CORDIS - EU research results

Floor Washing Robot for Professional Users

Periodic Reporting for period 3 - FLOBOT (Floor Washing Robot for Professional Users)

Reporting period: 2017-01-01 to 2018-06-30

Supermarkets, as well as other industrial, commercial and public use premises, have huge floor surfaces that have to be cleaned daily or even several times during the day. Cleaning those surfaces is time demanding and requires human effort, in terms of repetitive actions. Therefore, floor washing activities are best suited for robotization. Currently, there exists no robot that satisfies the requirements of the professional users and cleaning service providers. FLOBOT addresses these requirements by integrating existing & new results to produce a professional, robotic scrubber-dryer. The system consists of a mobile robotic platform (the robot) and a docking station for automatically refilling the robot’s water tank, emptying the dirty water tank & recharging the battery. The robot and the whole concept has been validated in 4 different pilot sites: a large supermarket and a hospital in Italy, as well as in the Lyon airport and in a logistics warehouse in France.
Overall project objectives are summarised below:
• To develop an autonomous, professional washing machine, ready for real-world use shortly after project completion
• To develop FLOBOT as a robotic system that increases the quality of service and lowers the cost of operation, while preserving competitiveness and respecting workers’ rights & health, abiding to all related regulations
• To integrate low cost solutions for mapping, localisation, autonomous navigation, object detection & human tracking, building on research results and commercial know-how
• To develop user interfaces for easy mission & tasks reprogramming, and remote control of the robot
• To prove the usability of the robot by testing it in 4 real-world use-case scenarios

Conclusions: The FLOBOT has delivered a validated robot prototype that is ready for the last steps towards industrialisation and commercialisation. The system is complemented by a docking station, a tablet user interface and a data management system.
One of the first project activities (WP2) has been the identification of the user & manufacturer requirements. After defining the requirements, the conceptual design of the FLOBOT and the docking station was prepared. In parallel, issues related to safety, standards, risk analysis & future certification of the FLOBOT were also examined (WP3). Regarding hardware development, first a preliminary robot prototype was prepared. This, Prototype Zero (PT0), was used for several tests. Based on the interfaces defined and on the sensors selected, the CAD of the mobile platform and the docking station was prepared. This was used to build the first operational robot prototype (PT1) and the first prototype of the docking station. The PT1 went through several upgrades after the various testing phases of the project. In terms of software development (WP5), all modules were successfully delivered. The navigation & mapping module functionality is based both on a 3D Lidar and on conventional odometry. Visual floor inspection (for cleanliness) is based on data gathered either by the RGBD-camera or the stereo setup. A friendly user interface has also been prepared. This runs on an Android tablet. Regarding the proactive safety module development, this has also been finalized, as well as the software module responsible for interfacing to ERPs or similar legacy systems.
All modules were integrated on the final robot version and the robot was first tested in the lab (WP6) and then used for field validations in 2018. First real-world tests took place in FIMAP’s warehouse. Then, the robot was used for real-world testing in a Carrefour supermarket, in the Lyon airport, in a private cargo warehouse and in the Imola hospital.
FLOBOT has also been quite active in terms of dissemination activities (WP8). It has been presented in several events, including different editions of the IROS conference and the ISSA-Interclean exhibition. Information about the project was published in several newspapers and magazines. The project website & social accounts are regularly updated. A business plan has been successfully prepared and the industrial partners are working towards exploitation of the project results.
Main results include:
TRL 8: ERP interfacing module, qualified and ready for commercial exploitation
TRL 7: Robot hardware, mapping & navigation modules, human tracking module, tablet GUI, proactive safety module, Object & dirt detection modules, docking station & procedure. All of them are ready for exploitation by the commercial partners, requiring relatively small effort for optimisation and full qualification
TRL 6: Environment reasoning module. Module successfully tested. Requires further testing and optimisation before reaching the market, even though, this is an optional module that will not hinder commercialisation of the main FLOBOT system
The project answers the need of the cleaning industry for a robotic scrubber that can operate unattended for a long period of time. Progress beyond the state of the art is in several aspects. Regarding localisation & mapping, several modules were already available in the consortium before the start of the project, but the use of the 3D Lidar is a new approach for robots of this type. Regarding image analysis for floor cleanliness detection, the project is advancing the state of the art by using the approach described by Bormann et al. in 2013, adapted to the FLOBOT case. Additionally, regarding object detection on the floor, a basic plane detection & height based object detection algorithm were implemented. As an option, we also advanced the basic approach with an experimental algorithm that simultaneously segments & reconstructs the area of operation. This technique is still not applied in industrial prototypes.
Regarding human tracking for safety, FLOBOT’s human tracking module is capable of detecting & tracking multiple people. Safety during FLOBOT operation within a complex environment with humans is of primary importance. The solution proposed is the projection of warning signs on floors in an intuitive way, by controlling the position of these signs. The docking station design also includes mechanical design novelties.
FLOBOT is expected to have significant impact on labour aspects. Currently, professional cleaning is a hard labour, low skills job, where competitiveness & profit margin are at the expense of the workers’ conditions. Additionally, about 74% of the people employed in the sector are women and there are difficulties in the recruitment of new personnel. With the adoption of FLOBOT, the worker’s role will evolve from low skills to that of a use & maintenance operator of robotic machines. At the same time, the robot will relief the physically weaker workers from part of the hard labour.
FLOBOT is also expected to support the EU industry producing professional cleaning machines, that has to deal with the highly competitive Asian & American companies. The consortium end-users are large companies that, with the adoption of FLOBOT, will allow the FLOBOT manufacturer to cover an important market share. Furthermore, FLOBOT has already had a great impact in terms of Industry-Academia cross-fertilisation.
FLOBOT rendering in a supermarket
FLOBOT rendering inside an airport
First FLOBOT operational prototype
FLOBOT testing in Carrefour
FLOBOT prototype in action
FLOBOT conceptual drawing
FLOBOT rendering in a hospital