---

PeraSwarm: Simultaneous Localization and Mapping in Mixed Reality Environment

---

Description

The objective of this project is to use swarm robots to efficiently map and navigate complex environments. These autonomous robots work together in a coordinated fashion, leveraging swarm intelligence to create detailed maps, contributing to advancements in autonomous exploration and mapping technology.

Overview

PeraSwarm is a project that explores the field of swarm robotics, where multiple robots collaborate to achieve tasks in a decentralized and distributed manner. The project focuses on the exploration behavior of swarm robotics in unknown environments, aiming to develop efficient mapping strategies for autonomous mobile robots.

Key Features

• Integration of physical and virtual robots in a mixed reality environment
• Cost-effective sensor implementation in multi-robot systems
• Decentralized communication approach for enhanced robustness and scalability
• Development of new algorithms for improved efficiency and effectiveness

Methodology

1. Physical Robots: Low-cost robots with differential drive, sensors (distance, accelerometer, gyroscope, magnetometer), and modular C++ firmware design.
2. Virtual Robots: Java-based virtual robot simulations for scalability and platform independence.
3. Visualizer: Techniques for visualizing robot behavior in both virtual and augmented reality environments.
4. Simulator: A real-time integration framework for inter-reality communication, localization, and mapping.
5. Occupancy Grid Mapping (OGM): Explicit environment modeling approach for robustness and scalability.

Experiments and Results

• Tested algorithms: Random Movement, Heuristic Approach, Algorithm Based on Nearest Unexplored Cell, and Voronoi Coverage.
• Performance metrics: Time of full coverage, probability of success, and accuracy of exploration using ground truth comparison.
• Results showed the Algorithm Based on Nearest Unexplored Cell outperformed others in terms of faster exploration times and higher success probabilities.

Future Work

• Explore advanced cooperative localization methods
• Integrate machine learning techniques for improved mapping and localization
• Address challenges in complex sensor usage, decentralized mapping, scalability, and robustness to dynamic environments.

Team Members

1. E/18/077 - Nipun Dharmarathne, Website, Email
2. E/18/150 - Yojith Jayarathna, Website, Email
3. E/18/227 - Dinuka Mudalige, Website, Email

Supervisors

1. Prof. Roshan Ragel, Website, Email
2. Dr. Isuru Nawinne, Website, Email
3. Mr. Nuwan Jaliyagoda, Website, Email

Links

1. Project Repository
2. Project Page
3. Pera Swarm - GitHub Organization
4. Pera Swarm - Website
5. Department of Computer Engineering
6. University of Peradeniya