代表系統Featured Systems
語音識別機械手軟夾控制應用Voice-controlled robotic soft gripper
回到專案總覽Back to project archive語音控制軟夾展示環境Project image 1 
手臂控制與系統執行畫面Project image 2
整合語音指令、物件辨識與 ROS 2 MoveIt MTC，讓機器手可依語音任務規劃抓取動作。Integrates voice commands, object recognition and ROS 2 MoveIt MTC so a robot arm can plan and execute grasping tasks from spoken instructions.
專案內容Project scope
- 系統整合 OpenAI Whisper、Qwen3B 與 YOLOE，將自然語言任務轉為可執行的抓取目標。
- 使用 PCL 點雲轉換生成夾取姿態，並透過 ROS 2 MoveIt MTC 規劃手臂路徑。
- 目前紀錄顯示成功率約 90%，整體反應時間約 10 秒。
- The system integrates OpenAI Whisper, Qwen3B and YOLOE to convert natural-language tasks into executable grasp targets.
- Point-cloud conversion generates grasp poses, while ROS 2 MoveIt MTC plans arm motion.
- Recorded tests show approximately 90% success and around 10 seconds of overall response time.
我的工作範圍My contribution
我主要負責Owned
- ROS 2 通訊設置。
- 手臂控制。
- 夾取姿態設計。
- 夾取流程設計。
- 將辨識結果銜接至 MoveIt MTC 手臂路徑。
我協作的部分Collaborated
- 團隊整合 OpenAI Whisper、Qwen3B 與 YOLOE 建立語音指令物件辨識系統。
非我主要負責Not my primary scope
- 語音與影像辨識模型主要由其他成員負責。
Owned
- ROS 2 communication setup.
- Robot-arm control.
- Grasp-pose design.
- Grasp workflow design.
- Connecting recognition results to MoveIt MTC motion planning.
Collaborated
- The team integrated OpenAI Whisper, Qwen3B and YOLOE for voice-command object recognition.
Not my primary scope
- Voice and image-recognition models were primarily handled by other team members.
專案影像Project evidence
技術與回顧Technology and review
- 技術深度Technical depth
- 待評分Not rated
- 個人主責Personal ownership
- 待評分Not rated
- 展示成熟度Presentation maturity
- 待評分Not rated
- 回顧價值Review value
- 待評分Not rated