Development of a Modular Robotic Adapter for Autonomous Navigation Using Jetson Nano

Abstract

AdaptX is an AI-powered autonomous ground robot designed for autonomous navigation, high-resolution 3D environmental mapping, and real-time decision-making. AdaptX, which was created with scalability and adaptability in mind, combines robotic technologies and cutting-edge artificial intelligence to address important issues in autonomous mobility. The system makes use of YOLOv9 for fast object recognition, Simultaneous Localization and Mapping (SLAM) for precise real-time mapping and localization, and a web-based control interface for smooth remote monitoring and communication. AdaptX uses a combination of LiDAR and camera sensors, powered by the NVIDIA Jetson Nano, to provide dependable autonomous navigation and strong environmental awareness. AdaptX\'s plug-and-play hardware architecture is a noteworthy feature that enables quick integration and replacement of sensor modules or computing components without requiring significant software changes. This modular design makes platform adaption, maintenance, and upgrades easier while guaranteeing versatility for a wide range of applications. The system\'s potential for quick deployment in dynamic contexts and across many use cases is further increased by its plug-and-play functionality. This study provides a thorough analysis of the software architecture, system design, and implementation approach of AdaptX. The system\'s ability to autonomously navigate complex and dynamic landscapes, recognize and react to impediments with high precision, and create real-time 3D maps for spatial awareness is demonstrated via performance evaluation in a variety of terrain situations. Results from experiments confirm the robot\'s dependability and efficiency in real-world situations. AdaptX has great potential in a number of fields, such as automated surveillance, smart agriculture, industrial automation, and disaster relief. To further cement AdaptX\'s position as a flexible solution for next-generation autonomous robotic systems, future developments will concentrate on incorporating deep learning models for enhanced contextual comprehension, behavior prediction, and adaptive decision-making.

Introduction

AdaptX is a robust, scalable autonomous ground robot designed for real-time navigation, 3D mapping, and object detection in dynamic environments. It leverages advanced AI, sensor fusion, and real-time decision-making to function across sectors like agriculture, surveillance, logistics, defense, and disaster response — with minimal human intervention.

???? Core Technologies

• SLAM (Simultaneous Localization and Mapping) – Enables precise, real-time environmental mapping and localization without odometry.

• YOLO (You Only Look Once) – Used for fast and accurate object detection.

• Jetson Nano – Serves as the central processing unit, executing AI models and sensor data analysis.

• Web-based Flask Interface – Provides remote monitoring, real-time visualization, and manual control.

?? System Architecture

A. Modular Subsystems

1. Perception & Sensing – Uses LiDAR and a camera to gather environmental data for SLAM and object detection.

2. Computation & Decision-Making – Handles SLAM, YOLO processing, and motion planning on Jetson Nano.

3. Control & Actuation – Controls motor movements based on AI decisions; enables remote user interaction via the web interface.

B. Hardware Components

• Jetson Nano: Runs SLAM, YOLO, and control logic.

• RPLiDAR X2: Generates 360° point cloud data for mapping.

• Camera Module: Captures video input for object detection.

• Arduino + Motor Driver: Executes movement based on commands from Jetson Nano.

C. Software Stack

• ROS (Robot Operating System): Integrates sensors, SLAM, and system communication.

• OpenCV: Handles image processing for YOLO.

• YOLOv5: Performs object detection in real-time.

• Hector SLAM: Efficient LiDAR-based mapping.

• Flask: Hosts the interactive web interface.

D. Workflow Summary

1. Data collection from LiDAR & camera.

2. SLAM maps the environment.

3. YOLO identifies obstacles.

4. Motion planning triggers navigation.

5. Web interface provides real-time feedback and manual control options.

????? Implementation & Testing

• Platform: RC car equipped with all hardware.

• Simulations: Conducted in Gazebo to verify SLAM and detection before real-world use.

• Indoor Tests: Hallway navigation and obstacle avoidance with map accuracy checks.

• Outdoor Tests: Assessed performance on rough terrain and varying lighting.

• Dynamic Testing: AdaptX responded to moving obstacles and updated paths in real time.

???? Results

• Mapping: High-fidelity, real-time maps using Hector SLAM.

• Object Detection: YOLOv9 accurately detected obstacles with low latency.

• Navigation: Effective in avoiding static and moving obstacles; reliable decision-making.

• System Responsiveness: Maintained low latency across perception, planning, and actuation.

• User Interface: Web-based interface was responsive, user-friendly, and informative.

???? Future Work

A. Optimizations

• Integrate multi-sensor fusion (e.g., stereo vision) for enhanced SLAM accuracy.

• Use temporal-aware detection models and larger, more diverse datasets for improved YOLO performance.

B. Platform Expansion

• Drones: For aerial mapping and surveillance.

• Industrial robots: Warehouse automation and inventory management.

• Autonomous vehicles: Urban navigation and smart traffic handling.

C. Advanced AI Integration

• Use reinforcement learning for adaptive navigation strategies.

• Apply semantic segmentation for richer scene understanding.

• Implement deep sensor fusion for improved performance in complex environments.

Conclusion

AdaptX represents a significant advancement in autonomous robotic systems, combining real-time decision-making, SLAM-based navigation, object detection, and web-based control into a seamless and adaptable module. The research highlights the design, development, and performance evaluation of AdaptX, demonstrating its ability to operate autonomously in dynamic environments. By integrating Hector SLAM for mapping, YOLO for object detection, and a Flask-based interface for remote control, the system showcases a well-rounded approach to modern robotic autonomy. The results indicate that AdaptX is highly effective in generating accurate maps, detecting objects in real time, and responding efficiently to environmental changes. Its modularity allows it to be deployed across various robotic platforms, making it a versatile solution for applications such as search and rescue, surveillance, industrial automation, and smart mobility. The system’s adaptability and performance in real-world scenarios underscore its potential to enhance the capabilities of autonomous robotics. Moving forward, AdaptX has immense potential for further development, including deep learning integration, improved sensor fusion, and expansion to aerial and industrial robotic platforms. With continuous improvements, it could become a benchmark for autonomous modules, contributing to the advancement of AI-driven robotics.

References

[1] C. Wei and Z. Jian, \"Application of intelligent UAV onboard LiDAR measurement technology in topographic mapping,\" 2021 IEEE International Conference on Emergency Science and Information Technology (ICESIT), Chongqing, China, 2021, pp. 942-945, doi: 10.1109/ICESIT53460.2021.9696811. [2] B. Cao, R. C. Mendoza, A. Philipp and D. Göhring, \"LiDAR-Based Object-Level SLAM for Autonomous Vehicles,\" 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Prague, Czech Republic, 2021, pp. 4397-4404, doi: 10.1109/IROS51168.2021.9636299. [3] Yaseen, Muhammad. (2024). What is YOLOv9: An In-Depth Exploration of the Internal Features of the Next-Generation Object Detector. 10.48550/arXiv.2409.07813. [4] M. U. Khan, S. A. A. Zaidi, A. Ishtiaq, S. U. R. Bukhari, S. Samer and A. Farman, \"A Comparative Survey of LiDAR-SLAM and LiDAR based Sensor Technologies,\" 2021 Mohammad Ali Jinnah University International Conference on Computing (MAJICC), Karachi, Pakistan, 2021, pp. 1-8, doi: 10.1109/MAJICC53071.2021.9526266. [5] Thale, Sumegh & Prabhu, Mihir & Thakur, Pranjali & Kadam, Pratik. (2020). ROS based SLAM implementation for Autonomous navigation using Turtlebot. ITM Web of Conferences. 32. 01011. 10.1051/itmconf/20203201011. [6] Iram Noreen, Amna Khan and Zulfiqar Habib, “Optimal Path Planning using RRT* based Approaches: A Survey and Future Directions” International Journal of Advanced Computer Science and Applications(IJACSA), 7(11), 2016. http://dx.doi.org/10.14569/IJACSA.2016.071114. [7] Landa, Jaromir & Procházka, David & Stastny, Jiri. (2013). Point cloud processing for smart systems. Acta Universitatis Agriculturae et Silviculturae Mendelianae Brunensis. 61. 2415-2421. 10.11118/actaun201361072415. [8] Wang, Xin & Pan, HuaZhi & Guo, Kai & Yang, Xinli & Luo, Sheng. (2020). The evolution of LiDAR and its application in high precision measurement. IOP Conference Series: Earth and Environmental Science. 502. 012008. 10.1088/1755-1315/502/1/012008. [9] Quigley, Morgan & Conley, Ken & Gerkey, Brian & Faust, Josh & Foote, Tully & Leibs, Jeremy & Wheeler, Rob & Ng, Andrew. (2009). ROS: an open-source Robot Operating System. ICRA Workshop on Open Source Software.

Copyright

Copyright © 2025 Sayali Belhe, Disha Chordiya, Vedant Badre, Siddharth Bhimpure, Pratham Chandratre. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.