AI-Based Smart Robotic Arm for Automatically Sorting Objects

Dr P V Rama Krishna1, N Shyam Sundar Sagar2, Palakurthi Vamshi3 ,Anusha4 ,Dhayam Shashank5
Department of Electronics and Communication Engineering, Gandhi Institute of Technology and Management (GITAM),
Hyderabad, Telangana, India.

Abstract Methodology Results

Object recognition combined with sorting based on a single device includes • AI Algorithm Development •The robotic arm accurately detects and classifies multiple objects
the AI-powered robotic arm system as an object recognition system. This one - Develop and train an AI model for object detection (e.g., using YOLOv8). (eraser, NodeMCU, and ping pong ball) in real-time using the YOLOv8n
incorporates a computational apparatus combining sophisticated image - Label dataset, train the model, and deployment on Raspberry Pi. model on the Raspberry Pi.
analysis, object recognition, and categorization. The main goal here is to • Robotic Arm Design and Construction •The project effectively utilizes the lightweight YOLOv8n model,
increase effectiveness and allow for application variety with the help of - Design a 6-DOF robotic arm capable of flexible movements. enabling high-speed object detection on a low-power device (Raspberry
automation of the given sorting challenges . A robotic arm with a camera - Integrate external power sources and mounts for stability and efficiency. Pi) with minimal latency.
module is the core component of the described system. This makes it possible • System Integration •Optimizations to the camera setup successfully minimized latency,
to pick objects in real-time images over its arm operational workspace at - Integrate the Raspberry Pi, servos, and camera. improving response time and fluidity in the robotic arm's movements.
given points in time. Such images are then analysed by some algorithms of • Software Development •Seamless integration between servos, Raspberry Pi, external power
Artificial Intelligence, which classify the object from the image based on the - Write code to control servo movements and capture camera feed. supply, and USB camera demonstrates a robust hardware-software
features visible to the eyes. As such, the system can easily distinguish so many - Implement object detection using the AI model to guide arm movements. synergy.
various types of objects. Classification must be done accurately in order to - Develop calibration algorithms to align arm actions with detected objects. .
sort materials with excellent efficiency. • Prototype Testing and Validation
- Test the robotic arm in real-world scenarios for object detection and
• Automation in object recognition
Introduction
and sorting is essential for improving sorting.
efficiency in industrial processes. • Optimization and Finalization
• Traditional sorting systems lack flexibility and often require complex - Reduce latency in object detection and arm response.
setups, which can limit their application scope.
• AI, particularly deep learning, offers significant advancements in accuracy
and adaptability for sorting tasks.
• Robotic arms combined with image processing enable real-time object
classification and sorting, reducing human intervention. FIGURE 6. Labelled Output Image
• Using YOLO models for object recognition provides high-speed and
accurate sorting, ideal for resource-limited platforms like Raspberry Pi.
• This study demonstrates the integration of a robotic arm with AI-powered
object detection, showcasing a cost-effective, efficient solution for
automated sorting applications.

FIGURE 4. Methodology
FIGURE 5.Target identification and object detection FIGURE 7. YOLO Model Training and Validation
Algorithm Metrics
• YOLOv8n: The "Nano" version of YOLOv8, designed to be the lightest and
fastest in the YOLOv8 family, making it ideal for edge devices like Raspberry
Pi. Conclusions
• Real-Time Object Detection: YOLOv8n provides real-time object detection • The project highlights successful integration between hardware (servos,
Figure 1. The Raspberry Pi 4 capabilities, detecting objects within a frame as they appear. camera, Raspberry Pi) and software (YOLOv8n, servo control), underscoring
• High Speed and Low Latency: The algorithm is optimized for speed, the importance of cohesive design in achieving smooth and accurate
allowing it to process images quickly and detect objects with minimal robotic movements.
latency, even on low-powered hardware. • While the project meets its objectives, further improvements in speed,
• Single Shot Detection: YOLOv8n performs object detection in a single pass accuracy, and camera latency would enhance performance. Future
through the network, which makes it both fast and efficient. developments could explore using a dedicated microcontroller for servo
• Transfer Learning Capability: YOLOv8n supports fine-tuning on custom control, adding a PWM driver, or incorporating additional sensors for more
datasets, allowing developers to adapt it to specific applications, such as complex tasks.
sorting objects by type in robotics. • The project demonstrates that AI-based object detection and sorting can
FIGURE 2. USB Camera Module HD 1080p
• Small Model Size: YOLOv8n has low computational demands, making it be efficiently implemented on low-cost, low-power hardware like the
FIGURE 3. The 6DOF Robotic Arm. suitable for devices with limited processing power or memory. Raspberry Pi, making it accessible and practical for real-world applications.

Team Members References

1. Russell, S., & Norvig, P. (2021). Artificial Intelligence: A Modern Approach (4th ed.). Pearson.
Palakurthi Vamshi(HU21EECE0100134) 2. Siciliano, B., & Khatib, O. (2016). Springer Handbook of Robotics (2nd ed.). Springer.
Anusha (HU21EECE0100570) 3. Bishop, C. M. (2006). Pattern Recognition and Machine Learning. Springer.
Dhayam Shashank (BU21EECE0100526) 4. Corke, P. (2017). Robotics, Vision and Control: Fundamental Algorithms in MATLAB (2nd ed.). Springer.
5. Szeliski, R. (2022). Computer Vision: Algorithms and Applications (2nd ed.). Springer.
6. Kumar, R., & Jain, M. (2021). "Gesture-Controlled Robotic Arm Using Machine Learning," in Proceedings of the International Conference on Advances in
Computing and Communication Engineering (ICACCE), IEEE.