快速重新規劃無濺擾軌跡與流出液體動態補償控制於移動機械臂應用

Abstract

在工業自動化領域，雖然處理剛性物體已經取得了成功，但是操控像液體這樣可變形的物體仍然是一項挑戰。本論文專注於利用移動式機械手臂（mobile manipulator）提出創新的解決方案，以應對液體運輸和澆注任務。為應對液體運輸挑戰，我們利用全身模型預測控制（whole-body MPC）將機械手臂和移動底座的運動進行同步。同時提出了基於模型預測控制(MPC)的架構以生成最佳參考無濺擾軌跡，利用濺擾抑制的模擬驗證展示於移動機械手臂的適用性，並且優於現有基於軌跡最佳化(TO)的方法，成功減少了44％的方均根誤差（RMSE）。對於澆注任務，我們將複雜的導航和定位算法整合到自定義移動底座中，利用視覺補償來校正低導航精度引起的基座偏移。實現了一個前饋補償器來達成澆注位置控制任務，通過影像處理技術驗證，取得了令人滿意的結果。這些進展為智慧機器人中的液體操作任務提供了有希望的解決方案，增強了工業自動化能力。

In industrial automation, while handling rigid objects has seen success, manipulating deformable objects like liquids remains challenging. This thesis focuses on introducing innovative solutions for liquid transport and pouring tasks using a mobile manipulator (MM).To address liquid transport challenges, we synchronize the manipulator and mobile base using whole-body Model Predictive Control (MPC). The proposed slosh-free MPC framework generates optimal reference trajectories, outperforming existing Trajectory Optimization (TO) methods. It demonstrates its suitability for MM with slosh suppression through validation simulations, achieving a 44% reduction in Root Mean Square Error (RMSE). For pouring tasks, we integrate well-established navigation algorithms into a customized mobile base, utilizing visual compensation to correct the base shift due to the inherent uncertainties. A feedforward compensator is implemented to achieved pouring position control task, yielding a satisfactory result validated through image processing techniques. These advancements offer promising solutions for liquid manipulation tasks in intelligent robotics.