課程時長:16週(每週3小時,包含講座、實驗室與練習)
先修條件:具備AI的基礎知識。熟悉空間資料處理有助益但非必要。
課程目標:培養學生設計、開發與評估基於AI視覺的無障礙輔助科技的能力,重點涵蓋導航、空間智慧、環境安全以及視障人士、認知或情感需求的支援。
課程結構
課程分為五個模組,每個模組聚焦於AI視覺與輔助科技的關鍵領域。每個模組包含講座、實作練習與專案,以強化學習。課程大綱將提供的科目與練習整合成一個連貫的課程。
模組一:AI視覺與空間智慧基礎(第1-3週)
目標:理解AI視覺、空間智慧及其在輔助科技中的應用基礎。
第1週:AI視覺與空間智慧簡介
- Topics:
- Overview of AI vision: Object detection, image segmentation, and scene understanding
- AI needs spatial intelligence: Concepts of spatial awareness and mapping
- AI needs spatial intelligence 2: Advanced spatial reasoning with AI
- AI needs spatial intelligence 3: Detect obstacles: Techniques for obstacle detection using computer vision
- 活動:
- 講座:介紹電腦視覺框架(例如OpenCV、TensorFlow、PyTorch)
- 實驗室:設置電腦視覺環境並處理範例圖像
- Exercise: Blind spatial intelligence (Introduction to spatial data visualization for accessibility)
第2週:障礙與危險偵測
- Topics:
- Detecting obstacles: Algorithms for real-time obstacle detection (e.g., YOLO, SSD)
- Detecting hazards: Identifying environmental dangers using AI vision
- Perspective transformation: Converting 2D images to 3D spatial models
- 活動:
- 講座:深度學習應用於物件與危險偵測
- 實驗室:使用預訓練數據集訓練簡單的障礙偵測模型
- Exercise: Obstacle Avoidance 1 (Simulate obstacle detection in a controlled environment)
第3週:空間資訊與識別輔助
- Topics:
- Spatial information and recognition assistance: Building spatial maps for navigation
- Spatial sketching and spatial maps: Techniques for creating 2D/3D spatial representations
- Real world, spatial intelligence: Applying spatial intelligence in real-world scenarios
- 活動:
- 講座:Voronoi圖與空間分割應用於導航
- 實驗室:使用感測器數據建立基礎空間地圖
- Exercise: Blind spatial intelligence (Develop a simple spatial map for a virtual environment)
模組二:視障導航與安全之AI應用(第4-7週)
目標:開發針對視障人士的AI視覺導航與安全系統。
第4週:視障導航之AI視覺
- Topics:
- Blind navigation image recognition: Image-based navigation systems for the visually impaired
- Blind navigation image recognition 2 (safe route): Path planning for safe navigation
- AI blind visual assistive technology: Tools and frameworks for assistive navigation
- 活動:
- 講座:感測器融合(攝影機、LiDAR、超聲波)應用於導航
- 實驗室:實現基礎圖像識別導航系統
- Exercise: Blind AI navigation (Simulate a navigation scenario for a visually impaired user)
第5週:進階導航與安全路徑規劃
- Topics:
- AI Vision and LLM for Safe Routing Recommendation: Integrating large language models for route planning
- AI Vision and LLM for Safe Routing Planning OmniSafe (Voronoi-based): Voronoi-based path planning
- OmniSafe: Follow Voronoi 1 (New Scenario Campus Trail): Campus trail navigation
- 活動:
- 講座:Voronoi圖在安全路徑規劃中的應用
- 實驗室:實現基於Voronoi的導航演算法
- Exercise: Blind AI navigation (Design a safe route for a campus trail scenario)
第6週:OmniSafe導航場景
- Topics:
- OmniSafe: Follow Voronoi 2 (New Scenario Corridor): Corridor navigation
- OmniSafe: Follow Voronoi 3 (Country Alley): Rural alley navigation
- OmniSafe: Follow Voronoi 4 (Rural Road): Rural road navigation
- Advanced Navigation for People with Visual Impairments: Multimodal navigation systems
- 活動:
- 講座:不同導航環境中的挑戰
- 實驗室:在模擬的鄉村與城市環境中測試導航演算法
- Exercise: Obstacle Avoidance 2 (Test navigation in a complex environment with obstacles)
第7週:自動駕駛與無障礙
- Topics:
- Autopilot: AI-driven autonomous navigation systems
- Accessible space detection: Identifying accessible routes and spaces
- 活動:
- 講座:自動駕駛系統及其在無障礙中的適應
- 實驗室:模擬針對視障人士的自動駕駛系統
- Exercise: Blind AI navigation (Integrate autopilot features into a navigation prototype)
模組三:認知與情感輔助科技(第8-10週)
目標:探索AI在認知支援與情感分析中的輔助科技應用。
第8週:認知支援與多工任務
- Topics:
- Information processing, cognitive disability life assistance: AI for cognitive support
- Multitasking training, delaying dementia: Cognitive exercises using AI
- 活動:
- 講座:AI在認知療法與多工任務中的應用
- 實驗室:設計簡單的認知訓練應用程式
- Exercise: Assistive device design (Prototype an AI-based cognitive aid)
第9週:AI情感分析
- Topics:
- AI-assisted emotional analysis: Sentiment and emotion detection using AI
- AI-assisted emotional analysis (II) (Boundary-themed app): Focused emotional analysis applications
- 活動:
- 講座:面部表情識別與自然語言處理應用於情感分析
- 實驗室:使用面部或語音數據建立情感偵測模型
- Exercise: Emotion, mood, artificial intelligence (Analyze emotional data from sample inputs)
第10週:進階情感與認知應用
- Topics:
- Integration of emotional and cognitive assistive technologies
- Ethical considerations in AI for emotional and cognitive support
- 活動:
- 講座:輔助AI系統的倫理設計
- 實驗室:將情感與認知支援整合至單一原型
- Exercise: Advanced AI applications (Develop an integrated emotional-cognitive assistive tool)
模組四:AI、虛擬實境與遊戲化無障礙應用(第11-13週)
目標:利用AI、虛擬實境(VR)與遊戲化技術開發創新輔助應用。
第11週:AI + VR + 遊戲化無障礙應用
- Topics:
- AI + VR + Gaming: Service applications of major technology trends: Trends in assistive tech
- G4H app creation, media creation: Gamification for health and accessibility
- 活動:
- 講座:VR與遊戲化在無障礙中的應用
- 實驗室:開發簡單的VR輔助應用程式
- Exercise: G4H app creation, media creation (Design a gamified accessibility app)
第12週:生成式AI在環境安全中的應用
- Topics:
- Generative AI in environmental safety applications: Generating safety visualizations
- Safety Visualization: Visualizing environmental risks
- Safety Visualization for All People: Inclusive safety design
- 活動:
- 講座:生成式AI用於視覺化(例如DALL·E、Stable Diffusion)
- 實驗室:使用生成式AI創建安全視覺化
- Exercise: Advanced AI applications (Generate a safety visualization for a public space)
第13週:運籌學與演算法設計
- Topics:
- Operational Research: Optimization for assistive technologies
- Dog fetching ball by the pool without getting wet: Path optimization algorithms
- 活動:
- 講座:優化技術(例如Dijkstra、A*、Voronoi)
- 實驗室:解決路徑優化問題
- Exercise: Algorithm (Design an algorithm for a navigation scenario)
模組五:進階應用與實作(第14-16週)
目標:整合知識,開發全面的輔助科技應用。
第14週:專案提案與設計
- Topics:
- Review of AI vision, navigation, cognitive, and emotional assistive technologies
- Project planning and requirements gathering
- 活動:
- 講座:專案設計與評估指標
- 實驗室:頭腦風暴與提出應用專案
- Exercise: Advanced AI applications (Draft a project proposal)
第15週:專案開發
- Topics:
- Prototyping and iterative development
- Testing AI vision and navigation systems in real-world scenarios
- 活動:
- 實驗室:建立與測試專案組件
- Exercise: Blind AI navigation (Integrate navigation and safety features into the project)
第16週:專案展示與評估
- Topics:
- Presenting assistive technology solutions
- Evaluating accessibility and usability
- 活動:
- 專案展示與同儕評審
- 提交最終專案
- Exercise: Advanced AI applications (Demonstrate a fully functional assistive technology prototype)
評估與計分
- 参与度與出勤:15%
- 每週練習:45%(包含 Blind spatial intelligence, Blind AI navigation, Assistive device design, Emotion, mood, artificial intelligence, G4H app creation, Algorithm, Advanced AI applications)
- 實驗室與作業:40%
學習成果
課程結束後,學生將能夠:
- 設計並實現用於障礙偵測、危險識別與安全導航的AI視覺系統。
- 使用空間智慧與Voronoi路徑規劃,開發針對視障人士的輔助科技。
- 使用AI與多模態數據創建認知與情感支援系統。
- 整合AI、VR與遊戲化技術,開發創新的無障礙解決方案。
- 應用運籌學與優化技術解決現實世界的輔助科技挑戰。
- 建立並評估滿足無障礙需求的全面輔助科技原型。
注意事項
- 無障礙:所有課程材料將提供無障礙格式(例如支援螢幕閱讀器的文件、字幕影片)。
- 倫理考量:學生將探討AI在無障礙中的倫理影響,包括隱私、偏見與包容性。
- 客製化:學生可根據特定人群(例如視障、認知障礙)或環境(例如城市、鄉村)調整專案。
沒有留言:
張貼留言