Docs: mark tactile latency CORRECTED in-data (no loader compensation needed)
Browse files- README.md +3 -3
- tasks.json +4 -1
README.md
CHANGED
|
@@ -134,9 +134,9 @@ sample = ds[0]
|
|
| 134 |
```
|
| 135 |
`mode="segment"` iterates clean spans (no bad frames by construction); `mode="window"` slides over whole episodes and skips `bad_frames.json` intervals. Backend: PyAV (install `decord` for faster random access).
|
| 136 |
|
| 137 |
-
##
|
| 138 |
|
| 139 |
-
Recordings **up to and including 2026-06-18**
|
| 140 |
lag of **≈15 frames (~0.5 s)**: the tactile stream at index `i` was physically
|
| 141 |
captured ~15 frames *before* the camera/pose at the same index. Cause: a
|
| 142 |
recording-side `cv2.VideoCapture` V4L2 buffer that was never flushed
|
|
@@ -144,7 +144,7 @@ recording-side `cv2.VideoCapture` V4L2 buffer that was never flushed
|
|
| 144 |
on 2026-06-27; **future recordings will not have this lag**.
|
| 145 |
|
| 146 |
The streams are stored frame-aligned by tick index, so this lag is baked in but
|
| 147 |
-
**
|
| 148 |
|
| 149 |
```python
|
| 150 |
ds = ReactVideoDataset("data/motherboard", tactile_latency=15) # pairs view[i] with tactile[i+15]
|
|
|
|
| 134 |
```
|
| 135 |
`mode="segment"` iterates clean spans (no bad frames by construction); `mode="window"` slides over whole episodes and skips `bad_frames.json` intervals. Backend: PyAV (install `decord` for faster random access).
|
| 136 |
|
| 137 |
+
## ✅ Tactile latency corrected (was ~15 frames)
|
| 138 |
|
| 139 |
+
Recordings **up to and including 2026-06-18** HAD a GelSight-vs-camera capture
|
| 140 |
lag of **≈15 frames (~0.5 s)**: the tactile stream at index `i` was physically
|
| 141 |
captured ~15 frames *before* the camera/pose at the same index. Cause: a
|
| 142 |
recording-side `cv2.VideoCapture` V4L2 buffer that was never flushed
|
|
|
|
| 144 |
on 2026-06-27; **future recordings will not have this lag**.
|
| 145 |
|
| 146 |
The streams are stored frame-aligned by tick index, so this lag is baked in but
|
| 147 |
+
**now corrected in the published data** (tactile shifted +15f, rebuilt from raw H5). No loader flag needed. The loader still accepts `tactile_latency=` for raw data:
|
| 148 |
|
| 149 |
```python
|
| 150 |
ds = ReactVideoDataset("data/motherboard", tactile_latency=15) # pairs view[i] with tactile[i+15]
|
tasks.json
CHANGED
|
@@ -97,6 +97,9 @@
|
|
| 97 |
"fixed_in_rig": "2026-06-27",
|
| 98 |
"cause": "recording-side cv2.VideoCapture V4L2 buffer never flushed (throttled reads + no BUFFERSIZE=1 + default pixfmt)",
|
| 99 |
"compensation": "ReactVideoDataset(tactile_latency=15) pairs view[i] with tactile[i+15]; shifts tactile videos + contact scalars only",
|
| 100 |
-
"note": "
|
|
|
|
|
|
|
|
|
|
| 101 |
}
|
| 102 |
}
|
|
|
|
| 97 |
"fixed_in_rig": "2026-06-27",
|
| 98 |
"cause": "recording-side cv2.VideoCapture V4L2 buffer never flushed (throttled reads + no BUFFERSIZE=1 + default pixfmt)",
|
| 99 |
"compensation": "ReactVideoDataset(tactile_latency=15) pairs view[i] with tactile[i+15]; shifts tactile videos + contact scalars only",
|
| 100 |
+
"note": "Recordings up to 2026-06-18 had a ~15-frame GelSight buffer lag; it has been baked out of the published tactile videos + contact scalars. No loader compensation needed. Rig fixed 2026-06-27.",
|
| 101 |
+
"status": "CORRECTED_IN_DATA",
|
| 102 |
+
"frames_shifted": 15,
|
| 103 |
+
"method": "tactile streams shifted +15 frames (rebuilt from raw H5) so tactile[i] aligns with view[i]/pose[i]"
|
| 104 |
}
|
| 105 |
}
|