Dynamic AGV routing depending on sensor-based collision avoidance

Abstract

AGVs have seen an upward trend in development over the last 60 years. The technology has developed from mechanical bumpers and guided wire navigation to contactless sensors and free navigation technique in the current age. Further, the control on AGVs has moved from central control system to local intelligence which opens up various possibilities with respect to operations as well as applications. The growing trend of AGVs has been due to the sudden growth in digital technology and the ever-increasing demand to reduce human intervention in operations. This has resulted in increased research regarding the implementation of intelligent AGVs in areas of application that have not yet been explored, namely, light metal and forging industry. The major reasons for indulging in autonomous equipment are, increased productivity, reliability and safety since human involvement is either eliminated or largely reduced. However, a major share of the research about intelligent AGVs has been confined to warehouses and port logistics. Therefore, through this research, another area of application is investigated, namely, the light metal and forging industry, and more specifically, the potroom of an aluminium smelter. Hence, the objective of the research is as stated: With the introduction of intelligent AGVs, stochastic behavior needs to be addressed, that is, how would these AGVs react to disturbances created by such random human behaviour and process interference? Therefore, the research focusses on the routing problem in such situations which are dynamic in nature. The research aims to provide a planning approach in terms of dynamic AGV routing under the assistance of a sensor based system that can detect obstacles. The dynamic re-routing of AGVs is addressed using a mathematical formulation as well as a graphical representation. In order to solve the problem at hand, the graphical approach is followed and the objective has been simplified for research purposes. It is simplified as: For instance, if a certain pathway is blocked in the potroom of an aluminium smelter due to such stochastic behavior, how would the AGV find the optimal path? An algorithm is devised in order to answer the above research question and further implemented in Python. Various scenarios of stochastic disturbances is analysed and evaluated accordingly. Therefore, this research develops an algorithm and a subsequent model that is implemented using Python which is used to evaluate the routing of an AGV in the presence of stochastic behavior. It acts as a proof of concept for the problem at hand as it restricts the work to a simplified situation of a single AGV operation. Although this research uses the case of a light metal and forging industry, the same can be applied to industries with similar challenges such as, cement industry, power generation industry, aerospace, construction and so on. Cement, construction and power generation industry deal with environments similar to the light metal and forging industry. Further, in all these application areas, the use of AGVs for material handling would improve productivity and reliability, while improving safety of operations as well. This research focuses on a simplified situation, however, the basis of this work can be further extended to solve a more detailed real world scenario with a fleet of AGVs, and this can be done by the use of advanced heuristics.