PhD, Engineering Mathematics工程数学博士
University of BristolUniversity of Bristol
Thesis: Human-Inspired Neuromorphic Tactile Capabilities. Expected completion: March 2027.论文题目：受人类启发的神经形态触觉能力研究。预计 2027 年 3 月毕业。Experience经历
Research depth. Engineering practice.研究积累与工程实践
From tactile perception research to VLA fine-tuning and robot deployment, with industrial integration, teaching and research communication experience.我的研发工作覆盖触觉传感器原型、实验数据分析、机器人模型微调与本地部署。此前在越疆机器人承担真机测试和技术支持，也在布里斯托大学负责课程教学与科研视频制作。
Experience trace科研与工作经历
- Bristol, UK英国 · 布里斯托
Project Lead · VLA Development项目负责人 · VLA 研发
Dual-arm VTLA Manipulation Project
- Lead the VLA workstream on a dual-Franka leader–follower platform, collaborating on hardware and demonstration collection; define interfaces for front, wrist and paired tactile RGB with robot state and actions.负责双 Franka 主从平台的 VLA 路线，与团队协作开展硬件与示教采集；定义前视、腕部、双路触觉 RGB 及机器人状态、动作的数据接口。
- Completed three independent π0.5-DROID LoRA fine-tunes of 5,000 steps each, covering a visual baseline, paired tactile image inputs and a human-reviewed data export.完成三组独立 π0.5-DROID LoRA 微调，每组 5,000 步，覆盖视觉基线、双路触觉图像输入和人工审核数据导出。
- Built deterministic data checks and shared training/inference image preprocessing; retained 8,116 frames for the baseline comparison and established a separate 24/6-episode training/validation split with training-only normalisation.建立确定性数据检查与训练／推理共享图像预处理；为基线对照保留 8,116 帧，另为审核导出设置 24／6 段示教的训练／验证划分及仅基于训练数据的归一化。
- Set up RTX 5090 policy inference and documented checkpoint/adapter compatibility; prepare a common closed-loop robot evaluation with fixed success criteria and per-trial records.搭建 RTX 5090 策略推理环境，记录 checkpoint 与适配器的兼容关系；准备统一的真机闭环评测，固定成功判据并逐次记录结果。
- Bristol, UK英国 · 布里斯托
Researcher研究员
Bristol Robotics Laboratory · Dexterous Group
- Designed six robot exploratory motions and evaluated NeuroTac event streams over 50–1,000 ms windows. Under varying speeds and contact depths, sliding + rotation achieved 87.33% texture-classification accuracy versus 79.67% for sliding, a 7.66 percentage-point gain at a 1,000 ms window.设计六种机械臂探索动作，评估 50–1,000 ms 窗口下的 NeuroTac 事件流；在变运动速度与变接触深度条件下，1,000 ms 窗口的“滑动＋旋转”纹理分类准确率为 87.33%，较纯滑动的 79.67% 提升 7.66 个百分点。
- Processed tactile events through spatial cropping, spatiotemporal pooling and temporal windowing; built a lightweight IF-neuron SCNN with 2Conv + 2Pool + 3FC for ten-class texture recognition.对触觉事件流进行空间裁剪、时空聚合池化与时间截取，基于 IF 脉冲神经元构建 2Conv + 2Pool + 3FC 轻量化 SCNN，完成十类纹理识别。
- Quantised and mapped the SCNN with Sinabs for real-time on-chip inference on SynSense Speck2f; measured 8.04 mW average on-chip operating power for sliding + rotation inputs, excluding the sensor and wider robot system.使用 Sinabs 完成 SCNN 量化与权重映射，在 SynSense Speck2f 上完成实时片上推理；“滑动＋旋转”输入下实测平均片上运行功耗为 8.04 mW，不代表传感器或整套机器人系统功耗。
- Designed TacTip / NeuroTac housings, optical supports and robot flanges in Fusion 360, iterated tolerances and fabricated PolyJet / FDM sensor prototypes for robot integration and tactile data collection.使用 Fusion 360 设计 TacTip / NeuroTac 壳体、光学支架与机械臂法兰，迭代公差并通过 PolyJet / FDM 制作传感器原型，完成机械臂集成并支撑触觉数据采集。
- Bristol, UK英国 · 布里斯托
Teaching Assistant助教
University of Bristol
- Lead engineering and robotics labs, tutorials and assessment.承担工程与机器人课程的实验课、辅导课，按课程评分标准评阅作业与技术报告，并给出具体修改建议。
- Co-develop teaching material and give structured feedback.与课程负责人共同开发教学材料，将理论内容转化为实践任务，跟进学生项目进度并支持课程运行。
- Bristol, UK英国 · 布里斯托
Communication Assistant学院传播助理
Faculty of Science & Engineering · University of Bristol
- Directed, filmed and edited stories about university facilities and research.负责院系视频的策划、拍摄与剪辑，与多个部门协作完成科研人员、研究设施和校园信息等内容制作。
- Managed social media and supported prospective international students.运营学部官方社交媒体，通过微信群组织新生行前问答，为国际学生提供入学信息。
- Shenzhen, China中国 · 深圳
Technical Support Engineer Intern技术支持工程师（实习）
Shenzhen Yuejiang Technology · Dobot
- Developed communication and control logic connecting Siemens S7-1200 / 1500 PLCs with Dobot arms, including hand–eye calibration, conveyor tracking, pick-and-place and palletising workflows.开发 Siemens S7-1200 / 1500 PLC 与 Dobot 机械臂的通信及逻辑控制方案，完成手眼标定、传送带跟踪抓取、搬运与码垛流程。
- Tested Magician, M1 and M1 Pro hardware, calibrated repeatability and diagnosed faults; updated technical documentation, developed training cases and delivered PLC–robot integration training for university and enterprise customers.开展 Magician、M1 和 M1 Pro 硬件测试、重复定位精度标定与故障诊断；更新技术文档，编写实训案例，并面向高校与企业客户开展 PLC 与机器人系统集成培训。
MSc, Robotics机器人学硕士
University of Bristol & University of the West of EnglandUniversity of Bristol & University of the West of England
Distinction优秀Dissertation: Texture Exploration Strategies.毕业论文：纹理探索策略。BSc, Automation自动化学士
Northeastern University, China东北大学