added blogposts

_config.yml

@@ -1,5 +1,5 @@
 # Site Settings
-title: BIGLab
+title: BIG
 tagline: roBot Intelligence Group
 email: 
 description: >
@@ -16,7 +16,7 @@ disqus:
  # Leave shortname blank to disable comments site-wide.
  # Disable comments for any post by adding `comments: false` to that post's YAML Front Matter.
  # shortname: my_disqus_shortname
- shortname: biglab
+ shortname: big
 
 plugins:
  - jekyll-sitemap
@@ -54,7 +54,7 @@ defaults:
  path: ""
  type: "posts"
  values:
- author: "BIGLab"
+ author: "BIG"
  layout: post
  show_sidebar: true
 

_data/navigation.yml

@@ -2,8 +2,8 @@
  link: /research/
 - name: Publications
  link: /publications/
-- name: News
- link: /news/
+# - name: News
+#  link: /news/
 - name: Team
  link: /current-members/
  dropdown:
@@ -13,5 +13,5 @@
  link: /alumni/
 - name: Contact
  link: /contact/
-- name: Openings
- link: /openings/
+# - name: Openings
+#  link: /openings/

_includes/footer.html

@@ -31,7 +31,7 @@
 <footer class="footer">
  <div class="container">
  <div id="blocks">
- <div id="block1"><img src="/img/logos/large.png" alt="BIG Lab Logo" style="width:30%;"></div>
+ <div id="block1"><img src="/img/logos/large.png" alt="BIG Logo" style="width:30%;"></div>
  <div id="block2">
  <center>
  <a class="button" itemprop="twitter" href="https://twitter.com/CMUBigLab" target="_blank">
@@ -42,7 +42,7 @@
  </a>
  <br>
  <br>
- <p class="">&copy; 2024 | BIGLab</a></p>
+ <p class="">&copy; 2024 | BIG</a></p>
  </center>
  </div>
  <div id="block3"><img src="/img/riLogo2019.svg" alt="RI Logo" style="float: right;"></div>

_posts/2023-03-19-SoRTS.md

@@ -11,188 +11,8 @@ hide_hero: true
 permalink: /sorts/
 image: /img/posts/2023-07-01-SoRTS/model.png
 link-new-tab: true
+redirect: https://navars.xyz/sorts/
 ---
-
-<h3> 
-Learned Tree Search for Long-Horizon Social Robot Navigation in Shared Airspace
-</h3>
-
-[Ingrid Navarro](https://navars.xyz) *, [Jay Patrikar](https://www.jaypatrikar.me) *, Joao P. A. Dantas, 
-Rohan Baijal, Ian Higgins, [Sebastian Scherer](https://theairlab.org/team/sebastian/) and 
-[Jean Oh](https://www.cs.cmu.edu/~./jeanoh/) 
-
- *Denotes equal contribution 
-
-<left>
- <a class="button" itemprop="github" href="https://github.com/cmubig/sorts" target="_blank">
- <i class="fab fa-github fa-lg"></i>
- </a>
- <a class="button" itemprop="youtube" href="https://www.youtube.com/watch?v=PBE3O4cW2rI" target="_blank">
- <i class="fab fa-youtube fa-lg"></i>
- </a>
- <a class="button" itemprop="paper" href="https://arxiv.org/pdf/2304.01428.pdf" target="_blank">
- <i class="fas fa-file fa-lg"></i> 
- </a>
-</left>
-
-# Abstract
-
-The fast-growing demand for fully autonomous aerial operations in shared spaces necessitates developing
-trustworthy agents that can safely and seamlessly navigate in crowded, dynamic spaces. In this work, 
-we propose **Social Robot Tree Search (SoRTS)**, an algorithm for the safe navigation of mobile robots in 
-social domains. SoRTS aims to augment existing socially-aware trajectory prediction policies with a
-Monte Carlo Tree Search planner for improved downstream navigation of mobile robots. To evaluate the 
-performance of our method, we choose the use case of social navigation for general aviation. To aid 
-this evaluation, within this work, we also introduce **X-PlaneROS**, a high-fidelity aerial simulator,
-to enable more research in full-scale aerial autonomy. By conducting a user study based on the assessments 
-of **26 FAA-certified pilots**, we show that SoRTS performs comparably to a competent human pilot, 
-significantly outperforming our baseline algorithm. We further complement these results with
-self-play experiments, showcasing our algorithm’s behavior in scenarios with increasing complexity.
-
-<hr>
-
-# Video Overview
-
-<p align="center">
-<iframe width="1280" height="720" src="https://www.youtube.com/embed/PBE3O4cW2rI" frameborder="0" allowfullscreen> </iframe>
-</p>
-
-<hr>
-
-# Method
-
-We present **Social Robot Tree Search (SoRTS)**, an algorithm for safe navigation of mobile robots in 
-social settings. SoRTS aims to augment existing data-driven socially-aware motion prediction models 
-with a Monte Carlo Tree Search (MCTS) planner, enabling long-horizon social navigation. Although the 
-algorithm is designed to be domain-independent, in our work we use general aviation in non-towered 
-airports as our case of study. 
-
-<p align="center">
- <img width="1280" src="/img/posts/2023-07-01-SoRTS/model.png" alt="SoRTS">
-</p>
-
-
-SoRTS' tree search is biased by **social** and a **reference** modules. The **social** module aims to 
-handle *short-horizon agent-to-agent dynamics*. It does so by leveraging socially-aware motion 
-prediction model which was trained offline on trajectory data.
-
-<p align="center">
- <img width="1280" src="/img/posts/2023-07-01-SoRTS/social_module.png" alt="SoRTS">
-</p>
-
-The **reference** module provides a global path to the agent which embodies a navigation 
-guideline *(e.g., an FAA-established flying pattern)* that the agent should stick to whenever there 
-aren't any *social conflicts* to resolve. This module also provides a cost map that represents a
-frequence count map to encourage the agent to visit most-frequent regions. 
-
-<p align="center">
- <img width="1280" src="/img/posts/2023-07-01-SoRTS/reference_module.png" alt="SoRTS">
-</p>
-
-Finally, SoRTS's combines the decision modalities of these two modules via **Monte Carlo Tree Search 
-(MCTS)** to achieve long-horizon navigation by backpropagating most successful outcomes. The figure 
-below shows a *toy-example* of two aircraft merging on a single path, where each can execute three 
-possible motion primitives <sup>1</sup>: forward, speed up, and slow down. MCTS, in its forward simulations, 
-not only prunes branches that lead to future collisions but also uses the social module to choose between 
-cutting-in-front (socially undesirable) and yielding (socially desirable), thereby producing socially 
-compliant and safe behavior.
-
-<p align="center">
- <img width="1280" src="/img/posts/2023-07-01-SoRTS/mcts_module.png" alt="SoRTS">
-</p>
-
-
-# Results
-
-We support our method with two evaluation settings: a user study and self-play experiments. 
-
-### User-Study 
-
-We designed a **within-subject** study with 26 participants, all of which were FAA-certified. 
-Using our flight simulator, [**XPlaneROS**](https://theairlab.org/xplane_ros/), we asked users to 
-land an aircraft at a specified runway while interacting with a second pilot instructed to complete
-the same task. The task required both pilots to coordinate with each other, i.e., being 
-socially-compliant and follow FAA flying patterns. As shown in the figure below, the second pilot
-was either a human, SoRTS or a baseline<sup>2</sup> algorithm. In each trial, the algorithm<sup>3</sup> 
-order was randomly chosen.
-
-<p align="center">
- <img width="600" src="/img/posts/2023-07-01-SoRTS/user_study.png" alt="SoRTS">
-</p>
-
-The study aimed at analyzing the second pilot's **navigational performance** (e.g, ability to follow
-rules, navigate smoothly, etc) and **safety** (e.g, comfort, collision risk, etc) by analyzing how 
-the users perceived the other pilot, but also by studying the trajectory information from each trial.
-
-Overall, we found that users rated SoRTS’ and the human’s competence similarly, while the baseline was 
-perceived as less competent. The figure below shows examples of two experiments. **Left:** Each row shows the 
-resulting trajectories of a user interacting with our human pilot, SoRTS, and the baseline. 
-The top row shows a user that successfully followed the expected guideline; we can see that the 
-baseline did not fly as smoothly as SoRTS, and abruptly cut short close to the goal. The 
-bottom row shows a user that did not follow the expected guideline; our algorithm still 
-managed to navigate properly, whereas the baseline behaved erratically, unsafely crossing 
-over the runway. **Right:** Box-plots showing the *per-algorithm* distribution 
-of the results from the user study: the top ones show the average scores given by the user 
-for navigation performance (a) and safety (b). The bottom ones show the average reference error 
-(a) and loss-of-separation (b) metrics obtained from the trajectory data.
-
-<p align="center">
- <img width="1280" src="/img/posts/2023-07-01-SoRTS/user_study_results.png" alt="SoRTS">
-</p>
-
-
-### Self-Play Experiments
-
-We also performed self-play experiments to further examine the behavior of our algorithm and its 
-baseline. These experiments follow the same design as our user study but we also analyze experiments 
-with 3 and 4 agents in each scenario. The figure below shows animated examples from our simulations. 
-We also provide quantitative evaluations, performed on the resulting trajectories. 
-
-<table align="center">
- <tr>
- <td>Baseline</td>
- <td>SoRTS</td>
- </tr>
- <tr>
- <td><img src="/img/posts/2023-07-01-SoRTS/baseline_2agents.gif" width=400 ></td>
- <td><img src="/img/posts/2023-07-01-SoRTS/sorts_2agents.gif" width=400 ></td>
- </tr>
- <tr>
- <td><img src="/img/posts/2023-07-01-SoRTS/baseline_4agents.gif" width=400 ></td>
- <td><img src="/img/posts/2023-07-01-SoRTS/sorts_4agents.gif" width=400 ></td>
- </tr>
-</table>
-
-
-
-| Num Agents | Algorithms | Success Rate | Timeout | Loss-Of-Separation | Offtrack | Reference Error |
-|:-----:|:-----:|:-----:|:-----:|:-----:|:-----:|:-----:|
-| 2 | Baseline | 0.760 | 0.190 | 0.000 | 0.050 | 19.34 |
-| | SoRTS | 0.950 | 0.040 | 0.000 | 0.010 | 13.69 |
-| 3 | Baseline | 0.747 | 0.218 | 0.000 | 0.067 | 19.34 |
-| | SoRTS | 0.943 | 0.030 | 0.000 | 0.027 | 15.68 |
-| 4 | Baseline | 0.483 | 0.510 | 0.000 | 0.007 | 19.34 | 
-| | SoRTS | 0.932 | 0.055 | 0.000 | 0.013 | 15.68 |
-
-
-<hr>
-
-#### Footnotes 
-
-1. In our method, we actually leverage a set of 252 motion primitives. 
-2. The baseline algorithm uses the same **social** and **reference** components but combines them
-naively by weighting them together. 
-3. We use second pilot and algorithm interchangeably.
-
-
-# BibTeX
-
-```
-@inproceedings{navarro2023learned,
- title={Learned Tree Search for Long-Horizon Social Robot Navigation in Shared Airspace},
- author={Navarro Ingrid, Jay Patrikar, Joao P. A. Dantas, Rohan Baijal, Ian Higgins, Sebastian Scherer and Jean Oh},
- booktitle={Arxiv},
- year={2023}
- }
-
-```
+<script>
+window.location.href = 'https://navars.xyz/sorts/';
+</script>

_posts/2024-04-02-CoFRIDA.md

@@ -0,0 +1,17 @@
+---
+layout: post
+title: "CoFRIDA: Self-Supervised Fine-Tuning for Human-Robot Co-Painting"
+date: 2024-04-02 00:00:00
+categories: research
+author: "Peter Schaldenbrand"
+published: true
+sidebar: false
+hide_hero: true
+permalink: /cofrida/
+image: /img/posts/2024-04-02-CoFRIDA/model.png
+link-new-tab: true
+redirect: https://pschaldenbrand.github.io/cofrida/
+---
+<script>
+window.location.href = 'https://pschaldenbrand.github.io/cofrida/';
+</script>

_posts/2024-04-02-SCoFT.md

@@ -0,0 +1,17 @@
+---
+layout: post
+title: "SCoFT: Self-Contrastive Fine-Tuning for Equitable Image Generation"
+date: 2024-04-02 00:00:00
+categories: research
+author: "Zhixuan Liu"
+published: true
+sidebar: false
+hide_hero: true
+permalink: /scoft/
+image: /img/posts/2024-04-02-SCoFT/model.png
+link-new-tab: true
+redirect: https://ariannaliu.github.io/SCoFT/
+---
+<script>
+window.location.href = 'https://ariannaliu.github.io/SCoFT/';
+</script>

deploy.sh

@@ -4,10 +4,10 @@ DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
 cd $DIR
 rm -rf biglab-temp
 
-git clone [email protected]:castacks/website-dev.git biglab-temp
+git clone [email protected]:cmubig/cmubig.github.io.git biglab-temp
 cd biglab-temp
 last_commit=$(git log -1 --format="%at" | xargs -I{} date -d @{} +%s)
-git clone [email protected]:castacks/castacks.github.io.git _site
+git clone [email protected]:cmubig/cmubig.github.io.git _site
 cd _site
 last_push=$(git log -1 --format="%at" | xargs -I{} date -d @{} +%s)
 if [ $last_commit -ge $last_push ]; then
@@ -20,10 +20,10 @@ if [ $last_commit -ge $last_push ]; then
  git remote add origin [email protected]:cmubig/cmubig.github.io.git
  touch .nojekyll
  touch CNAME
- echo "web.github.io" >> CNAME
+ echo "cmubig.github.io" >> CNAME
  git add -A
  git commit -m "Auto deploy"
- git push origin main -f
+ git push -f origin main
  cd ../../
  rm -rf biglab-temp
  echo "Pushed changes!"

index.md

@@ -8,9 +8,9 @@ hide_footer: false
 hero_height: is-large
 hero_image: ./img/big-homepage_gif.gif
 hero_link: /publications/
-hero_link_text: Our Research
+hero_link_text: Research
 hero_link2: /current-members/
-hero_link_text2: Our Team
+hero_link_text2: Team
 ---
 
 # About Us

news.md

@@ -11,7 +11,7 @@ hide_footer: false
 <table>
  <tr>
  <td> 
-  <a class="twitter-timeline" data-width="600" data-height="600" href="https://twitter.com/CMUBigLab">Tweets by CMU BIG Lab</a>
+  <a class="twitter-timeline" data-width="600" data-height="600" href="https://twitter.com/CMUBigLab">Tweets by CMU BIG</a>
   <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>
  </td>
  <td>