diff --git a/_sass/custom.scss b/_sass/custom.scss
index 3f7e13fba..57ffce259 100644
--- a/_sass/custom.scss
+++ b/_sass/custom.scss
@@ -1985,7 +1985,7 @@ body.imuchallenge-home .content a.imu-workshop-text-link:hover {
.imu-workshop-partners {
display: grid;
- grid-template-columns: repeat(3, minmax(0, 1fr));
+ grid-template-columns: repeat(4, minmax(0, 1fr));
gap: 1rem;
}
@@ -2018,28 +2018,6 @@ body.imuchallenge-home .content a.imu-workshop-text-link:hover {
object-fit: contain;
}
-.imu-workshop-amazon {
- gap: 0.8rem;
-}
-
-.imu-workshop-amazon span {
- color: var(--imu-home-ink);
- font-size: 1.45rem;
- font-weight: 800;
-}
-
-.imu-workshop-amazon i {
- width: 1px;
- height: 2rem;
- background: var(--imu-home-line);
-}
-
-.imu-workshop-amazon strong {
- color: var(--imu-home-accent) !important;
- font-size: 1.25rem;
- font-weight: 800;
-}
-
.imu-workshop-inline-links {
display: flex;
flex-wrap: wrap;
@@ -2201,6 +2179,10 @@ body.imuchallenge-home .content .imu-workshop-citation-block code {
margin-top: 1.1rem;
}
+ .imu-workshop-partners {
+ grid-template-columns: repeat(2, minmax(0, 1fr));
+ }
+
body.imuchallenge-home .content a.imu-workshop-action {
min-height: 42px;
padding: 0.6rem 0.7rem;
diff --git a/img/logos/amazon-far.png b/img/logos/amazon-far.png
new file mode 100644
index 000000000..d466c7923
Binary files /dev/null and b/img/logos/amazon-far.png differ
diff --git a/img/logos/sii-logo.png b/img/logos/sii-logo.png
new file mode 100644
index 000000000..4d1af130b
Binary files /dev/null and b/img/logos/sii-logo.png differ
diff --git a/img/social/wechat_communities_2026.png b/img/social/wechat_communities_2026.png
index 3785f2978..3436d05a2 100644
Binary files a/img/social/wechat_communities_2026.png and b/img/social/wechat_communities_2026.png differ
diff --git a/imuchallenge/about.md b/imuchallenge/about.md
index 1b3519ed6..35a65a7ac 100644
--- a/imuchallenge/about.md
+++ b/imuchallenge/about.md
@@ -43,4 +43,4 @@ Challenge organization is coordinated by **CMU AirLab**, with benchmark, dataset
## Contact
-For challenge questions, post in the [Kaggle discussion forum](https://www.kaggle.com/competitions/tartan-imu-challenge-iros2026/discussion){:target="_blank" rel="noopener"} so answers remain visible to every team. Use the [team registration form](https://forms.gle/7rUg7ENctuBU9kTN8){:target="_blank" rel="noopener"} for registration details.
+For challenge questions, post in the [Kaggle discussion forum](https://www.kaggle.com/competitions/tartan-imu-challenge-iros2026/discussion){:target="_blank" rel="noopener"} so answers remain visible to every team. To register, join the competition on Kaggle and create your team there — see the [setup guide](/imuchallenge/setup/#quick-start) for the full walkthrough.
diff --git a/imuchallenge/data.md b/imuchallenge/data.md
index 976b07f11..aa32f53b6 100644
--- a/imuchallenge/data.md
+++ b/imuchallenge/data.md
@@ -38,7 +38,7 @@ Window indices and per-window targets are in `index/`: `train_windows.csv` / `va
At competition scale, the dataset totals **136,289 one-second windows** across all platforms — **81,931 train / 23,714 val / 30,644 held-out test** — for about **37.9 hours** of recorded motion in total, ranging from about 5.2 hours (drone) to 13.4 hours (human) of motion per platform across all three splits.
-**Metric — macro-averaged Absolute Trajectory Error (ATE), lower is better:** you submit one body-frame velocity per test window. The organizers rotate each prediction into the world frame using the ground-truth orientation (used only for scoring, never as model input), accumulate the per-window displacements into a path, align that path to ground truth with an SE(3) Umeyama alignment (rotation + translation, no scale), and compute ATE as the RMS position error between the aligned estimate and ground truth. Per-platform ATE is the mean over that platform's test trajectories; the final score is the equal-weight mean of the four per-platform ATEs, so no platform dominates.
+**Metric — TartanIMU Score `0.6 × (AVE / 0.7356384388) + 0.4 × (ATE20 / 3.1160277267)`, lower is better:** you submit one body-frame velocity per test window. AVE (60% of the score) is the mean per-window Euclidean error `‖v_pred − v_gt‖` between your predicted and ground-truth body-frame velocity, in m/s — no alignment or trajectory accumulation involved. ATE20 (40% of the score) is a trajectory-level metric: the organizers rotate each prediction into the world frame using the ground-truth orientation (used only for scoring, never as model input), accumulate the per-window displacements into a path, cut the ground-truth path into ~20 m segments of travelled distance, align each segment independently to ground truth with an SE(3) Umeyama alignment (rotation + translation, no scale), and take the RMS position error within each segment; a trajectory's ATE20 is the mean over its segments. The score is macro-averaged over the four platforms, each contributing 25%.
Splits are deduplicated at the trajectory level (SHA-256 of raw IMU content); train / val / public-test / private-test share no recording. The public leaderboard is scored on the Public test trajectories; final standings use the held-out Private trajectories. The Public/Private split is at the whole-trajectory level, never per window.
diff --git a/imuchallenge/index.md b/imuchallenge/index.md
index f9f956fef..4dd905a75 100644
--- a/imuchallenge/index.md
+++ b/imuchallenge/index.md
@@ -182,9 +182,6 @@ permalink: /imuchallenge/
TartanIMU Code
-
- Register Team
- Setup Guide
@@ -200,8 +197,8 @@ permalink: /imuchallenge/
Important reminder: eligibility for the final ranking
This applies to every team, including those who have already submitted to Kaggle. The Kaggle leaderboard is provisional: official final standings are decided only after the organizers re-execute the top teams' submitted models. To be eligible for the final ranking you must, in addition to submitting predictions:
-
Complete the Submission Form for each submission you want considered, attaching the prediction CSV you uploaded.
-
Submit a technical report (PDF). Download it here — the zip contains main.tex and ieeeconf.cls. On Overleaf: New Project → Upload Project → select the zip. Maximum 6 pages excluding references, 7 in total; the appendix does not count. The report is due 23:59 US Eastern Time, 23 September 2026 — three days after the leaderboard closes — and is submitted by attaching it at the end of the challenge Form. All numbers in Tables III–V must come from the official scoring service, not a local re-implementation.
+
Complete the Submission Form for each submission you want considered, attaching the prediction CSV you uploaded.
+
Submit a technical report (PDF). Download it here — the zip contains main.tex and ieeeconf.cls. On Overleaf: New Project → Upload Project → select the zip. Maximum 6 pages excluding references, 7 in total; the appendix does not count. The report is due 23:59 US Eastern Time, 23 September 2026 — three days after the leaderboard closes — and is submitted by attaching it at the end of the challenge Form. All numbers in Tables III–V must come from the official scoring service, not a local re-implementation.
Publish your single unified model's weights and inference code as a public Hugging Face model repository, within the same 7 days.
Teams missing any of the three, or whose model cannot be re-executed, are removed from the final ranking regardless of leaderboard position. Full details on the Rules and Evaluation tabs.
@@ -358,19 +355,22 @@ permalink: /imuchallenge/
Organizers and partners
Built by teams working on robust robot perception.
-
The challenge is organized by CMU AirLab and Amazon FAR in conjunction with the IROS 2026 workshop on interoceptive perception.
+
The challenge is organized by CMU AirLab and Amazon FAR in conjunction with the IROS 2026 workshop on interoceptive perception, with support from the Shanghai Innovation Institute.
-**Register your team.** [Fill out the registration form.](https://forms.gle/7rUg7ENctuBU9kTN8){:target="_blank" rel="noopener"} Then join the competition on Kaggle: [TartanIMU Challenge: Multi-Platform Inertial Odometry](https://www.kaggle.com/competitions/tartan-imu-challenge-iros2026){:target="_blank" rel="noopener"}.
+**Register your team.** Join the competition on Kaggle and create your team there: [TartanIMU Challenge: Multi-Platform Inertial Odometry](https://www.kaggle.com/competitions/tartan-imu-challenge-iros2026){:target="_blank" rel="noopener"}. There's no separate signup — your Kaggle team *is* your registration.
@@ -225,6 +225,10 @@ finetune:
**Improve and [submit](https://www.kaggle.com/competitions/tartan-imu-challenge-iros2026/submissions){:target="_blank" rel="noopener"}.** Experiment with your own methods to push performance on the leaderboard.
+
+**Log your submission and report.** The Kaggle leaderboard alone doesn't make you eligible for the final ranking or workshop presentation. Complete the [Submission Form](https://docs.google.com/forms/d/e/1FAIpQLSc_3Pq2Bs4ytd0lfHHLsU_qoO5z0BP6Psnsqo7e5nOyBTlCjQ/viewform){:target="_blank" rel="noopener"} for each submission you want considered, and again to attach your technical report before the deadline. See [Final ranking eligibility](#schedule-rules-and-leaderboard) for the full requirements.
+
+
## Schedule, Rules, and Leaderboard
@@ -241,7 +245,7 @@ Competition dates, team and submission limits, and any external-data policy can
The Kaggle leaderboard is **provisional**. Official final standings are decided only after the organizers re-execute the top teams' submitted models. To be eligible for the final ranking you must, in addition to submitting predictions:
-1. Complete the [Submission Form](https://forms.gle/7rUg7ENctuBU9kTN8){:target="_blank" rel="noopener"} for each submission you want considered, attaching the prediction CSV you uploaded.
+1. Complete the [Submission Form](https://docs.google.com/forms/d/e/1FAIpQLSc_3Pq2Bs4ytd0lfHHLsU_qoO5z0BP6Psnsqo7e5nOyBTlCjQ/viewform){:target="_blank" rel="noopener"} for each submission you want considered, attaching the prediction CSV you uploaded.
2. Submit a technical report (PDF) — see [Technical report](#technical-report) below. All numbers in Tables III–V must come from the [official scoring service](https://huggingface.co/spaces/Tartan-IMU/imu_odometry_challenge_scoring){:target="_blank" rel="noopener"}, not a local re-implementation.
3. Publish your single unified model's weights and inference code as a public Hugging Face model repository, within the same 7 days.
diff --git a/index.md b/index.md
index d99cba40b..0ea8761ac 100644
--- a/index.md
+++ b/index.md
@@ -80,10 +80,6 @@ hero_buttons:
font-size: 0.92rem;
line-height: 1.5;
}
- #home-wechat-qr[hidden],
- #home-wechat-expired[hidden] {
- display: none;
- }
#home-wechat-qr img {
display: block;
width: min(100%, 660px);
@@ -156,20 +152,20 @@ hero_buttons:
Join Our WeChat Communities
-
Scan to join the 2026 IMU Challenge or Proprioception community.
+
Scan to join the Proprioception community.
-
+
- These invitation QR codes are valid until August 12, 2026.
+ This invitation QR code is valid until September 12, 2026.
- 
+
+
+ 
+
+