無人機對於動態未知之目標物體之估測與追蹤控制

Estimation and Tracking of a Moving Target by Unmanned Aerial Vehicles

李俊明(Jun-Ming Li)

指導教授 ： 程登湖

國立交通大學/工學院/機械工程系所/碩士(2019年)

摘要

本研究使用無人機針對未知動態目標物體進行視覺追蹤控制並且對目標物體的動態進行即時估測，利用You-Only-Look-Once (YOLO) 神經網路於影像中辨識出所需追蹤之目標物體並生成對應的bounding box，獲得目標物體與無人機的相對資訊，進而實現由影像達成即時的估測與回授追蹤控制。追蹤之目標物體的動態是未知且無法被量測的，本研究使用等速度動態模型去模擬目標物體的動態並結合Unscented Kalman Filter (UKF)進行估測。由於UKF中的process noise是不確定的且易受系統估測的穩定度影響，因此本研究建立一個移動窗口(moving window) 紀錄部分UKF的估測資料同時對process noise變異數矩陣進行線上估測。

當無人機的畫面中無法獲取目標物體bounding box資訊時(e.g.,YOLO辨識失敗、出現遮蔽物)，估測的狀態可能會隨時間而發散。因此我們會使用先前UKF估測而得的目標物體動態去擬合出一條目標物體的移動軌跡，在無法獲取bounding box資訊時作短時間的預測，如此一來便能提供追蹤控制器更穩定的回授資料，並且加入估測而得的目標物體速度減少追蹤誤差增進追蹤表現。本研究所使用的追蹤控制器為非線性模型預測控制器(NMPC)，與傳統的回授控制器相比，非線性模型預測控制器考量了攝影機視野與控制指令上下界的限制，可增加目標物體於追蹤中維持於攝影機視野中的機率，並且使無人機的追蹤動態更為穩定平滑。由於本研究並未假設目標物體有任何動態上的限制，因此此追蹤控制器可用於追蹤各種動態目標物，模擬與實驗結果展示了估測與追蹤控制在各種情境上的表現。

關鍵字

無人機 ； 追蹤控制 ； 視覺追蹤 ； 狀態估測

並列摘要

An image-based control strategy along with estimation of target motion is developed to track dynamic targets without motion constraints. To the best of our knowledge, this is the first controller that can track moving targets based on a bounding box of the target detected by a deep neural network using the you-only-look-once (YOLO) method. The features generated from the deep neural network can relax the assumption of continuous availability of the feature points in most literature and minimize the gap for applications.

One of the challenges is that the motion pattern of the target is unknown and modeling its dynamics is infeasible. To resolve these issues, the dynamics of the target is modeled by a constant-velocity model and is employed as a process model in the Unscented Kalman Filter (UKF), but process noise is uncertain and sensitive to system instability. To ensure convergence of the estimate error, the noise covariance matrix is estimated according to history data within a moving window.

Another challenge is that when occlusion is present, the bounding box of the moving target becomes unobtainable and makes the estimation diverge. To solve this, a motion model derived by quadratic programming is employed as a process model in the UKF, wherein the estimated velocity is implemented as a feedforward term in the developed tracking controller so as to enhance the tracking performance.

A Nonlinear Model Predictive Controller (NMPC) is designed in this work to achieved target tracking. Compared to the existing Image-based Visual Servo (IBVS) control methods, NMPC consider the constraints of the image features and control input to ensure better tracking performance. Since no motion constraint is assumed for the target, the developed controller can be applied to track various moving targets. Simulations are used to demonstrate the performance of the developed estimator and controller in the presence of occlusion. Experiments are also conducted to verify the efficacy of the developed estimator and controller.

並列關鍵字

UAV ； tracking control ； visual control ； state estimation

參考文獻

[1] V. Chitrakaran, D. M. Dawson, W. E. Dixon, and J. Chen, "Identification of a moving object’s velocity with a fixed camera," Automatica, vol. 41, pp. 553–562, 2005.

[2] J. Schlichenmaier, N. Selvaraj, M. Stolz, and C. Waldschmidt, "Template matching for radar-based orientation and position estimation in automotive scenarios," in Proc. IEEE MTT-S Int. Conf. Microw. Intell. Mobility (ICMIM), Nagoya, Japan, 2017, pp. 95-98.

[3] A. Stroupe, M. Martin, and T. Balch, "Distributed sensor fusion for object position estimation by multi-robot systems," in Proc. IEEE Int. Conf. Robot. Autom., Seoul, South Korea, 2001, pp. 1092-1098.

[4] A. Kamthe, L. Jiang, M. Dudys, and A. Cerpa, "Scopes: Smart cameras object position estimation system," in EWSN ’09, Berlin, Heidelberg, 2009, pp. 279-295.

[5] H. S. Ahn and K. H. Ko, "Simple pedestrian localization algorithms based on distributed wireless sensor networks," IEEE Trans. Ind. Electron., vol. 56, no. 10, pp. 4296–4302, Mar. 2009.