لطفا منتظر بمانید ...
0% Complete
صفحه اصلی
/
یازدهمین کنفرانس بین المللی کنترل ، ابزار دقیق و اتوماسیون
A Hybrid Inverse Reinforcement Learning and Deep Deterministic Policy Gradient Framework for Model-Free Control of Robotic Manipulators
نویسندگان :
Samim Kamyab
1
Ghoncheh Zand
2
Masoud Shirzadeh
3
1- University of Tehran
2- Amirkabir University of Technology (Tehran Polytechnic)
3- Amirkabir University of Technology (Tehran Polytechnic)
کلمات کلیدی :
Deep Deterministic Policy Gradient،Reinforcement Learning
چکیده :
This paper introduces a novel hybrid control framework for robotic manipulators that integrates inverse reinforcement learning, Behavior Cloning (BC), and Deep Deterministic Policy Gradient (DDPG). A reward function was constructed using data generated by a Proportional Integral Derivative (PID) controller acting as the expert agent. To accelerate learning and improve policy stability, the actor network was pre-trained with BC to imitate the expert’s behavior and subsequently refined using DDPG under the learned reward function. Results show that the proposed method improves the integral of absolute error by 35% and the integral of squared error by approximately 30% in the controlled connections compared to the classical PID controller. Moreover, Unlike model-based approaches, the proposed framework does not require exact dynamic model or explicit cost functions, which makes it more suitable for real-world applications. the hybrid design also leads to faster convergence and superior performance compared to standard DDPG methods.
لیست مقالات
لیست مقالات بایگانی شده
Implemention of LQR controller for HVAC systems and investigating effect of gains on improving performance
Ali Parsaie kia - Paria Moradi
شبیه سازی بازوی 5urfd در متلب و v-rep
رضا جلالی رومی - علیرضا نقی پور - فراز جباری ایلخچی
Observability-Driven Transceiver Design for Over-the-Air Control Systems
Amir Hossein Mansouri - Younes Esmaeili - Mohammad Haeri
Integrating Vision and Robotics: Automated Object Localization and Fault Detection with a 6-DoF Manipulator
Masoumeh Azadegan - Alaleh Vafaei
تحلیل پایداری وابسته و مستقل از تاخیر سیستم های سوئیچینگ با تاخیر زمانی زمان-متغیر مبتنی بر تابع لیاپانوف چندگانه
حسین چهاردولی
Frequency Regulation Improvement in AC Microgrids: A Fuzzy-Based Extended Virtual Synchronous Generator Control
Arman Jafari - Sharara Rehimi - Hassan Bevrani
A comparison study on the dynamic control of OpenMANIPULATOR-X by PD with gravity compensation tuned by oscillation damping based on the phase-trajectory-length concept
Amirhossein Dadbin - Ahmad Kalhor - Mehdi Tale Masouleh
H∞ Robust Control Design for Three-Element Industrial Boiler Supervisory System
Narges Ahooghalandari - Reza Shadi - Ahmad Fakharian
A Novel Cyber-Secure Algorithm in Cooperative Vehicles Platoon
Abolfazl Saadati Moghadam - Mohammad Haeri
Image-based Visual Servoing for Dynamic Positioning of a Work-class ROV with Depth Computation
Ali Nourmohammadi - Mehdi Loueipour - Alireza Hosseinnajad
بیشتر
ثمین همایش، سامانه مدیریت کنفرانس ها و جشنواره ها - نگارش 44.9.5