The traditional approaches to robotic gripper control often depend on predefined grasping strategies. This limitation of robotic gripper results in inefficient manipulation and increased risk of object damage during handling. This research aims to develop an adaptive grasping strategy using Reinforcement Learning (RL) to allow robotic grippers to learn and adapt their grasping techniques in real-time based on environmental feedback. The proposed solution is focused on training a model to optimize grip strength, angle, and approach based on the characteristics of the objects. This method is shown initiatives on training a neural network to adjust the grip parameters dynamically which allows the gripper to learn optimal strategies for various objects in real time.