對於自走車來說，要如何從一個佈滿各種障礙物的環境中，找到一條合適的路徑，並確保自己能成功的從起始點到達目標點，且行經的路程要是最短的，是個值得研究的方向。 本篇文章提出一種結合人工魚群演算法(AFSA)和快速擴展隨機樹(RRT)的新型演算法。和過去傳統的人工魚群演算法用於規畫路徑不同的地方在於，它是藉由類似樹枝生長的方式來去增加延伸點，在數個增加的點相互比較後，選擇最佳的延伸點去當做下一條魚的移動位置。 這個方法改善了人工魚群演算法後期收斂時間慢、不易收斂於全體最佳解的缺點，加上基本的快速擴展隨機樹演算法每次在規畫路徑上並不是那麼穩定，所以本文就利用模擬快速擴展隨機樹演算法的方式去加入到人工魚群演算法，再配合危險度地圖觀念去有效閃避障礙物。 經過測試，我們可以由數據上顯示這個新式演算法比單一個人工魚群演算法或是快速擴展隨機樹演算法應用在路徑規劃上都好且更穩定，且路徑長度也能是最短的。 ;In terms of self-propelled vehicles, how to find a suitable route from the environment filled with varied obstacles and how to ensure the target point from the starting point with the minimum passing away can be successfully reached is a direction worthy of research In this thesis, the new algorithm in a combination of Artificial Fish Swarm Algorithm (AFSA) and Rapidly-Exploring Random Tree (RRT) is proposed. Compared with the traditional AFSA, the difference of the new algorithm in path planning is it is by similar branch growing up to increase extension points, and after the increase points are compared with each other, the best extension point is chosen as the point of the next moving fish. The new algorithm improves the shortcomings of the later-period AFSA in slow convergence time and being hard to be converged in the optimal solution. In addition, the basic RRT algorithm is not so stable on each path planning, so in this paper, the simulation of the RRT algorithm is used to be included in AFSA, with the concept of danger degree map to effectively dodge obstacles. After testing, it is shown that the data of the new algorithm on path planning are better than those of either AFSA or RRT and the path is the shortest.
