近年來隨著感測、通訊及控制技術的進步，不同型態的載具開始朝向自動駕駛的方向發展，自主航行船舶亦成為熱門研究項目之一。實現自主航行需包含許多項目的開發，其中，在障礙物閃避又尤其重要。避障的規則需遵守國際公約組織IMO所提出之國際海上避碰規則公約(COLREGs)來避免海上交通事故的發生，本研究結合人工勢場法和最接近點法，進行船舶自主避障的開發及研究，目標為根據不同的船體大小、性能、障礙物大小及可容忍與障礙物的距離來調整系統參數，達到自主避障效果。此方法利用兩船的行進方向與速度，取得最接近點距離(DCPA)和最接近點時間(TCPA)，並結合人工勢場法，推算出導引船舶安全行進方向，最後將成果模組化應用於5公尺及13公尺等兩種不同船型進行模擬測試。根據避碰對象分為兩種模式，靜態避碰模式和動態避碰模式。當偵測範圍中存在靜態障礙物時，在勢場中產生排斥力，目標點產生吸引力，依合力方向修正導引點，協助船舶閃避障礙物；動態避碰則結合平行修正導航法偏移導引點，並考慮障礙物與船體自身的相對關係，使船舶避碰滿足國際海上避碰規則公約。本研究使用操船模擬器所建置之每秒30次的風、浪、流模擬，同時考量船體運動特性來驗證演算法之可行性。根據所模擬的船體操縱性，在三級風浪狀況下，可以安全閃避來自各方向的動態障礙物。由於避碰方法考量環境的干擾和船體動力學，本研究成果未來將可依不同的船體大小及性能，改變避碰參數以應用於不同船型的導航系統之中。

Along with the advancement of the sensing, communication and control technology, autopilots have been developed for various types of vehicles. To achieve autonomous piloting of marine vessels, the obstacle avoidance is a safety issue to be investigated. The obstacle avoidance functions must follow the Convention on the International Regulations for Preventing Collisions at Sea (COLREGs) which is addressed by International Maritime Organization (IMO) to avoid marine accidents. This study provides an efficient way to avoid obstacles by combining the artificial potential field method (APF) and closest point of approach method (CPA). Obstacle avoidance could be achieved by implementing this method and modifying system parameters according to the size and performance of vessels, size of obstacles and the distance between vessels and obstacles. A safest sailing direction could also be provided by utilizing artificial potential field, distance at the closest point of approach (DCPA) and time to closest point of approach (TCPA), which are derived from two vessel’s velocity and orientation. Results are then applied into two vessel models with length respectively of 5 m and 13 m for performance simulations. The algorithm includes two modes, the static avoidance mode and the dynamic avoidance mode. A repulsion force is generated in the potential field if an obstacle exists within the detection range, while a target point generates attraction force. The synthesis of potential field forces is employed to assist in modifying guidance point for a vessel to avoid obstacles. To avoid the local mini-mum, dynamic avoidance mode combines parallel correction navigation method additionally, and also considers the related distance between vessel and obstacle by offsetting the guidance points. Hence, the collision avoidance is compliant with COLREGs. The feasibility of this research is verified by implementing vessel’s motion characteristics into a ship simulator, which involves simulations of the wind, wave and current influences. The result shows that the vessel is able to avoid dynamic obstacles in sea state 3. Since both environmental interference and vessel dynamics are considered, the result of this research could be applied into navigation system of different vessels by modifying their vessel sizes and performance parameters.