Scholarlink

<p>This study presents a miniature autonomous humanoid robotic system that not only performs real-time environmental monitoring and atmospheric water harvesting in extreme and unstructured environments, but also introduces several novel system-level innovations. The robot establishes a new Artificial Intelligence–Internet of Things (AI–IoT)–robot coordination architecture that integrates YOLO-based visual perception, ultrasonic ranging, and environmental sensing into a unified decision-making workflow, enabling multi-modal data fusion for adaptive navigation. A compact ESP32-CAM module combined with a customized YOLO detector achieves a 97% F1-score in target recognition and a 91% success rate in dynamic obstacle avoidance. Furthermore, the proposed system incorporates a micro-scale bioinspired water harvesting module, redesigned for mobile operation, which utilizes 100&#xa0;g of silica gel to collect up to 25 mL of moisture daily under 23&#xa0;°C and 75% relative humidity, and yields up to 77.6&#xa0;L annually when scaled to 1000&#xa0;g with efficiency taken into account. To optimize adsorption performance, this study develops a humidity-driven collection-efficiency model that links real-time sensor inputs with water harvesting predictions and supports path-planning decisions that guide the robot toward high-humidity zones. Environmental parameters—including temperature, humidity, pressure, and volatile organic compounds (VOCs)—are captured by onboard sensors and transmitted to a cloud platform via message queuing telemetry transport (MQTT) and hypertext transfer protocol (HTTP) for real-time visualization, mission adaptation, and autonomous task refinement. These innovations collectively form a new integration workflow that enhances environmental awareness, mobility robustness, and water harvesting efficiency. Experimental validations confirm the feasibility of the system for autonomous deployment in harsh, remote, or post-disaster conditions. Future work will incorporate swarm intelligence to extend multi-robot cooperation and resilience under climate-challenged environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Miniature autonomous humanoid robot for environmental sensing and atmospheric water harvesting using bioinspired materials and AI-based vision

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月26日

查看全文
  • Hwa-Dong Liu,
  • Chen-Wei Su,
  • Chia-Hsun Chang,
  • Cheng-Ze Li,
  • Ping-Jui Lin

摘要

This study presents a miniature autonomous humanoid robotic system that not only performs real-time environmental monitoring and atmospheric water harvesting in extreme and unstructured environments, but also introduces several novel system-level innovations. The robot establishes a new Artificial Intelligence–Internet of Things (AI–IoT)–robot coordination architecture that integrates YOLO-based visual perception, ultrasonic ranging, and environmental sensing into a unified decision-making workflow, enabling multi-modal data fusion for adaptive navigation. A compact ESP32-CAM module combined with a customized YOLO detector achieves a 97% F1-score in target recognition and a 91% success rate in dynamic obstacle avoidance. Furthermore, the proposed system incorporates a micro-scale bioinspired water harvesting module, redesigned for mobile operation, which utilizes 100 g of silica gel to collect up to 25 mL of moisture daily under 23 °C and 75% relative humidity, and yields up to 77.6 L annually when scaled to 1000 g with efficiency taken into account. To optimize adsorption performance, this study develops a humidity-driven collection-efficiency model that links real-time sensor inputs with water harvesting predictions and supports path-planning decisions that guide the robot toward high-humidity zones. Environmental parameters—including temperature, humidity, pressure, and volatile organic compounds (VOCs)—are captured by onboard sensors and transmitted to a cloud platform via message queuing telemetry transport (MQTT) and hypertext transfer protocol (HTTP) for real-time visualization, mission adaptation, and autonomous task refinement. These innovations collectively form a new integration workflow that enhances environmental awareness, mobility robustness, and water harvesting efficiency. Experimental validations confirm the feasibility of the system for autonomous deployment in harsh, remote, or post-disaster conditions. Future work will incorporate swarm intelligence to extend multi-robot cooperation and resilience under climate-challenged environments.