Nâng cao hiệu quả điều khiển robot, sử dụng học tăng cường kết hợp học sâu.=Increased effective of control robot using reinforcement combine with deep learning.

0 Người đánh giá. Xếp hạng trung bình 0

Tác giả: Thị Thảo Hiếu Lương, Thị Thùy Phạm

Ngôn ngữ: vie

Ký hiệu phân loại:

Thông tin xuất bản: Khoa học & Công nghệ (ĐH Kinh tế Kỹ thuật Công nghiệp), 2024

Mô tả vật lý: tr.54-63

Bộ sưu tập: Metadata

ID: 475771

Thêm vào giỏ

Although deep learning can solve problems that cannot be done by tradition machine learning algorithms, it requires huge a amount of data, which is not always available in control problems. Reinforcement learning is a good solution in robot control, the data that is generated when the agent interacts with environment. Along with the develope of neural networks, many reseracher have focused on combine neural network with reinforcement learning to create deep reinforcement learning. In this paper, we propose a new deep reinforcement learning model based on the improvement of the traditional Deep Q Learning algorithm by combining techniques: Fixed_Q Target, Double Deep Q, Prioritized Experience Replay with CNN network (VGG16), apply to control the robot with state and agents designed by using Unity ML-Agents. Experimentally, comparing and evaluating the effectiveness of the proposed model with the original on the simulation environment, the test results show that the proposed method fix the overestimation q value and fast convergence.Mặc dù học sâu có thể giải quyết các bài toán mà các thuật toán học máy cũ không giải quyết được nhưng cần lượng dữ liệu rất lớn và trong thực tế dữ liệu không phải lúc nào cũng có sẵn trong bài toán điều khiển. Học tăng cường là một giải pháp tốt trong bài toán điều khiển robot, dữ liệu được tạo ra khi tác tử tương tác với môi trường. Cùng với sự ra đời của mạng neural, nhiều nghiên cứu đã tập trung kết hợp mạng neural vào học tăng cường tạo nên học tăng cường sâu. Trong bài báo này chúng tôi đề xuất mô hình học tăng cường sâu mới dựa trên sự cải tiến thuật giải Deep Q Learning truyền thống bằng cách kết hợp các kỹ thuật: Fixed_Q Target, Double Deep Q, Prioritized Experience Replay, với mô hình mạng VGG16, ứng dụng điều khiển robot xếp hàng hóa với không gian trạng thái tự thiết kế sử dụng Unity ML-Agents. Thực nghiệm, so sánh đánh giá hiệu quả mô hình đề xuất so với mô hình ban đầu, kết quả cho thấy phương pháp đề xuất hội tụ nhanh và khắc phục được hiện tượng ước lượng quá mức giá trị q.

1. Đọc trực tuyến

1. Đọc trực tuyến

Tạo bộ sưu tập với mã QR