Research Work Details

EN 中文

← Back to Home

Research Work Introduction

DexHand Sensing System: A System Concept Note (Sensor Stack × Multi-rate Loops × Learning)

Type: System Design Note | Focus: Dexterous robotics & visuotactile sensing | [Read]

This note summarizes a structured sensing-system design for dexterous hands: what to sense, how to synchronize and calibrate multimodal streams, and how to build learning-ready data schemas. It emphasizes a three-loop view (fast reflex / mid tactile coordination / slow semantics) to make grasp stability and recovery robust in contact-rich manipulation.

Highlights
  • Sensor stack from minimum viable proprioception to visuotactile / dynamic tactile arrays
  • Multi-rate closed loop decomposition: 1 kHz safety/reflex, 60–200 Hz tactile control, 5–15 Hz planning
  • Timestamp-at-source + ring buffer alignment for replayable data
  • Learning-ready schema: tactile + proprioception + action + QC labels (replay_ok, recovery_event)

Also see: Thinking Board (a lightweight local message board for future notes).

MoiréTac: A Dual-Mode Visuotactile Sensor for Multidimensional Perception Using Moiré Pattern Amplification

Status: Accepted | Conference: IEEE ICRA 2026 | arXiv: 2509.12714 | [Paper]

This research develops a dual-mode visuotactile sensor based on moiré pattern amplification, enabling multidimensional perception. The sensor supports multi-DoF force/pose estimation and pre-grasp cues, providing continuous tactile field perception through vision-through-contact fusion technology.

Key Contributions
  • Innovatively applied moiré pattern amplification technology to tactile perception
  • Achieved vision-through-contact dual-mode fusion perception
  • Supports multi-DoF force/pose estimation
  • Provides pre-grasp cues to enhance robotic manipulation capabilities
MoiréTac Sensor

UltraTac: Integrated Ultrasound-Augmented Visuotactile Sensor for Enhanced Robotic Perception

Status: Accepted | Conference: IEEE/RSJ IROS 2025 | [arXiv]

This research proposes an integrated ultrasound-augmented visuotactile sensor for enhanced robotic perception. The sensor combines vision, tactile, and ultrasound technologies to provide more accurate perception information in complex environments.

Technical Highlights
  • Multimodal fusion (Visuotactile + ultrasound)
  • Enhanced perception accuracy and robustness
  • Applicable to robotic manipulation in complex environments
UltraTac Sensor

A Bio-Inspired Event-Driven Mechanoluminescent Visuotactile Sensor for Intelligent Interactions

Journal: Advanced Functional Materials | Year: 2024/2025 | [DOI]

This research develops a bio-inspired event-driven mechanoluminescent visuotactile sensor for intelligent interactions. The sensor features low power consumption and strong ambient-light immunity, validated for human-robot interaction applications on a quadruped platform.

Innovations
  • Event-driven mechanism for reduced power consumption
  • Mechanoluminescent materials for self-illuminated perception
  • Strong ambient-light immunity
  • Validated on quadruped robot platform
Mechanoluminescent Sensor

FlexLock: Bionic Snake-Inspired Variable-Stiffness Flexible Robotic Arm

Status: Under Review | Journal: Cyborg and Bionic Systems (SCI Q1)

This research proposes a snake-inspired variable-stiffness flexible robotic arm, combining shape memory alloy (SMA) and particle jamming technologies for stiffness modulation. The design is suitable for safe human-robot interaction and high-load manipulation scenarios.

Design Features
  • Bio-inspired design inspired by snakes
  • Stiffness modulation using SMA and particle jamming technology
  • Safe human-robot interaction capabilities
  • High-load manipulation capabilities
FlexLock Robotic Arm

论文工作介绍

灵巧手传感系统构想:传感栈 × 多频闭环 × 学习数据结构

类型: 系统设计笔记 | 方向: 灵巧操作与视触觉 | [阅读]

该笔记总结了灵巧手传感系统的结构化设计:需要感知哪些信号、如何做多模态同步与标定,以及如何把数据组织成面向 VLA/RL 的训练单元。 核心是“三层闭环”:快反射/安全(1kHz)、触觉协调(60–200Hz)、慢语义规划(5–15Hz),从而提升接触密集任务中的稳定性与恢复能力。

要点
  • 从最小可用本体感知到视触觉/动态触觉阵列的传感栈
  • 多频闭环拆分:快安全、中触觉、慢语义
  • 设备端打戳 + 环形缓冲对齐,让数据可复现
  • 面向学习的数据 schema:触觉+本体+动作+质控标签(replay_ok / recovery_event)

也可以看:Thinking Board(一個輕量留言版,用來記錄技術/產品/經濟等未來想法)。

MoiréTac: 基于莫尔图案放大的双模式视触觉传感器

状态: 已接收 | 会议: IEEE ICRA 2026 | arXiv: 2509.12714 | [论文]

本研究开发了一种基于莫尔图案放大的双模式视触觉传感器,能够实现多维感知。该传感器支持多自由度力/姿态估计和预抓取线索,通过视觉-接触融合技术提供连续触觉场感知。

主要贡献
  • 创新性地将莫尔图案放大技术应用于触觉感知
  • 实现了视觉-接触双模式融合感知
  • 支持多自由度力/姿态估计
  • 提供预抓取线索,提升机器人操作能力
MoiréTac Sensor

UltraTac: 集成超声增强的视触觉传感器

状态: 已接收 | 会议: IEEE/RSJ IROS 2025 | [arXiv]

本研究提出了一种集成超声增强的视触觉传感器,用于增强机器人感知能力。该传感器结合了视觉、触觉和超声技术,能够在复杂环境中提供更准确的感知信息。

技术亮点
  • 多模态融合(视触觉+超声)
  • 增强的感知精度和鲁棒性
  • 适用于复杂环境下的机器人操作
UltraTac Sensor

受生物启发的事件驱动机械发光视触觉传感器

期刊: Advanced Functional Materials | 年份: 2024/2025 | [DOI]

本研究开发了一种受生物启发的事件驱动机械发光视触觉传感器,用于智能交互。该传感器具有低功耗和强环境光抗干扰能力,已在四足机器人平台上验证了人机交互应用。

创新点
  • 事件驱动机制,降低功耗
  • 机械发光材料,实现自发光感知
  • 强环境光抗干扰能力
  • 在四足机器人平台上验证
Mechanoluminescent Sensor

FlexLock: 受蛇类启发的可变刚度柔性机械臂

状态: 审稿中 | 期刊: Cyborg and Bionic Systems (SCI Q1)

本研究提出了一种受蛇类启发的可变刚度柔性机械臂,结合了形状记忆合金(SMA)和颗粒阻塞技术,实现刚度调制。该设计适用于安全的人机交互和高负载操作场景。

设计特点
  • 受蛇类生物启发的仿生设计
  • SMA和颗粒阻塞技术实现刚度调制
  • 安全的人机交互能力
  • 高负载操作能力
FlexLock Robotic Arm