SoftRafts: Floating and Adaptive Soft Modular Robots

🚀 Overview

SoftRafts is a modular soft robotic system that integrates rigid 3D-printed components, soft foam, a cable-driven actuation mechanism, and a propeller for aquatic mobility. Using permanent magnets for fast inter-module connections, each unit can bend, steer, connect, and operate collectively in a swarm. It serves multiple functions, such as:

• Acting as a gripper to retrieve debris from water
• Assembling into a floating raft for drone landing
• Forming a snake-like chain that transitions between land and water
• Working collaboratively to transport payloads

This platform demonstrates how soft deformation and modularity enable adaptive, multifunctional robotics for real-world aquatic environments.

---

📁 Repository Structure

SoftRafts/
│
├── CAD/                     # STL models for 3D printing
│   └── softraft.png         # Render of full assembled raft
│
├── firmware/                # ESP32-S2 microcontroller firmware
│   └── firmware_softrafts.ino
│
├── software/                # Control, analysis, and visualization notebooks
│   ├── SoftRafts-control.ipynb
│   ├── snake--curve.ipynb
│   └── gaits_data_processing.ipynb
│
├── raw_gaits_data/          # Raw positional and orientational data (6 gaits)
│   ├── forward_data.json
│   ├── backward_data.json
│   ├── left_front_data.json
│   ├── right_front_data.json
│   ├── left_back_data.json
│   ├── right_back_data.json
│   └── README.md
│
├── figures/                 # Auto-generated plots (e.g., Figure 8)
│   └── plot_comparison.png
│
├── LICENSE
└── README.md

---

🧱 CAD Models

The CAD/ folder contains 3D-printable STL models for all robot parts:

• Whole_SoftRaft.stl — complete raft assembly (for reference)
• Board.stl, Middle.stl, Tail.stl, Head.stl — core structural modules
• Motor.stl, Motor Cover.stl — motor housing components
• Winch.stl — for cable-driven actuation
• Cover.stl — protective enclosure for electronics
• softraft.png — visual rendering of the assembled robot

---

🔧 Firmware

Located in the firmware/ folder, firmware_softrafts.ino is the control firmware for the ESP32-S2 microcontroller.

Features:

• Wi-Fi web server interface for motor control (port 80)
• OTA (Over-the-Air) firmware updates via ArduinoOTA
• Encoder-based position feedback and calibration
• REST-style HTTP commands for live control (/set, /status, /pause, etc.)

Upload via the Arduino IDE with the ESP32-S2 board selected.

---

🧠 Software

The software/ folder contains all Python-based notebooks for robot control, gait data analysis, and visualization.

Included Notebooks:

• SoftRafts-control.ipynb — Wi-Fi control interface for the SoftRaft robot
• snake--curve.ipynb — Geometry and curvature visualization of chained modules
• gaits_data_processing.ipynb — Reproducible gait data processing and analysis notebook for Figure 8

The analysis notebook:

• Loads raw data from raw_gaits_data/
• Performs coordinate transformation to the local reference frame
• Removes outliers and computes cluster centers
• Generates the comparison figure automatically (figures/plot_comparison.png)

Requires Python ≥ 3.7 with numpy, matplotlib, requests, pandas, and notebook.

---

📊 Raw Data

The raw_gaits_data/ folder contains six JSON files corresponding to different gaits
(forward, backward, left-front, right-front, left-back, right-back).
Each file contains 35 recorded trials of positional and orientational data (in cm and degrees).

A detailed README.md in that folder explains:

• The meaning and units of each field
• How to reproduce Figure 8 using the analysis notebook in software/

---

⚙️ Setup Instructions

1. Clone the repository:

   git clone https://github.com/SMILE-Robotics-Lab/SoftRafts.git
   cd SoftRafts/software

2. Create and activate a virtual environment:

   python3 -m venv venv
   source venv/bin/activate

3. Install dependencies:

   pip install numpy matplotlib pandas requests notebook

4. Launch Jupyter:

   jupyter notebook

---

📬 Contact

For questions, collaboration, or reproducibility assistance:

📧 Luyang Zhao — [email protected]
📧 Yitao Jiang — [email protected]

---

📄 License

This project is licensed under the MIT License.