Real-Time Object Detection and Navigation in Harsh Environments Using YOLOv7 and IoT Sensors

Navigating and detecting obstacles accurately in harsh environments like mines, disaster zones, or remote areas poses a significant challenges due to low visibility, rugged terrain, and unreliable connectivity. This paper introduces a solution to tackle these issues by combining Ai/MI with IoT. However, YOLOv7 is proposed for real-time object detection, accurately in challenging conditions such as dust, fog, and smoke. Also, IoT sensors, including LiDAR and Orbbec cameras, that collect environmental data. The system processes this information locally to ensure real-time decision-making without relying on external servers. For data transmission in areas with little to no connectivity, the system uses LoRa technology, enabling long-range and energy-efficient data transmission. However, it is challenging to accurately detect and classify object in such scenarios. In order to deal with this, camera has been used which provides three different kind of format image such as RGB, Infra Red (IR) and Thermal image. The proposed scheme implies YOLOv7 in which features from RGB, Infra Red (IR) and Thermal images are considered for decision making and subsequent accurate object detection. This approach has proven to enhance safety, reliability, and efficiency across industries such as mining, defense, and search-and - rescue operations.