Motion Planning of a Flexible Robotic Arm Based on Neural Network Under Simulated Conditions

Abstract

Flexible robotic systems are gaining significant attention due to their ability to adapt to complex environments, offering advantages in fields such as medical robotics and autonomous manipulation. This study focuses on developing a neural network-based control framework to enhance the motion planning and tracking capabilities of a flexible robotic manipulator. The proposed controller utilizes simulated datasets to train a neural network that maps the current state of the manipulator's actuators and the target end-effector position to the next state, enabling precise single-point positioning and trajectory tracking. Experimental results demonstrate minimal tracking errors, with the mean squared error (MSE) converging close to zero, indicating high accuracy across training, validation, and test phases. The system successfully follows both static targets and dynamic paths, as confirmed by the alignment between actual and target trajectories. Additionally, the controller shows stable performance, ensuring smooth motion without oscillations or delays. This research highlights the potential of neural networks to improve control precision in flexible robotic systems, bridging the gap between simulation and real-world applications. The study lays a solid foundation for further advancements in intelligent control frameworks for flexible robotic systems, contributing to the development of adaptive and autonomous robotic platforms.