REFILLS proposes to develop solutions allowing robots to improve logistic processes in a supermarket, revolutionising their current structure. The conventional in-store logistics processes are split in three scenarios. In the first scenario, mobile robots would inspect shelves and generate semantic environment maps for layout identification and store monitoring. The second scenario employs robots for three tasks: autonomous sorting of cases from mixed case pallets in the backroom; autonomous transportation of trolleys from the backroom to the shop floor, and assistance to clerks. In the third and final scenario, the robots handle a wide variety of products and refill shelves autonomously.
Supermarket clerks perform a number of time-consuming, repetitive, unergonomic, monotonous and wearing tasks. REFILLS aims at providing physical and cognitive help by performing autonomously some operations (shelf monitoring, depalletizing, pre-sorting, transporting) and assisting the clerks in other operations (spot finding, products replenishment).
Even while automation is improving the experience of customers at retail shops both in terms of ordering and customer comfort, there is still a lack of automation in the logistics management of the retail stores. Most of the logistics costs arise from items handling, items transportation, shelves replenishment and backroom management. The automation introduced by REFILLS may not only reduce the cost of logistics, but it also enables retail stores to function as a hub that can be used by online and delivery services thanks to robotic system solutions.
REFILLS aspires to create autonomous, cooperating and collaborative logistics robots for supermarkets by facing the following scientific and technological objectives:
- Cognition-enabled control for shelf management — Autonomous mobile robots with visual sensors generate large amounts of valuable data. These data are used to build semantic maps of the store and manipulation knowledge bases to perform combined task and motion planning for in-store manipulation activities.
- Perception and control for interacting with objects — Multimodal sensing capability (e.g. visual, force, tactile and proximity) is exploited to allow the robot to recognize objects and find their pose, as well as to safely grasp and manipulate a large variety of objects even in the presence of uncertainties.
- Design of a unique set of robotic modules and integration for in-store logistics application — A new set of robotic modules, built on existing, commercially available platforms and controllers and conceived with the following criteria: low-cost, modularity, control parametrization, optimal kinematics, dependability in operation.