readme finish for 3d printing

Former-commit-id: 9324734e67e417f46e284aa0661098b1517fbcda
ai4ce · Oct 18, 2024 · 2cdd8fe · 2cdd8fe
1 parent 9ca61c8
commit 2cdd8fe
Show file tree

Hide file tree

Showing 3 changed files with 41 additions and 4 deletions.
diff --git a/README.md b/README.md
@@ -1,5 +1,7 @@
-:new: [2024-10-15] *Installation for Everything Robotics Updated. More Documentation Ongoing for VLM Reasoning and Real World Experiment. The README Needs a Lot of Cleaning*
+**More Documentation Ongoing for VLM Reasoning and Real World Experiment. The README Needs a Lot of Cleaning and Update**
 
+:new: [2024-10-17] *Installation for Hardware Integration/3D Printing Updated.*
+:new: [2024-10-15] *Installation for Robotics Software Updated.*
 :new: [2024-10-11] *Made Public*
 # FusionSense
 ### [[Page](https://ai4ce.github.io/FusionSense/)] | [[Paper](https://arxiv.org/abs/2410.08282)] | [[Video](https://youtu.be/thC0PeAQxe0)]
@@ -15,11 +17,12 @@ FusionSense is a novel 3D reconstruction framework that enables robots to fuse p
 ## Preparation 
 
 ### Step 0: Install Everything Robotics
-We utilize a depth camera mounted on a robot arm powered by `ROS2` to acquire pictures with accurate pose information. We also used a tactile sensor for <b>Active Touch Selection</b>.
+We used a depth camera mounted on a robot arm powered by `ROS2` to acquire pictures with accurate pose information. We also used a tactile sensor for <b>Active Touch Selection</b>.
 
 If you have no need for this part, feel free to jump into [Step 1](https://github.com/ai4ce/FusionSense/blob/main/README.md#step-1-install-3d-gaussian-dependencies-and-nerfstudio) for the 3D Gaussian pipeline of <b>Robust Global Shape Representation</b> and <b>Local Geometric Optimization</b>.
 
-For installing everything robotics, please see this [documentation](./instructions/install_robotics.md).
+- For installing robotics software, please see [Robotics Software Installation](./instructions/install_robotics.md). 
+- For hardware integration, please see [3D Printing Instructions](./instructions/3d_printing.md).
 
 ### Step 1: Install 3D Gaussian Dependencies and Nerfstudio
 **Note:** Because our major dependencies, `Nerfstudio` and `Grounded-SAM-2`, officially support two different CUDA version (11.8 vs. 12.1), we will have to create two separate environments. We hope to resolve this in the future when `Nerfstudio` bump its official CUDA support version.

diff --git a/instructions/3d_printing.md b/instructions/3d_printing.md
@@ -3,4 +3,29 @@
 ## Object
 - The `.stl` for printing the Stanford Bunny can be found at `/assets/3d_printing/object/stanford_bunny.stl`.
 - The print file can be found at `/assets/3d_printing/object/stanford_bunny.form`
-- The printing is done on a Formlabs Form 3 SLA 3D printer. The material for the transparent bunny is Clear Resin V4. The material for the black bunny is Black Resin V4.  
+- The printing is done on a Formlabs Form 3 SLA 3D printer. The material used are: 
+    - Transparent bunny: Clear Resin V4
+    - Black Bunny: Black Resin V4
+    - Reflective Bunny: Tough 2000 V1 + Generic reflective paint from Amazon
+- Some tips for printing:
+    1. Try to align the supports at the bottom of the bunny, which is invisible during experiment and won't be touched, because supports will leave some artifacts (bump) on the surface.
+    2. Try not to use the cure station for the transparent bunny, because curing will cause the clear resin to become yellowish.
+    3. For the same reason, we did not cure the black and reflective bunnies as we want to maintain maximum comparability between them. Instead, we use a fan to blow on the three bunnies for 24 hours.
+
+## Camera and Tactile Sensor Mount
+- The `.stl` files for the mount that connect the sensor/camera to the robot are located at:
+    - `/assets/3d_printing/sensor_mount/xArm6_Interface_V2.STL`
+    - `/assets/3d_printing/sensor_mount/xArm6_Sensor_Mount_V2.STL`
+- The design source file from Solidworks is also in the same folder
+- The interface is to connect to the robot EEF flange, while the mount will carry the camera and the sensor. The mount will be insert on top of the interface.
+- The assembly file, which put the models of mount, interface, sensor and camera together, can be found in the `/assets/3d_printing/sensor_mount/assembly` folder.
+    - This can be used for visualization in RViz, or for collision avoidance. For collision avoidance, I recommend downsample the assembly stl to reduce face number. There is a sample in the provided xArm ROS2 package
+- Two M2 screws to connect the GelSight Mini tactile sensor. Two M3 screws to connect the Intel RealSense D405 camera.
+- **Important:** The dimension of the mount is measured in Solidworks and put to the configuration file in the [RealSense interface package](https://github.com/ai4ce/realsense_ROS2_interface) and the [GelSight tactile sensor interface package](https://github.com/ai4ce/gelsight_ROS2_interface), so that we can accurately acquire the pose of camera and the sensor in the world coordinate by adding offsets to the EEF's coordinate.
+    - Therefore, if you modify the parts' dimension, please update the configuration in the ROS2 packages accordingly. Specifically, the `realsense_ROS2_interface/realsense_capture/config/calibration.yaml` and `gelsight_ROS2_interface/gelsight_capture/config/gsmini.yaml`.
+
+## Object Mounting Platform
+- The `.stl` file for the mounting platform that connect the object to the tripod can be found at:
+    - `/assets/3d_printing/tripod_mount/Tripod_Plate_180MM_V2.STL`
+    - `/assets/3d_printing/tripod_mount/Tripod_Plate_180MM_V2_LID.STL`
+- The `Tripod_Plate_180MM_V2.STL` can be inserted into the groove of the Arca-Swiss plate as we mentioned in the Appendix of the Arxiv paper. The `LID` then connect to the former so object can be fastened on this platform.
diff --git a/instructions/install_robotics.md b/instructions/install_robotics.md
@@ -1,11 +1,20 @@
+# Robotics Software Installation
+
+## ROS 2
 1. Install ROS2 Humble according to the [official instruction](https://docs.ros.org/en/humble/Installation.html). 
     - Note that while this project is developed under Humble, we do not explicitly use any Humble-specific feature, so other distro should work in principle.
 
+## RealSense D405 Depth Camera
 2. Install the [RealSense depth camera interface package](https://github.com/ai4ce/realsense_ROS2_interface). This is a custom ROS2 package built upon Intel's official ROS2 wrapper.
 
+## GelSight Mini Tactile Sensor
 3. Install the [GelSight tactile sensor interface package](https://github.com/ai4ce/gelsight_ROS2_interface). 
     - **Unfortunately**, this package is built upon a codebase that has not been open-source. We will update the link when the dependency is fully-open within a few weeks. Stay tuned!
     - Note that this is not the same as the official GelSight implementation. We made some tweaks and (hopefully) improvement, so it's strongly encouraged that you use our own package.
+
+**Note:** See [3D Printing Instructions](3d_printing.md) for some caveats on the dimension of the camera/sensor mount and the world-frame coordinates of the camera and the tactile sensor.
+
+## Robot Servoing and Teleoperation.
 4. Install the [Ufactory xArm6 servoing and teleoperation package](https://github.com/ai4ce/xarm_ros2). This is a custom ROS2 package built upon the official UFactory ROS2 packages.
     - We also have a [UR10e equivalent](https://github.com/ai4ce/ur_ros2) available. 
     - If you are using a different robot, while we may not have a ROS2 package readily available, as long as your robot works with MoveIt 2, you should be able to adapt my code fairly easily.