Design and Development of a Modular Snake Robot Digital-Twin Simulator for learning multi-terrain locomotion

Background and Goals of the Project

Snake robots have gained significant attention in recent years due to their ability to traverse complex and challenging terrains with unparalleled flexibility and maneuverability. These robotic systems, inspired by the locomotion of real snakes, offer unique advantages in various applications such as search and rescue missions, exploration of hazardous environments, and inspection of confined spaces.

To optimize the design and performance of snake robots, researchers and engineers rely on simulation tools that provide an accurate representation of their locomotion capabilities. Digital-twin simulators have emerged as valuable tools in the field of robotics, enabling the evaluation and refinement of robot designs in a virtual environment before physical implementation. They allow for rapid prototyping, testing of various locomotion strategies, and analysis of system performance under different terrains and operating conditions.

The key features of our digital-twin simulator include a physics-based simulation engine that models the dynamic behaviour of the snake robot in a gazebo-like simulator, realistic terrain models that accurately represent various types of surfaces and a modular design that allows for easy customization and integration of different robot modules. Furthermore, the simulator provides a comprehensive set of performance metrics, enabling quantitative evaluation and comparison of different locomotion strategies.

Please find more information about this master project and the figures referred to in the text in the attachment below.

Goals of the Project:

In this project, we present the design and development of a modular snake robot digital-twin simulator specifically tailored for multi-terrain locomotion. The simulator aims to provide an efficient and realistic platform for evaluating capabilities and training the snake robot to locomote across a wide range of terrains, including multi-friction, rough terrains, inclined surfaces, obstacles, and confined spaces. By accurately simulating the complex interactions between the snake robot and its environment, the simulator enables researchers and engineers to optimize robot design parameters, control algorithms, and motion planning strategies.

• Simulation of Snake robot locomotion in the multi-terrain environment. The simulation can be carried out in Gazebo using ROS.
• Design and development of the frictional skin for effective locomotion on a flat surface. And testing the developed skin in a simulator as well as with a physical snake robot.
• Development of a learning-based controller to train the snake robot locomotion for multi-terrain locomotion.
• Develop a digital twin for a modular snake robot, which combines the physical snake robot and its environment with a virtual simulation. This integration allows for real-time interaction and synchronization between the physical system and its digital counterpart.

Potential applications of snake robots: Snake robots possess significant potential for various real-world applications due to their unique locomotion capabilities and adaptability. Here are some of the key potentials of snake robots. 1. Search and Rescue Operations, 2. Inspection of Confined Spaces, 3. Exploration in Extreme Environments, 4. Medical Applications, 5. Agriculture and Farming. As technology continues to advance, snake robots have the potential to revolutionize industries, improve safety, and overcome challenges in environments where traditional robots struggle to operate effectively.

Steps and methods during project implementation

The project is divided into three verticals where Gazebo, ROS and Physical robot are involved. The overall project framework is depicted in fig. 2, And, the digital-twin framework of the snake robot connecting both the virtual world and the physical world is shown in fig. 3. The methods and implementation of the project are elaborated below:

• Locomotion in the multi-terrain environment:
  Develop a multi-terrain environment in Gazebo to simulate the snake robot using ROS. And observe the performance of the snake robot in different environments and while doing environment transitions.
• Designing a frictional skin: Design and develop an effective anisotropic frictional skin for the snake body to locomote on flat terrain.
• Optimize the locomotion parameters: Based on the locomotion performance in a multi-terrain environment, we need to optimize the locomotion parameters for effective locomotion in and during terrain transition.
• Develop a Deep Reinforcement-based control: In the next step, a DRL-based controller will be developed to locomote the snake robot in a simulated (Gazebo) environment using ROS. The generalized DRL framework for snake robot is shown in fig.4.
• Digital-Twin for multi-terrain snake robot: Integration of simulated robot, and environment with the physical snake robot and test the performance as shown in fig.3. 

Schedule

The schedule of the proposed work is divided into 32 weeks and shown in Figure. 5.

Expected Importance for the Ongoing Ph.D. Project

This project holds immense importance within the ongoing project of the Multi-terrain Modular Selfreconfigurable Snake Robot. The successful implementation Digital-Twin Simulator for multi-terrain locomotion serves as a crucial interface for the testing and validating of different mechanical and control design implementations. It eliminates the need for complex interfaces, and extensive and expensive testing, allowing researchers to seamlessly validate designed control and mechanical systems. The successful implementation of this project within the digital twin-based snake robot simulator project opens up opportunities for a broader research community in various academics and industries. It paves the way for new innovative ideas to test and train snake robots for various environments.

Plan for Dissemination of Results

The project findings will be compiled into a research paper targeting reputable conferences/journals in the field of robotics, Deep-Reinforcement Learning, and Control systems. These papers will detail the methodology, experimental results, and contributions of the project.

Any plan for further applications for external funding

This project is highly relevant to Collaborative robots for Search and Rescue, as it focuses on developing a multi-terrain locomotion-capable snake robot. It enables the robots to reach narrow and untraceable locations during SAR missions. Further, this technology can be applied to Oil and Gas industries for pipe inspection. The project’s significance extends to the Agder Region in Norway, where it fosters innovation, attracts investments, creates job opportunities, and positions the region as a leader in responsible robotics research and development. Therefore, we will target the RFF funding call Regional cooperation project (KSP) 2023. Link: https://www.regionaleforskningsfond.no/fondsregion-agder/rffagder/2023/regionaltsamarbeidsprosjekt-ksp-2023/