6 | |
|---|---|---|---|---|---|---|
| A1 | 1.000 | -0.005 | -0.063 | -0.003 | 0.237 | -0.104 |
| A2 | -0.005 | 1.000 | -0.404 | 0.176 | -0.283 | 0.195 |
| A3 | -0.063 | -0.404 | 1.000 | -0.114 | -0.159 | -0.004 |
| A4 | -0.003 | 0.176 | -0.114 | 1.000 | -0.685 | 0.384 |
| A5 | 0.237 | -0.283 | -0.159 | -0.685 | 1.000 | -0.689 |
| A6 | -0.104 | 0.195 | -0.004 | 0.384 | -0.689 | 1.000 |
| Current correlation health score | 74/100 |
| Overall health | 77/100 (B) |
Which parameter, if frozen (removed from optimization), gives the biggest health improvement?
Single-Parameter Freeze Impact
| Rank | Parameter | Corr Health If Frozen | Change |
|---|---|---|---|
| 1 | link5/offset_rz | 81 | +7 |
| 2 | link6/offset_ry | 81 | +7 |
| 3 | link1/offset_rz | 79 | +5 |
| 4 | link2/offset_ry | 79 | +5 |
| 5 | link6/offset_z | 76 | +2 |
| 6 | link6/offset_rx | 76 | +2 |
| 7 | link3/offset_ry | 75 | +1 |
| 8 | link6/offset_x | 75 | +1 |
| 9 | link3/offset_x | 74 | +0 |
| 10 | link4/offset_rx | 74 | +0 |
| 11 | link5/offset_x | 74 | +0 |
| 12 | link5/offset_z | 74 | +0 |
| 13 | link2/offset_rx | 73 | -1 |
| 14 | link2/offset_rz | 73 | -1 |
| 15 | link3/offset_rz | 73 | -1 |
| 16 | link5/offset_ry | 73 | -1 |
| 17 | link6/offset_rz | 73 | -1 |
| 18 | link1/offset_rx | 72 | -2 |
| 19 | link1/offset_ry | 72 | -2 |
| 20 | link3/offset_rx | 72 | -2 |
| 21 | link4/offset_ry | 72 | -2 |
| 22 | link4/offset_rz | 72 | -2 |
Greedy Freeze Sequence
| Step | Freeze Parameter | Corr Health After | Step Gain | Total Gain | Params Left |
|---|---|---|---|---|---|
| 1 | link5/offset_rz | 81 | +7 | +7 | 21 |
| 2 | link1/offset_rz | 88 | +7 | +14 | 20 |
| 3 | link3/offset_rz | 90 | +2 | +16 | 19 |
| 4 | link2/offset_rx | 92 | +2 | +18 | 18 |
| 5 | link1/offset_ry | 93 | +1 | +19 | 17 |
[2026-05-05T16:47:58] Run ID: 20260505_164735_fb4832
[2026-05-05T16:47:58] Timestamp: 2026-05-05T16:47:58
[2026-05-05T16:47:58] Status: success
[2026-05-05T16:47:58] Duration: 20.98s
[2026-05-05T16:47:58] Iterations: 8
[2026-05-05T16:47:58] Final cost: 3.4846785488e+00
[2026-05-05T16:47:58] Stage configuration:
Stage #1: photogrammetry_init (enabled)
Data sources: photogrammetry
Parameter link types: base, camera
Error term: camera_pose (type=camera_pose, weight=1.000000, rotation=1.000000, touch=1.000000, optimize_centers=False)
Total samples: 78
Stage #2: photogrammetry_full (enabled)
Data sources: photogrammetry
Parameter link types: link1, link2, link3, link4, link5, link6
Error term: camera_pose (type=camera_pose, weight=1.000000, rotation=1.000000, touch=1.000000, optimize_centers=False)
Total samples: 78
[2026-05-05T16:47:58] Stage execution summary:
Stage: photogrammetry_init
Iterations: 4, final_cost: 2.1029613906e+01
Cost reduction: —
Converged: True (Converged: cost change below tolerance)
Optimized params: 12
Data sources: 1, samples: 78
Elapsed: 8.79s
Stage: photogrammetry_full
Iterations: 4, final_cost: 3.4846785488e+00
Cost reduction: —
Converged: True (Converged: cost change below tolerance)
Optimized params: 22
Data sources: 1, samples: 78
Elapsed: 10.92s
[2026-05-05T16:47:58] Pose error improvement by stage (camera_pose):
Stage: photogrammetry_init
Position mean: 107.757802 -> 0.478247 (+99.56%)
Position std : 47.775076 -> 0.202213 (+99.58%)
Position max : 215.457560 -> 1.131757 (+99.47%)
Orientation: N/A (no informative RPY measurements for this stage)
Stage: photogrammetry_full
Position mean: 0.478247 -> 0.192926 (+59.66%)
Position std : 0.202213 -> 0.086343 (+57.30%)
Position max : 1.131757 -> 0.440136 (+61.11%)
Orientation: N/A (no informative RPY measurements for this stage)
Rotation weight: 0.000000
[2026-05-05T16:47:58] Parameter changes:
link1/offset_x: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link1/offset_y: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link1/offset_z: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link1/offset_rx: 0.00000000 → 0.01669853 (Δ=+0.01669853)
link1/offset_ry: 0.00000000 → -0.02188579 (Δ=-0.02188579)
link1/offset_rz: 0.00000000 → -0.03129690 (Δ=-0.03129690)
link2/offset_x: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link2/offset_y: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link2/offset_z: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link2/offset_rx: 0.00000000 → -0.01157515 (Δ=-0.01157515)
link2/offset_ry: 0.00000000 → -0.03141320 (Δ=-0.03141320)
link2/offset_rz: 0.00000000 → 0.00511982 (Δ=+0.00511982)
link3/offset_x: 0.00000000 → 0.01364120 (Δ=+0.01364120)
link3/offset_y: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link3/offset_z: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link3/offset_rx: 0.00000000 → 0.01070026 (Δ=+0.01070026)
link3/offset_ry: 0.00000000 → 0.01798839 (Δ=+0.01798839)
link3/offset_rz: 0.00000000 → 0.03285708 (Δ=+0.03285708)
link4/offset_x: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link4/offset_y: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link4/offset_z: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link4/offset_rx: 0.00000000 → 0.00150677 (Δ=+0.00150677)
link4/offset_ry: 0.00000000 → 0.00972680 (Δ=+0.00972680)
link4/offset_rz: 0.00000000 → -0.04165995 (Δ=-0.04165995)
link5/offset_x: 0.00000000 → 0.04480781 (Δ=+0.04480781)
link5/offset_y: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link5/offset_z: 0.00000000 → 0.39340333 (Δ=+0.39340333)
link5/offset_rx: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link5/offset_ry: 0.00000000 → 0.00010106 (Δ=+0.00010106)
link5/offset_rz: 0.00000000 → -0.04569140 (Δ=-0.04569140)
link6/offset_x: 0.00000000 → -0.53301134 (Δ=-0.53301134)
link6/offset_y: 0.00000000 → 0.00000000 (Δ=+0.00000000)
link6/offset_z: 0.00000000 → 0.02999485 (Δ=+0.02999485)
link6/offset_rx: 0.00000000 → -0.07284822 (Δ=-0.07284822)
link6/offset_ry: 0.00000000 → -0.10637707 (Δ=-0.10637707)
link6/offset_rz: 0.00000000 → -0.03105290 (Δ=-0.03105290)
base/x: -4347.13995974 → -4860.26533672 (Δ=-513.12537698)
base/y: 34.72651344 → -65.69470448 (Δ=-100.42121792)
base/z: -1366.37595269 → -1296.82912229 (Δ=+69.54683040)
base/rx: 2.78825079 → 0.75853597 (Δ=-2.02971482)
base/ry: -3.44856180 → 0.83465768 (Δ=+4.28321948)
base/rz: 5.70954890 → 3.82233297 (Δ=-1.88721594)
camera/x: -92.13383677 → 253.39157543 (Δ=+345.52541219)
camera/y: -0.32489083 → 3.89607563 (Δ=+4.22096645)
camera/z: 158.96389818 → 417.83763200 (Δ=+258.87373382)
camera/rx: -149.64923032 → -149.64923032 (Δ=+0.00000000)
camera/ry: -45.50998792 → -45.50998792 (Δ=-0.00000000)
camera/rz: 15.06683992 → 15.06683992 (Δ=-0.00000000)
Ready to close
the accuracy gap?
Thank you for your request. We will contact you shortly.