UGV - Frontier Exploration

Project Overview

This project presents a scalable, computationally efficient approach for incremental frontier detection in autonomous unmanned ground vehicles (UGVs), developed at DRDO Center for AI & Robotics. It advances exploration algorithms by introducing Safe Frontier Detection (SFD), which considers only safe and reachable frontiers using locally updated map data, improving execution time and reliability compared to traditional methods. The solution was experimentally validated in real-world-like environments with complex, narrow openings, demonstrating significant improvements in exploration speed, accuracy, and system robustness. The approach is designed for seamless integration with ROS and modern robotic platforms.

More Images

Key Contributions & Details

• Designed and implemented a novel Safe Frontier Detection (SFD) approach for autonomous exploration in large, complex maps.
• Achieved up to 24% reduction in exploration time and 30-40% increase in mission success rates across various map sizes compared to baseline.
• Incrementally updates frontier and obstacle data, reducing computation and improving real-time performance over traditional exhaustive methods.
• Ensures only safe and reachable frontiers are selected, preventing exploration failure and robot entrapment in narrow or unsafe regions.
• Integrated SFD with ROS ecosystem for seamless deployment on real unmanned ground vehicles.