Semi-Autonomous Robot for Post Disaster Human Locator and Emergency Response

Abstract

Search and Rescue operations (SAR) have become a significant area of concern in disaster scenarios, when quick action is vital for saving lives. This study illustrates the development of an edge-AI, semi-autonomous robot, designed to assist first responders by delivering clear and actionable information for effective decision making in disaster-stricken environments. By utilizing a modified rocker-bogie mechanism, the developed mechanical structure is capable of navigating obstacle-filled and rough terrain areas. A victim detection process, through advanced discrete viewpoint sampling, has been developed by utilizing a combination of AI assisted scanning & movement, and full autonomous scan modes, trained using YOLOv8n and YOLO11x-pose vision AI models on the COCO 2017 dataset. Through the ESP-NOW and Miracast wireless communication protocols, the robot maintains a data connection with the remote controller for real-time data transmission and control without dependence on external networks. All the detection and sensor data are transmitted to the remote controller unit, providing real-time updates on whether the environment is safe for entry to first responders. Extensive testing of the system has demonstrated reliable performance and effective output, confirming its ability to operate accurately in simulated post-disaster environments and respond as intended under various conditions.