The thesis focuses on decentralized multi-robot navigation, leveraging the Path-Optimized Rapidly-Exploring Random Tree (PO-RRT) algorithm to enhance path planning by minimizing path lengths and improving efficiency. It addresses limitations in traditional path planning algorithms by incorporating optimization techniques that consider robot dynamics and environmental constraints. Furthermore, it tackles the inefficiencies and risks of centralized communication models, such as latency and single points of failure, by adopting a decentralized peer-to-peer communication approach using Wi-Fi Direct. This setup enables robots to communicate directly, facilitating real-time coordination. This thesis integrates the PO-RRT algorithm with advanced communication protocols and sensor systems, including Raspberry Pi controllers, ultrasonic sensors, and LiDAR, for effective navigation in dynamic environments. Simulation results in MATLAB demonstrate the system's ability to generate optimized paths in various scenarios and environments, validating its performance against other algorithms such as RRT*, BI-RRT, and PRM. Comparative analyses highlight PO-RRT's superiority in creating shorter paths, enhancing multi-robot coordination, and reducing computational overhead.