未知环境中机器人巡航问题之研究 - 淡江大学.ppt

*/48 Particle Filter for Robot Localization Applied Electronics Technology, NTNU */48 Applied Electronics Technology, NTNU KF-Based State Estimation Kalman filter (KF) estimator Adapt the concept of recursive prediction and update estimate process. Prediction: (Linear prediction of states and measurements) Update: (Linear update equation for system states) Example: KF-based SLAM State vector of robot (camera) State vector of static objects State vector of moving objects Measurement models Ojk=[ojk sjk]T */48 Applied Electronics Technology, NTNU The camera is presumed to move at constant velocity (CV); The acceleration is caused by an impulse noise from the external force. Visual Sensors for SLAM Monocular vision Binocular vision Velocity noise: Camera carried by robot Free-moving camera */48 Applied Electronics Technology, NTNU 淡江機電系機器人實驗室 機器人視覺式同時定位、建圖、與移動物體追蹤(visual simultaneous localization, mapping, and moving-object tracking) SLAM SLAMMOT */48 Monocular SLAM Applied Electronics Technology, NTNU */48 Monocular SLAM Applied Electronics Technology, NTNU */48 Monocular People Detection and Tracking Applied Electronics Technology, NTNU */48 Binocular SLAM Applied Electronics Technology, NTNU */48 Binocular SLAM Applied Electronics Technology, NTNU */48 Binocular SLAM Applied Electronics Technology, NTNU */48 Applied Electronics Technology, NTNU Differential-drive Mobile Robot [2010] Binocular Vision Laser Range Finder Wheel Encoder PC-based Controller Sonar Mobile Robot */48 Visual SLAM of Mobile Robots Applied Electronics Technology, NTNU */48 Visual SLAM of Mobile Robots Applied Electronics Technology, NTNU */48 Visual SLAM of Mobile Robots Applied Electronics Technology, NTNU */48 結論與未來研究議題 理論上，同時定位、建圖、與移動物體追蹤的問題已經有解 實現技術方面，仍有挑戰性： 辨識移動物體 使用移動感測器偵測與追蹤移動物體 實際應用時，依需求選擇完整求解或簡化求解 解答的一致性(consistency) 計算複雜性(computational complexity) 與路徑規劃、運動控制器的結合 新感測器的發展與應用 新的應用領域 Applied Electronics Technology, NTNU */48 Visual SLAM of Robot Vacuum Cleane