LSWM:通过强化学习进行双脚运动的长短历史世界模型
Jie Xue1,2,3, Zhiyuan Liang1,2,3, Haiming Mou3
1School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China.
Biomimetics (Basel, Switzerland)
|January 27, 2026
概括
这项研究引入了长短世界模型 (LSWM),以改善双脚机器人在具有挑战性的地形上运动. LSWM增强了状态重建和未来预测,显著提高了机器人的性能和适应性.
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:
- 在非结构化的地形上,双脚机器人运动受到传感器噪音,缺失状态和未来预测不佳的限制.
- 传统的方法依赖于国家重建的长期历史,忽视短期动态和风险预测.
研究的目的:
- 提出和评估一个长短的世界模型 (LSWM) 来增强双脚机器人运动.
- 为了提高对传感器噪声的稳定性,并通过使用短期历史来增强状态的可观测性.
- 通过未来状态预测,提高适应复杂环境的能力.
主要方法:
- 开发了一个长短世界模型 (LSWM) 框架,包含两个模块:国家重建模块 (SRM) 和未来国家预测模块 (SPM).
- SRM利用长期历史记录来在短期历史记录中重建特权信息,提高稳定性和可观测性.
- SPM预测未来的特权信息,以提高适应能力和风险预测.
主要成果:
- 通过广泛的模拟和现实世界的实验,LSWM在复杂和动态环境中展示了显著的性能优势.
- 在室内爬楼梯任务中取得了94%的成功率,超过了最先进的方法至少34%.
结论:
- 拟议的LSWM框架有效地提高了双脚机器人在非结构化和动态环境中的移动能力.
- 对于需要先进状态估计和预测控制的现实世界机器人应用,LSWM提供了强大的解决方案.
更多相关视频
09:23JenaTron - An Experimental Approach to Study the Effects of Plant History and Soil History on Grassland Ecosystem Functioning
Published on: March 21, 2025
1.9K
10:43Eye-tracking Technology and Data-mining Techniques used for a Behavioral Analysis of Adults engaged in Learning Processes
Published on: June 10, 2021
5.8K
相关概念视频
What is Evolutionary History?
43.2K
Scientists record evolutionary history by analyzing fossil, morphological, and genetic data. The fossil record documents the history of life on Earth and provides evidence for evolution. However, both fossil and living organisms offer evidence that outlines Earth’s evolutionary history.
43.2K
Reinforcement
901
Positive and negative reinforcement are key concepts in operant conditioning, a learning process where the consequences of a behavior affect the likelihood of that behavior being repeated.
Positive reinforcement occurs when a behavior is followed by the presentation of a rewarding stimulus, increasing the frequency of that behavior. For example:
Positive reinforcement occurs when a behavior is followed by the presentation of a rewarding stimulus, increasing the frequency of that behavior. For example:
901
History of Microbiology
9.0K
Microbiology, a scientific field dedicated to the study of microorganisms, has undergone profound development since its inception in the 17th century. Its history is marked by key discoveries and technological advancements that have shaped our understanding of life at the microscopic level and transformed medicine, agriculture, and industry.Early Foundations of MicrobiologyThe early foundations of microbiology were built on groundbreaking observations and the development of pioneering...
9.0K
Life Histories
22.7K
Overview
22.7K
Corrosion of Reinforcement
567
The corrosion of steel reinforcement within concrete is a process influenced by the material's inherent properties and external factors. The high pH level of around 13, provided by calcium hydroxide present in concrete, initially protects the steel reinforcement by promoting the formation of a passive iron oxide layer on its surface.
However, over time and under certain conditions like carbonation, chloride ingress, and cracking this protective state can be compromised. Steel has areas with...
However, over time and under certain conditions like carbonation, chloride ingress, and cracking this protective state can be compromised. Steel has areas with...
567
Reinforcement Schedules
494
Positive reinforcement is a powerful method for teaching new behaviors to both animals and humans. B.F. Skinner demonstrated this with his experiments using rats in a Skinner box. When a rat pressed a lever, it received a food pellet. This immediate reward encouraged the rat to repeat the behavior. This method, where a reward follows every instance of the behavior, is known as continuous reinforcement. It is highly effective for establishing new behaviors quickly.
Once a behavior is learned,...
Once a behavior is learned,...
494
