Delft University of Technology
Faculty Mechanical, Maritime and Materials Engineering
Transport Technology / Logistic Engineering

B.R.C. Herijgers Design issues of AGV systems
Literature survey, Report 2000.LT.5433, Transport Technology, Logistic Engineering.

Material handling is the transfer of materials in the most economic manner from one location to another. These movements must be made safely, accurately, efficiently and without damage to the materials. Option for material handling equipment is an automated guided vehicle (AGV) system. Automated Guided Vehicles are now becoming popular in automated material handling systems.

Advantages of the AGV systems are its diversity of use, flexibility, ability of automatic interface with other facility systems and improvement of working conditions. Disadvantages are problems with design issues during the implementation, the costs involved in applying the system and the possibility of failure.

The design of an AGV system affects the performance of the system and the design can be very complicated. A number of interrelated decisions must be made in order to exploit the full benefits of the system. First the kind of automated guided vehicle has to be decided taking into account the characteristics of the vehicle like vehicle speed, load capacity and their docking equipment. AGV system design concerns also the determination of the guide path layout. The design of the guide path depends on three factors: location of pickup and delivery points, guide path layout and path type. The amounts of vehicles that will operate in the system have to be accurately chosen. Analytical determination as well as simulation can be used to find the required fleet size of the system. Beside these physical design aspects operational issues like vehicle dispatching, routing and scheduling are important. It is important to know in advance the kind of different dispatching and routing rules that can be used before designing an AGV system. These rules are in general based on traditional algorithms from the operations research area.

The dispatching rules used in AGV systems can be classified in several ways. The most common used classification is work-centre initiated rules versus vehicle-initiated rules. Work centre initiated dispatching deals with the problem of matching a task to a vehicle when multiple vehicles are idle and waiting for a task assignment. Vehicle initiated dispatching deals with the situation that a vehicle has the choice of a task when multiple loads are waiting for pickup simultaneously at different locations in a facility. The vehicle-initiated rule can be further classified into source-driven rules and demand-driven rules. The source driven rule operates on a push concept: an idle vehicle selects a part to move from an output queue that has the highest priority. The demand driven rule operates on a pull concept: an idle vehicle selects the part that has the highest demand from its succeeding work centres. Other classification is the distinction between dispatching rules based on one attribute and dispatching rules based on several attributes. A single-attribute dispatching rule cannot, in general, satisfy all the objectives. Some of the objectives may be in conflict due to the interdependencies within the manufacturing process. Multi-attribute dispatching rules take into account different criteria simultaneously in order to produce better performance. Most of the dispatching rules wait until a vehicle becomes empty and then assigns a transportation task to a vehicle. These dispatching rules do not take in account the status of the already assigned vehicles during the assignment process of a new task. As an alternative the Bid-based dynamic dispatching is proposed.

In routing a distinction can be drawn between static routing and dynamic routing. If routing is static, the path taken by an AGV between two given nodes is fixed. Dynamic routing is based on the current state of the system therefore dynamic routing offers a higher degree of flexibility. Dynamic routing can be classified into three major classes: centralised, local, distributed.

The last section of this report introduces a dispatching rule integrated with a routing technique. The dispatching rule that will be used is the Bid Based Dynamic Dispatching rule (B2D2) in combination with a semi-dynamic routing strategy. Normally first a transportation task will be matched with an AGV based on the shortest path. Subsequently the shortest route has to be scheduled to perform the transportation task in reality. In the case that will be given these two parts will be integrated. A transportation task will only be dispatched if a route can be scheduled that is not in conflict without earlier dispatched vehicles. The proposed procedure can be a starting point for further investigation. A simulation study can provide insight if the expected benefits, performance that can be exploited by use of this technique. The results can be compared with non-integrated dispatching and routing technique. The situations in which the proposed algorithm can be useful can be defined.

Reports on Logistic Engineering (in Dutch)
Modified: 2001.01.05; , TU Delft / 3mE / TT / LT.