Add wiki figures to website, minor changes associated with code cleanup

decargroup · Nov 22, 2024 · 4d44f76 · 4d44f76
1 parent b205dce
commit 4d44f76
Show file tree

Hide file tree

Showing 17 changed files with 22 additions and 23 deletions.
diff --git a/assets/anchor_constellation.jpg b/assets/anchor_constellation.jpg
diff --git a/assets/apriltag_det.png b/assets/apriltag_det.png
diff --git a/assets/banner_image.jpg b/assets/banner_image.jpg
diff --git a/assets/decar_logo.png b/assets/decar_logo.png
diff --git a/assets/ifo.jpg b/assets/ifo.jpg
diff --git a/assets/lazy_classifier_results.png b/assets/lazy_classifier_results.png
diff --git a/assets/setup.jpg b/assets/setup.jpg
diff --git a/assets/table.jpg b/assets/table.jpg
diff --git a/assets/trajectories.jpg b/assets/trajectories.jpg
diff --git a/docs/data.md b/docs/data.md
@@ -7,22 +7,23 @@ usemathjax: true
 # Experimental Setup
 ## Uvify IFO-S
 Each Uvify IFO-S quadcopter is equipped with an IMU, a front facing Intel RealSense d435i camera, an integrated downward facing camera, and two UWB transceivers, depicted below. The onboard flight computer is an NVIDIA Jetson Nano running PX4 autopiloting software.
-![](https://github.com/ndahdah/miluv_wiki/blob/main/doc/_static/ifo.jpg)
+![](https://decargroup.github.io/miluv/assets/ifo.jpg)
 
 ## Flight Arena
 The UAVs operated within an approximate 4m x 4m x 3m subsection of an enclosed flight arena. The arena comprises 12 motion capture cameras and six anchors with UWB transceivers. The experimental setup is illustrated below
-![](https://github.com/ndahdah/miluv_wiki/blob/main/doc/_static/setup.jpg)
+![](https://decargroup.github.io/miluv/assets/setup.png)
 
 The dataset includes experiments with three different anchor constellations, shown below
-![](https://github.com/ndahdah/miluv_wiki/blob/main/doc/_static/anchor_constellation.jpg)
+![](https://decargroup.github.io/miluv/assets/anchor_constellation.jpg)
 
 The primary constellation consists of anchors at varying heights, evenly spaced around the UAVs' operating area. The second constellation consists of anchors at the same positions as the primary constellation, but with each transceiver at the same height. The third constellation consists of three clusters of two anchors at varied heights. The location of each anchor is determined using the motion capture cameras. For experiments with obstacles, wood, plastic, and foam were placed in front of the UWB tags in order to disrupt line-of-sight to the UAVs' UWB transceivers.
 
 # Summary of Experiments
-![](https://github.com/ndahdah/miluv_wiki/blob/main/doc/_static/table.jpg)
+
+![](https://decargroup.github.io/miluv/assets/table.jpg)
 
 # Trajectories
-![](https://github.com/ndahdah/miluv_wiki/blob/main/doc/_static/trajectories.jpg)
+![](https://decargroup.github.io/miluv/assets/trajectories.jpg)
 
 <!---
 # Trajectory Videos

diff --git a/docs/examples/apriltag.md b/docs/examples/apriltag.md
@@ -13,7 +13,7 @@ $ python examples/ex_detect_apriltags.py
 in the repository's root directory.
 
 Note that users can safely ignore the following: "warning: count < 8 :(". Once the first image appears on-screen, the user may press any key other than 'escape' to advance the processed image stream forward in time or press 'escape' to end the image stream. A sample image from this example is show below.
-![](https://github.com/ndahdah/miluv_wiki/blob/main/doc/_static/apriltag_det.png)
+![](https://decargroup.github.io/miluv/assets/apriltag_det.png)
 
 ## More details
 This example demonstrates how to load images from the MILUV dataset, as shown below,

diff --git a/docs/examples/ekf/se23_one_robot.md b/docs/examples/ekf/se23_one_robot.md
@@ -28,14 +28,14 @@ We follow the same notation convention mentioned in the paper and assume the sam
 
 ## Importing Libraries and MILUV Utilities
 
-We start by importing the necessary libraries and utilities for this example as in the VINS example, with the only change being the EKF model we are using and the fact that we do not need to import `miluv.utils` separately as we do not need to process VINS data.
+We start by importing the necessary libraries and utilities for this example as in the VINS example, with the only change being the EKF model we are using.
 
 ```py
 import numpy as np
 import pandas as pd
 
 from miluv.data import DataLoader
-import utils.liegroups as liegroups
+import miluv.utils as utils
 import examples.ekfutils.imu_one_robot_models as model
 import examples.ekfutils.common as common
 ```
@@ -68,10 +68,10 @@ accel: pd.DataFrame = imu_at_query_timestamps["imu_px4"][["timestamp", "linear_a
 gyro: pd.DataFrame = imu_at_query_timestamps["imu_px4"][["timestamp", "angular_velocity.x", "angular_velocity.y", "angular_velocity.z"]]
 ```
 
-To be able to evaluate our EKF, we extract the ground truth pose data at these timestamps. The `DataLoader` class provides interpolated splines for the ground truth pose data, and the reason for that is that we can query the ground truth data at any timestamp and call the `derivative` method to get higher-order derivatives of the pose data. For example, here we use the first derivative of the pose data to get the linear velocity data, which is necessary to evaluate our $SE_2(3)$ EKF. We use a helper function from the `liegroups` module to convert the mocap pose data and its derivatives to a list of $SE_2(3)$ poses.
+To be able to evaluate our EKF, we extract the ground truth pose data at these timestamps. The `DataLoader` class provides interpolated splines for the ground truth pose data, and the reason for that is that we can query the ground truth data at any timestamp and call the `derivative` method to get higher-order derivatives of the pose data. For example, here we use the first derivative of the pose data to get the linear velocity data, which is necessary to evaluate our $SE_2(3)$ EKF. We use a helper function from the `utils` module to convert the mocap pose data and its derivatives to a list of $SE_2(3)$ poses.
 
 ```py
-gt_se23 = liegroups.get_se23_poses(
+gt_se23 = utils.get_se23_poses(
     data["mocap_quat"](query_timestamps), data["mocap_pos"].derivative(nu=1)(query_timestamps), data["mocap_pos"](query_timestamps)
 )
 ```

diff --git a/docs/examples/ekf/se23_three_robot.md b/docs/examples/ekf/se23_three_robot.md
@@ -20,7 +20,7 @@ import numpy as np
 import pandas as pd
 
 from miluv.data import DataLoader
-import utils.liegroups as liegroups
+import miluv.utils as utils
 import examples.ekfutils.imu_three_robots_models as model
 import examples.ekfutils.common as common
 ```
@@ -67,7 +67,7 @@ Lastly, we extract the ground truth poses and biases for all three robots at the
 
 ```py
 gt_se23 = {
-    robot: liegroups.get_se23_poses(
+    robot: utils.get_se23_poses(
         data[robot]["mocap_quat"](query_timestamps), data[robot]["mocap_pos"].derivative(nu=1)(query_timestamps), data[robot]["mocap_pos"](query_timestamps)
     )
     for robot in data.keys()

diff --git a/docs/examples/ekf/se3_one_robot.md b/docs/examples/ekf/se3_one_robot.md
@@ -42,10 +42,9 @@ We then import the `DataLoader` class from the `miluv` package, which provides a
 from miluv.data import DataLoader
 ```
 
-We also import the `liegroups` and `utils` libraries from the `miluv` package, which provide utilities for Lie groups that accompany and other helper functions. 
+We also import the `utils` module from the `miluv` package, which provide utilities for Lie groups that accompany and other helper functions. 
 
 ```py
-import utils.liegroups as liegroups
 import miluv.utils as utils
 ```
 
@@ -96,10 +95,10 @@ gyro: pd.DataFrame = imu_at_query_timestamps["imu_px4"][["timestamp", "angular_v
 vins_at_query_timestamps = utils.zero_order_hold(query_timestamps, vins)
 ```
 
-To be able to evaluate our EKF, we extract the ground truth pose data at these timestamps. The `DataLoader` class provides interpolated splines for the ground truth pose data, which we can use to get the ground truth poses at the query timestamps. We use a helper function from the `liegroups` module to convert the mocap pose data to a list of $SE(3)$ poses.
+To be able to evaluate our EKF, we extract the ground truth pose data at these timestamps. The `DataLoader` class provides interpolated splines for the ground truth pose data, which we can use to get the ground truth poses at the query timestamps. We use a helper function from the `utils` module to convert the mocap pose data to a list of $SE(3)$ poses.
 
 ```py
-gt_se3 = liegroups.get_se3_poses(
+gt_se3 = utils.get_se3_poses(
     data["mocap_quat"](query_timestamps), data["mocap_pos"](query_timestamps)
 )
 ```

diff --git a/docs/examples/ekf/se3_three_robot.md b/docs/examples/ekf/se3_three_robot.md
@@ -28,7 +28,6 @@ import numpy as np
 import pandas as pd
 
 from miluv.data import DataLoader
-import utils.liegroups as liegroups
 import miluv.utils as utils
 import examples.ekfutils.vins_one_robot_models as model
 import examples.ekfutils.common as common
@@ -94,7 +93,7 @@ Then, the ground truth data,
 
 ```py
 gt_se3 = {
-    robot: liegroups.get_se3_poses(data[robot]["mocap_quat"](query_timestamps), data[robot]["mocap_pos"](query_timestamps)) 
+    robot: utils.get_se3_poses(data[robot]["mocap_quat"](query_timestamps), data[robot]["mocap_pos"](query_timestamps)) 
     for robot in data.keys()
 }
 ```

diff --git a/docs/examples/losclassification.md b/docs/examples/losclassification.md
@@ -13,7 +13,7 @@ $ python examples/ex_los_nlos_classification.py
 in the repository's root directory.
 
 The output from this example is show below.
-![](https://github.com/ndahdah/miluv_wiki/blob/main/doc/_static/lazy_classifier_results.png)
+![](https://decargroup.github.io/miluv/assets/lazy_classifier_results.png)
 
 ## More details
 This example demonstrates how to use CIR data and how to set up MILUV data for machine learning purposes as shown below,

diff --git a/docs/examples/visualizeimu.md b/docs/examples/visualizeimu.md
@@ -74,7 +74,7 @@ axs[2].plot(time, gt_gyro[2, :], label="Ground Truth")
 ```
 
 <p align="center">
-<img src="https://decargroup.github.io/miluv/assets/imu/gyro.png" alt="gyro" width="400" class="center"/>
+<img src="https://decargroup.github.io/miluv/assets/imu/gyro.png" alt="gyro" width="600" class="center"/>
 </p>
 
 We then plot the measurement error and the compute ground truth bias for the gyroscope measurements.
@@ -92,7 +92,7 @@ axs[2].plot(time, imu_px4["gyro_bias.z"], label="IMU Bias")
 ```
 
 <p align="center">
-<img src="https://decargroup.github.io/miluv/assets/imu/gyro_bias.png" alt="gyro_bias" width="400" class="center"/>
+<img src="https://decargroup.github.io/miluv/assets/imu/gyro_bias.png" alt="gyro_bias" width="600" class="center"/>
 </p>
 
 Similarly, we plot the accelerometer measurements, the ground truth, the measurement error, and the computed ground truth bias as follows.
@@ -110,7 +110,7 @@ axs[2].plot(time, gt_accelerometer[2, :], label="Ground Truth")
 ```
 
 <p align="center">
-<img src="https://decargroup.github.io/miluv/assets/imu/accel.png" alt="accel" width="400" class="center"/>
+<img src="https://decargroup.github.io/miluv/assets/imu/accel.png" alt="accel" width="600" class="center"/>
 </p>
 
 ```py
@@ -126,5 +126,5 @@ axs[2].plot(time, imu_px4["accel_bias.z"], label="IMU Bias")
 ```
 
 <p align="center">
-<img src="https://decargroup.github.io/miluv/assets/imu/accel_bias.png" alt="accel_bias" width="400" class="center"/>
+<img src="https://decargroup.github.io/miluv/assets/imu/accel_bias.png" alt="accel_bias" width="600" class="center"/>
 </p>