Artificial Intelligence-Based Adaptive Gripping Techniques for Robotic Manipulation of Objects with Varying Material Properties

Authors:

                        Jeevan K L¹                                                                             Dr. Lokesha K²

     Department of Robotics and Artificial Intelligence¹                            Department of Robotics and Artificial Intelligence²

        M S Ramaiah Institute of Technology¹,                                              M S Ramaiah Institute of Technology²,
Bangalore – 5600541                                                                                   Bangalore – 560054

             jeevankl2327@gmail.com                                                                         lokesha.krish@msrit.edu

Abstract:

Robotic manipulation systems are increasingly being adopted in industrial automation, logistics, healthcare, and service robotics. However, conventional robotic grippers often apply a fixed gripping force, which can lead to object deformation, slippage, or damage when handling objects with different material properties. To address this challenge, this paper presents an Artificial Intelligence-Based Adaptive Gripping System for robotic pick-and-place applications. The proposed system integrates a robotic arm equipped with force and weight sensing mechanisms to identify object characteristics and automatically adjust gripping force according to the detected material properties.

A machine learning model is trained using data collected from various soft and hard objects, including their weight, gripping force, and sensor responses. The trained model classifies objects into different categories and determines the optimal gripping pressure required for safe manipulation. Based on the classification results, the robotic arm dynamically controls the gripper to ensure stable grasping while preventing object damage.

Experimental results demonstrate that the proposed approach achieves high object-classification accuracy and reliable grasping performance while maintaining adaptability to previously unseen objects. The integration of machine learning with adaptive force control significantly improves manipulation efficiency compared to conventional fixed-force gripping methods. The proposed system offers a low-cost and scalable solution for intelligent robotic manipulation and can be extended for advanced industrial and service-robotics applications.

Keywords: Robotic Arm, Adaptive Gripping, Machine Learning, Object Classification, Pick-and-Place, Force
Control, Artificial Intelligence, Smart Gripper, Industrial Automation.