Year 2020: 15 minutes at EROSKI* Distribution Center
Through this storytelling exercise, a future situation is described: a new product arrives to the Distribution Center and is successfully dealt with using the PICK-PLACE solution. *EROSKI is a Spanish supermarket chain that has several Distribution Centers and ULMA is one of its providers.
New product reference arrives to the Distribution Center
A new reference has to be defined in the Warehouse Management System (WMS) of Eroski Distribution Center. It is a perfume that will be distributed during the Christmas campaign, and the provider has facilitated the following information:
- Number of different variants (the pefume comes in 3 volumes: small, medium and big)
- Type of packaging: they will come in cardboard boxes with a cellophane cover
- Sizes and weight of the packaging (for each of the 3 variants)
- Arrangement of the products in the box when they arrive to the packaging area in the Distribution Center. They will come in carton boxes of AxBxC dimensions, in 3 layers of DxE units. Between layers there will be a cardboard sheet spacer
With this information and taking into account robot affordance, the Robot Grasping Planner decides automatically the best grasping strategy (points, parameters) for each item. This information will be used during the grasping operation, adapting it based on acquired real-time information.
Required abilities: Off-line and online grasping planner
The WMS has to manage a set of orders that have to be delivered to some small shops in the province. The orders are composed of multiple references and there are several units for each reference. The WMS has launched the commands to the Automatic Warehouse Dispatcher (AWD) to take the inbound boxes (with the items to be delivered inside) to the packaging area.
The Package Planner is in charge of:
- Calculating the size of the final carton box where all products have to be introduced (they use 5 different standard sizes
- Planning the disposal of each product in the box, as well as the order in which they have to be introduced. They call it mosaic
- It has distributed the packaging process between the human operator and the robot operator according to different criteria:
- The weight of the product
- The type of envelop and fragility
- The number of items
Required abilities: Logistic planner / Human-robot task allocation
Inbound boxes start arriving to the end-of-line station, as well as the outbound boxes where the references have to be put inside. There, the worker and the robot will start the picking-packing operations to prepare the multireference orders. The robot is a commercial-off-the-self robot offering some compliance features.
Inbound boxes are on the upper conveyor, outbound boxes in the bottom conveyor.
Required abilities: Safe manipulators for collaboration with people
The robot is equipped with a multifunctional gripper that allows handling products of different shapes, materials and sizes. It combines grasping and adherence technology in the same device, allowing dexterous manipulations, needed in both picking and packaging operations. To increase robotic arm reachability it is mounted on top of a mobile platform which provider higher flexibility to the solution.
Using the information of several embedded force and sliding sensors the grasping control algorithm allows manipulating a huge range of products.
A vision system on top of the inbound area allows the robot identifying the pose of the objects inside the inbound box.
A vision system on top of the outbound area allows monitoring the configuration of the mosaics and re-plan the mosaic if needed (sometimes the mosaic collapse or the human worker makes some mistakes in placing the part he/she is in charge of handling).
The robot takes a part from the inbound box and places it directly in the outbound box whenever possible. One of the orders includes products that can be placed in any position (randomly) as there is not any constraint due to size or fragility. In this case, the robot just drops the product inside the box (the throughput is much higher in these cases). But more often, it has to place with care in the order previously defined by the Planner.
Required abilities: Dexterous gripper / Flexible grasping / Robust object recognition
Though the planner tries to avoid any interference among the human and robot trajectories by scheduling their activities properly (time and space separation), it is not always possible due to the uncertainties introduced by human behaviour.
The worker has dropped the object he/she has taken out of the box. When trying to take it from the floor, he/she enters in the working area of the robot. Luckily, there is a perception system that tracks human activities (position, current and next operation, etc.) and modifies the robot trajectory accordingly to avoid any risky situation (e.g. collision or entrapment).
Furthermore, the force exerted by the robot in case of contact with an obstacle (object or person) is monitored and the control system will stop immediately in case of detection such an event.
Required abilities: Algorithms for safe motion in the presence of people / Dependable people detection and tracking techniques / Intrinsic safety for collaboration with people
The vision system on top of the outbound area has detected that the target position is not free. The Package Planner tries to find a new mosaic configuration based on the new product disposal and taking into account those products that remain pending for packaging.
When it is not possible to find such an alternative configuration, the human is requested to recompose the original mosaic.
Required abilities: Action Planning: Methods for whole system optimisation and risk aware scheduling and rescheduling
The supervisor has called the worker for a consultation. She/He has to leave the packaging station. He push a button and the Package Planner defines a new mosaic configuration and reschedules the remaining activities (including those originally assigned to the human worker), so the robot can continue completing the order even without the human presence (unless a product is unmanageable by the robot, e.g due to size, weight, fragility or any other characteristic).
This kind of operations (without human presence) are also implemented in two special packaging areas where high value products and pharmaceutical products are manipulated. When there is not human presence, the robot increases the working speed. In case of any error or unexpected situation the robot cannot handle, the presence of humans is requested but the process is not stopped and the worker can approach to the robot safely.
Required abilities: Action Planning: Methodologies for efficient on-line task allocation and scheduling / Methods for whole system optimisation and risk aware scheduling and rescheduling
The robot has introduced the last reference in one of the outbound boxes. It informs the WMS, which takes the box out of the station and organizes all resources to prepare a new order.