The objectives of this research are to provide a cost-effective solution for local industries, that can readily be installed on running manufacturing setup, minimizing change cost. The solution should carry a load up to 30Kg, autonomously schedule jobs of working shift and route material from one node to other. There were three major phases of project completion i.e., 1) Scheduling Algorithm Design 2) Routing Algorithm Design 3) Hardware Design and 4) Control Design Each of the phases are discussed in this section
Mover works on its two prominent features i.e., smart scheduling and smart routing. Due to these two features, it is designed to save time and cost hand in hand. Two algorithms were studied and applied, Shortest Job First (SJF) for smart scheduling and A* for smart routing.
2.1 Scheduling Algorithm: Shortest Job First (SJF)
SJF Scheduling Algorithm is particularly suitable for lot jobs that already have runtimes. Every process in a ready queue can be performed with this algorithm based on the shortest process time. This will lead to lower wait times for each process, and thus the reducing the average processing time. Different scheduling algorithms have various functions, and a particular class of processes may be preferable to another algorithm. We must consider the properties of each algorithm when choosing which algorithm to use in any particular situation. This proposed algorithm fulfils all programming requirements, such as maximum machine use, maximum performance, minimum turnaround time, minimum waiting time.
SJF initiates asking for a parameter of working hours, this first step is crucial as it decides the working shift of industry. The algorithm will try to schedule maximum number of jobs in given shift time. The second parameter is no. of job along with the machine type and processing time to schedule under given shift. The machine then sorts all jobs w.r.t type of machine and then schedule each machine with minimum processing time. The flow diagram of algorithms is shown in (Fig. 1).
2.2 Routing Algorithm: A*
The purpose of routing a Guided vehicle is to use the shortest path and save as much time as possible for each job. The route chosen must be free of congestion, conflict- and deadlock and the route must be short idling for vehicles, to determine if there exists a route that can take a vehicle from its origin to the destination. A* algorithm is considered as one of the most frequently used shortest path planning algorithm, this algorithm is chosen to find two nodes on the shortest route the parameter in this algorithm is positive and real number.
A* algorithm makes the process of finding the shortest route, simple and also efficient. This is a revised form of the Dijkstra algorithm. In computing the difference between both the current stage and the goal point one considers the main features that include the actual cost of the path, area for a search is considered as a set of nodes. Each edge is linked to each side of the path so the search route could be defined by the nodes of the space on each side. A* looks for the shortest path by selecting the node sequence from the start node to the target node, the cost function of A* is given in Eq. 1. The grid needs to be in AGV’s path search area, and the AGV’s environment map is separated into several uniform squares, each of them having a particular position in the environment. The A* search function and flow can be seen in (Fig. 2).
f(n) = g(n) + h(n) (1)
2.3 Hardware Design
The Mover is designed to carry to load of approx. 30Kgs, thus; the hardware designs the initial design consist of a single cart, however; to divide the load effectively it was redesigned and divided into two parts; the car and the load cage. The car itself would be the major compartment which would hold all the electronics, and would have the mechanism for producing torque. The cage would be designed using trusses to ensure a cart which could bear at least a load of 30kg on a long-term basis. The final working model can be seen in (Fig. 3).
Stress analysis was conducted on two core components of the AGV. Firstly, the AGV cage that would hold the load would be the first critical point and second critical component would be the shaft that attached the wheels to the motors. This has to hold up to 60kg load, that would include 30kg of the AGV itself and 30kg of the load that the AGV would carry. The analysis showed the optimized result and validate the fulfilment of objective.
2.4 Control Design
(Fig. 4) is the basic block diagram of Mover, it shows the basic model of the system, which includes three sensors and two motors being controlled by our microprocessor, Raspberry Pi.
These blocks are further expanded upon in the following figure. In each case, the result of each individual control system would be stored within the controller, and the AGV would be able to execute commands using the inputs. For the distance control system, the ultrasonic sensors would detect whenever the AGV is facing an object that is less than 2 meters away from the AGV. To avoid confusion between an obstacle and the AGV guides, the AGV would slow down and continue motion until the distance becomes 4 inches, which is the minimum distance between the AGV and the obstacle.
As for the proximity sensors, it has three sensors placed on the front end of the AGV. The sensors would be fixed above the centre of the steel strip, and on either side of the strip. If the left sensor turns off, it means the AGV has steered too far to the left, and must be turned right. If the right sensor turns off, it means the AGV is too far to the right, and must be steered left. The RFID tags would be used to identify and store in memory the current workstation that the AGV is at; this would be used to find the distance between the current workstation and all the other workstations. The control diagram of individual transducers is given in (Fig. 5).
In a nutshell the complete working of the mover can be seen in flow chart below. After scheduling when the job is assigned to it, the strain gauge sensor check wheatear weight is on the cage, after confirmation, RFID sensor checks the current position node to its target goal and calculate the shortest path, then the vehicle starts moving with the help of guided path if the obstacle object will be detected by an ultrasonic sensor that vehicle will be stopped if not then the vehicle will continue its motion and until it reaches the target location. The system flow diagram can be seen in (Fig. 6).