Context-Continuous Preference Learning for Exoskeleton Personalization
作者: Sunin Baek, Sungwoo Park, Daekyum Kim
分类: cs.LG, cs.RO
发布日期: 2026-09-23
备注: 23 pages, 11 figures, including supplementary materials
💡 一句话要点
提出上下文连续偏好学习以解决外骨骼个性化问题
🎯 匹配领域: 支柱二:RL算法与架构 (RL & Architecture)
关键词: 外骨骼 个性化学习 高斯过程 偏好建模 贝叶斯优化 用户反馈 机器人技术
📋 核心要点
- 现有方法在个性化外骨骼辅助时,收集用户反馈的时间和体力成本较高,限制了其应用。
- 本文提出上下文连续偏好学习(CCPL),通过高斯过程模型在相邻上下文间共享观察数据,以提高学习效率。
- 实验结果显示,CCPL在脚踝和肘部辅助的重建相关性上分别提高了0.076和0.050,且所需的反馈预算显著降低。
📝 摘要(中文)
个性化外骨骼辅助的挑战在于收集用户反馈所需的时间和体力。本文研究了用户偏好在不同操作条件下的平滑变化,并提出了上下文连续偏好学习(CCPL),一种高斯过程偏好模型,能够在保持上下文特定效用估计的同时共享相邻上下文的观察数据。通过对九名健康成人的脚踝和肘部外骨骼偏好数据进行模拟和回顾性分析,CCPL在偏好平滑变化时相较于独立学习显著提高了重建和基于偏好的贝叶斯优化效果。研究结果表明,上下文连续性为在有限反馈下共享偏好观察提供了基础,尽管在线个性化的实际效益尚待验证。
🔬 方法详解
问题定义:本文旨在解决个性化外骨骼辅助中用户反馈收集的高成本问题。现有方法往往依赖独立学习,无法有效利用相邻操作条件下的用户偏好信息。
核心思路:提出上下文连续偏好学习(CCPL),通过高斯过程模型在相邻上下文之间共享观察数据,从而在有限反馈下提高学习效率和准确性。
技术框架:CCPL的整体架构包括数据收集、偏好建模和优化三个主要模块。首先收集用户在不同操作条件下的反馈数据,然后利用高斯过程模型进行偏好估计,最后通过贝叶斯优化进行个性化调整。
关键创新:CCPL的核心创新在于其上下文连续性假设,允许在相邻上下文间共享偏好信息,从而克服了独立学习的局限性,尤其是在偏好变化平滑的情况下。
关键设计:在模型设计中,采用高斯过程作为偏好建模的基础,设置了上下文相关的超参数,以确保模型能够有效捕捉不同操作条件下的用户偏好变化。
🖼️ 关键图片
📊 实验亮点
实验结果表明,CCPL在脚踝辅助的重建相关性从0.644提高到0.720,在肘部辅助的重建相关性从0.476提高到0.526。此外,CCPL所需的反馈预算分别比独立学习低约37%和17%。
🎯 应用场景
该研究的潜在应用领域包括康复机器人、助行器和其他可穿戴设备的个性化设计。通过提高个性化的效率和准确性,CCPL有望在实际应用中显著改善用户体验,降低用户反馈收集的负担,推动智能辅助技术的发展。
📄 摘要(原文)
Personalizing exoskeleton assistance across operating conditions is constrained by the time and physical effort required to collect user feedback. We examined whether a user's preference landscape varies smoothly across operating conditions and when this continuity supports learning from limited feedback. We propose Context-Continuous Preference Learning (CCPL), a Gaussian-process preference model that shares observations across nearby contexts while retaining context-specific utility estimates. We evaluated CCPL through simulations and retrospective analyses of ankle and elbow exoskeleton preference data from nine healthy adults. In simulations, CCPL improved reconstruction and preference-based Bayesian optimization relative to independent learning when preferences varied smoothly, but showed negative transfer when continuity was weak. In both human studies, full-data reference landscapes estimated separately for each participant and context tended to be more similar between nearby operating conditions. With five exposures per context, CCPL increased mean reconstruction correlation with these references from 0.644 to 0.720 for ankle assistance and from 0.476 to 0.526 for elbow assistance relative to independent learning. The five-exposure budget was approximately 37% lower for ankle and 17% lower for elbow than the estimated independent-learning budgets needed to match these correlations. CCPL also improved held-out response prediction relative to independent learning, while benefits over pooled learning varied. These findings support context continuity as a basis for sharing preference observations under limited feedback, although benefits for online personalization in humans remain to be established.