隨著柔性制造系統的廣泛應用和物流自動化運輸系統的快速發展，AGV技術得到了快速發展。從一開始對單臺AGV的研究，發展到了對多AGV組成的物流系統的研究。而多AGV路徑規劃作為直接影響多AGV系統整體性能的重要部分，一直倍受廣大學者的關注。隨著研究的深入，國內外學者提出了很多計算模型和策略。韓國的Jung Hoon Lee等人將兩階段的交通控制策略應用于多AGV的無碰規劃，劉國棟等提出了多AGV調度系統中的兩階段動態路徑規劃的方法。兩階段控制策略離線生成路徑庫，減少了在線運算的負擔，但是隨著節點數的增多，動態規劃的負擔加重，不適用于大規模多AGV系統。其他如Petri網，遺傳算法，Tabu Search算法（禁忌搜索算法）等策略和算法，在系統節點數增多的情況下，也有同樣的缺陷。為了有效地共享系統路徑，時間窗(Time-window)方法被提出并用于解決多AGV最優路徑問題。然而使用時間窗實現多AGV路徑規劃也是一個NP完全問題，并且在使用時間窗的模型中，獲得時間窗的AGV占用路徑時間過長，容易導致關鍵路段發生擁堵，降低系統效率。
With the wide application of flexible manufacturing systems and the rapid development of logistics automated transportation systems, AGV technology has developed rapidly. From the beginning of the research on a single AGV, it has developed to the research on the logistics system composed of multiple AGVs. The multi-AGV path planning, as an important part that directly affects the overall performance of the multi-AGV system, has always attracted the attention of many scholars. With the deepening of research, scholars at home and abroad have proposed many computational models and strategies. South Korea's Jung Hoon Lee et al. applied the two-stage traffic control strategy to multi-AGV collision-free planning. Liu Guodong et al. proposed a two-stage dynamic path planning method in a multi-AGV dispatching system. The two-stage control strategy generates the path library offline, which reduces the burden of online computing, but with the increase of the number of nodes, the burden of dynamic programming increases, which is not suitable for large-scale multi-AGV systems. Other strategies and algorithms such as Petri Net, Genetic Algorithm, Tabu Search Algorithm (Tabu Search Algorithm), etc., also have the same defects when the number of system nodes increases. In order to effectively share the system path, a time-window method is proposed and used to solve the optimal path problem of multiple AGVs. However, the use of time windows to achieve multi-AGV path planning is also an NP-complete problem, and in the model using time windows, the AGVs that obtain the time windows occupy the path for too long, which may easily lead to congestion in key road sections and reduce system efficiency.
1) 靜態環境中確定AGV 路徑規劃
AGV 路徑規劃在智能控制系統中具有重要作用， 對于保證工作的安全性來說具有重要意義。一直以來，很多學者都對此進行孜孜不倦的探索，這也是機器人學中最新最熱的內容之一。主要研究的是在障礙物的環境下，機器人如何尋找到目標，也就是選擇合適的路徑規劃。智能控制下的AGV 路徑規劃較為重要的兩種形態，靜態環境中的路徑規劃以及動態環境中的路徑規劃。
靜態環境下的路徑規劃是假定在環境信息未被完全掌握的情況下，機器人是通過怎么樣的路徑感知環境，并且運用局部區域傳播算法。因此這種路徑一般會在環境中僅存在靜態已知障礙物的情況下被采用。但是要分析靜態環境中AGV 路徑規劃，需要解決的一個問題是在這種環境中什么樣的路徑才能夠被認為是合理的?？偠灾?，能夠使AGV 系統實現控制的就是合理路徑。合理的路徑由路徑的平滑程度決定，路徑越趨于平緩，則AGV 系統將會更容易實現。此時可以將路徑分為四個種類，第一類平滑程度非常低，表現為路徑的不連續性，此時很多存在位置會表現突變的特性， 這種情況下AGV 系統不容易被控制，因為這些曲線不連續，無法對其追蹤。第二類，這類曲線相對于第一種來說具有連續性，但是在切線方向有時也會發生突變現象。此時也不能夠被AGV 系統控制。第三類，這類曲線不僅具有連續性的特點， 而且還能在切線方向保持連續性，因此是較為合理的路徑規劃，一般情況下也常常被采用。第四類，將以上三類曲線的優點都集于一身，但是要生產這類曲線十分復雜，因此在實踐當中，這類曲線很難被采用。
1) Determine AGV path planning in static environment
AGV path planning plays an important role in the intelligent control system and is of great significance for ensuring the safety of work. For a long time, many scholars have been tirelessly exploring this, which is also one of the latest and hottest contents in robotics. The main research is how the robot finds the target in the environment of obstacles, that is, chooses the appropriate path planning. There are two important forms of AGV path planning under intelligent control, path planning in a static environment and path planning in a dynamic environment.
The path planning in the static environment assumes that the robot perceives the environment through what path when the environmental information is not fully grasped, and uses the local area propagation algorithm. Therefore such paths are generally taken in situations where there are only static known obstacles in the environment. However, to analyze the AGV path planning in a static environment, a problem that needs to be solved is what kind of path can be considered reasonable in this environment. All in all, what can make the AGV system control is a reasonable path. A reasonable path is determined by the smoothness of the path. The smoother the path, the easier the AGV system will be. At this time, the path can be divided into four types. The first type has a very low degree of smoothness, which is characterized by the discontinuity of the path. At this time, many existing positions will show the characteristics of sudden changes. In this case, the AGV system is not easy to be controlled, because the These curves are discontinuous and cannot be traced. The second type, this type of curve is continuous relative to the first type, but sometimes abrupt changes occur in the tangential direction. At this time, it cannot be controlled by the AGV system either. The third type, this type of curve not only has the characteristics of continuity, but also maintains continuity in the tangential direction, so it is a more reasonable path planning, and is often used in general. The fourth type combines the advantages of the above three types of curves, but it is very complicated to produce such curves, so in practice, such curves are difficult to use.
2) Path planning determined in a dynamic environment
Path planning in dynamic complex environments is different from path planning in static environments. Because after the environment changes, a lot of information cannot be grasped, and it is impossible to ensure optimality in this case. When planning a path, a balance should be made between safety and timeliness. In a more complex environment, no matter which performance index is applied, three aspects of target attraction, dynamic security and time constraints must be considered.