机器人操纵器的轨迹跟踪控制基于软演员-批判和生成对抗模仿学习
Jintao Hu1, Fujie Wang1, Xing Li1
1School of Computer Science and Technology, Dongguan University of Technology, Dongguan 523808, China.
Biomimetics (Basel, Switzerland)
|December 27, 2024
概括
这项研究引入了一种新的深度强化学习 (DRL) 方法SL-GAIL,用于机器人操纵器跟踪控制. 它即使在和约束和干扰的情况下也实现了稳定和有效的控制,超过了其他算法.
科学领域:
- 机器人技术 机器人技术 机器人技术
- 人工智能的人工智能
- 控制系统 控制系统
背景情况:
- 由于和限制和随机干扰,机器人操纵器在跟踪控制方面面临挑战.
- 现有的方法往往需要复杂的动态和动力学模型,这可能很难获得.
- 训练不稳定性是机器人控制强化学习的一个常见问题.
研究的目的:
- 为机器人操纵器跟踪控制提出一种新的深度强化学习 (DRL) 方法.
- 在没有明确的模型学习的情况下解决和约束和随机干扰.
- 提高培训稳定性和控制政策绩效.
主要方法:
- 软演员-关键 (SAC) 和长短期记忆 (LSTM) 组合 (SAC-LSTM) 已开发,通过捕捉联合位置趋势来提高训练稳定性.
- 采用了生成对抗模仿学习 (GAIL),使用SAC-LSTM策略作为专家数据来学习高级控制策略.
- 集成的SAC-LSTM-GAIL (SL-GAIL) 算法旨在直接从稳定的专家数据中学习,避免环境探索.
主要成果:
- 在模拟中,SL-GAIL算法有效地完成了机器人操纵器终端效应器跟踪任务.
- 与其他算法相比,拟议的方法表现出优越的稳定性,特别是在干扰环境中.
- 这种方法成功地管理了扭矩和关节角度和约束.
结论:
- 在具有挑战性的条件下,SL-GAIL算法为机器人操纵器跟踪控制提供了有效和稳定的解决方案.
- 这种无模型的方法通过利用专家数据简化了控制策略的学习过程.
- 这些发现突出了将LSTM,SAC和GAIL结合起来,用于先进的机器人控制应用的潜力.
相关概念视频
01:29Relative Motion Analysis using Rotating Axes-Problem Solving
388
Consider a crane whose telescopic boom rotates with an angular velocity of 0.04 rad/s and angular acceleration of 0.02 rad/s2. Along with the rotation, the boom also extends linearly with a uniform speed of 5 m/s. The extension of the boom is measured at point D, which is measured with respect to the fixed point C on the other end of the boom. For the given instant, the distance between points C and D is 60 meters.
Here, in order to determine the magnitude of velocity and acceleration for point...
Here, in order to determine the magnitude of velocity and acceleration for point...
388
01:24One-Degree-of-Freedom System
464
In mechanical engineering, one-degree-of-freedom systems form the basis of a wide range of electrical and mechanical components. Using these models, engineers can predict the behavior of various parts in a larger system, which gives them insight into how different forces interact with each other.
A one-degree-of-freedom system is defined by an independent variable that determines its state and behavior. One example of a one-degree-of-freedom system is a simple harmonic oscillator, such as a...
A one-degree-of-freedom system is defined by an independent variable that determines its state and behavior. One example of a one-degree-of-freedom system is a simple harmonic oscillator, such as a...
464
01:25Relative Motion Analysis using Rotating Axes
448
Consider a component AB undergoing a linear motion. Along with a linear motion, point B also rotates around point A. To comprehend this complex movement, position vectors for both points A and B are established using a stationary reference frame.
However, to express the relative position of point B relative to point A, an additional frame of reference, denoted as x'y', is necessary. This additional frame not only translates but also rotates relative to the fixed frame, making it...
However, to express the relative position of point B relative to point A, an additional frame of reference, denoted as x'y', is necessary. This additional frame not only translates but also rotates relative to the fixed frame, making it...
448