Automated Robotic Food Manufacturing System.
Lead Research Organisation:
University of Lincoln
Department Name: National Centre for Food Manufacturing
Abstract
The technology being developed in this project will revolutionise the industry as it will provide a truly robotic approach to batch mixing and cooking operations.
Three key developments include: 1) "Trajectory shaping" for ARFMS vessel movement to avoid instability and spillage. This work will have two aspects i) Software filter (work package 1) & ii) Vessel design (work package 2). 2) "Robot supervision" which is closely linked with the above, (work packages 1, 2, 3 and 4). This will involve monitoring with sensors what is going on inside the vessel movement wise with a view to developing a feedback loop to the robot to adjust trajectory and speed. 3) Hygiene compliance and BRC accreditation, (work package 5).
Food industry technical compliance expertise at the NCFM will be utilised together with an independent food hygiene consultancy to define the cleaning and technical control standards demanded by the food processing sector. This expertise will develop solutions for these challenges and will validate these solutions via a series of industrial scale trials in the NCFM manufacturing facility with an objective of full compliance to BRC Global Standard for Food, Issue 7.
ARFMS will utilise small processing vessels (weighing up to 1000Kgs when full) seamlessly moved by robot from process docking station to process docking station with no pumps or pipework and all instrumentation and control systems contained within the process workhead (lid), robot and installation. The robot will select and connect with the appropriate processing workhead, e.g. ingredient dispensing, mixing, homogenising, cooking, cooling and packing, depending on product being produced, and accurately locate onto the open vessel for activation of the processing cycle.
The project plan has been realistically constructed by the project team. It is proposed to comprise of six separate Work Packages and seven Project Milestones covering one or more work packages: WP1 (Milestones 1&2) - Modelling System Dynamics (5 months). Mathematical modelling of robot tasks, installation and use of small scale robot to validate algorithms generated. WP2 (MS3) - Optimal Vessel Design & Model Variances (4 months, 3 months running concurrently with WP1 ). This will ensure that movement speed, accuracy and efficiency are maximised without spillage of contents or system damage. WP3 (MS4) - Integration of System Model into Commercial Controls (3 months). Algorithms derived from WP1 and 2 will be integrated into the large size commercial Kuka Titan KR1000 robot. WP4 (MS5) - Full Scale Evaluation (4 months). This will evaluate the modelling and vessel developments on the large scale test system at NCFM's Holbeach facility producing real food products. WP5 (MS6) - Hygienic Design of Automated Robotic Food Manufacturing System (12 months running concurrently with other work packages). To investigate and implement suitable hygienic standard compliance measures into the system. WP6 (MS7) - Project conclusions, future development, dissemination events and commercial exploitation.
Key members of the project team are: OAL's Senior Management team giving support and guidance; Project Manager; Design, Engineers (Mechanical and Electrical); Software Engineers; ARFMS technician and operative. UoL's team consists of robotics, fluid dynamics, food manufacturing and science specialists and comprehensvie research facilities. Specialist sub-contractor, Food Fix Ltd, will provide independent hygiene expertise based on the British Retail Consortium's Global Food Safety Standard which is the internationally recognised food industry Technical Standard. The overall project, its partners and individual work packages will be closely managed by OAL's experienced Project Manager who has successfully managed other collaborative R&D projects.
Three key developments include: 1) "Trajectory shaping" for ARFMS vessel movement to avoid instability and spillage. This work will have two aspects i) Software filter (work package 1) & ii) Vessel design (work package 2). 2) "Robot supervision" which is closely linked with the above, (work packages 1, 2, 3 and 4). This will involve monitoring with sensors what is going on inside the vessel movement wise with a view to developing a feedback loop to the robot to adjust trajectory and speed. 3) Hygiene compliance and BRC accreditation, (work package 5).
Food industry technical compliance expertise at the NCFM will be utilised together with an independent food hygiene consultancy to define the cleaning and technical control standards demanded by the food processing sector. This expertise will develop solutions for these challenges and will validate these solutions via a series of industrial scale trials in the NCFM manufacturing facility with an objective of full compliance to BRC Global Standard for Food, Issue 7.
ARFMS will utilise small processing vessels (weighing up to 1000Kgs when full) seamlessly moved by robot from process docking station to process docking station with no pumps or pipework and all instrumentation and control systems contained within the process workhead (lid), robot and installation. The robot will select and connect with the appropriate processing workhead, e.g. ingredient dispensing, mixing, homogenising, cooking, cooling and packing, depending on product being produced, and accurately locate onto the open vessel for activation of the processing cycle.
The project plan has been realistically constructed by the project team. It is proposed to comprise of six separate Work Packages and seven Project Milestones covering one or more work packages: WP1 (Milestones 1&2) - Modelling System Dynamics (5 months). Mathematical modelling of robot tasks, installation and use of small scale robot to validate algorithms generated. WP2 (MS3) - Optimal Vessel Design & Model Variances (4 months, 3 months running concurrently with WP1 ). This will ensure that movement speed, accuracy and efficiency are maximised without spillage of contents or system damage. WP3 (MS4) - Integration of System Model into Commercial Controls (3 months). Algorithms derived from WP1 and 2 will be integrated into the large size commercial Kuka Titan KR1000 robot. WP4 (MS5) - Full Scale Evaluation (4 months). This will evaluate the modelling and vessel developments on the large scale test system at NCFM's Holbeach facility producing real food products. WP5 (MS6) - Hygienic Design of Automated Robotic Food Manufacturing System (12 months running concurrently with other work packages). To investigate and implement suitable hygienic standard compliance measures into the system. WP6 (MS7) - Project conclusions, future development, dissemination events and commercial exploitation.
Key members of the project team are: OAL's Senior Management team giving support and guidance; Project Manager; Design, Engineers (Mechanical and Electrical); Software Engineers; ARFMS technician and operative. UoL's team consists of robotics, fluid dynamics, food manufacturing and science specialists and comprehensvie research facilities. Specialist sub-contractor, Food Fix Ltd, will provide independent hygiene expertise based on the British Retail Consortium's Global Food Safety Standard which is the internationally recognised food industry Technical Standard. The overall project, its partners and individual work packages will be closely managed by OAL's experienced Project Manager who has successfully managed other collaborative R&D projects.
Planned Impact
There has only been incremental development of food processing systems during the last 30 years with minor improvements to individual pieces of equipment. The food industry is the UK's largest manufacturing sector forecast to be £113.1Bn in 2018-19 (Sources: Food Manufacture.co.uk and BDO), but unlike other manufacturing industries, e.g. automotive, electronics and aerospace, there has been little progress in automating food manufacturing due to the lack of enabling technology and low cost labour. The industry has relied on traditional cooking and processing methodologies which are extremely labour intensive, inefficient, and produce inconsistent and relatively low product quality with high wastage and energy costs. State of the art food processing generally consists, at best, of automated recipe control, guiding operators through the production process; some automated traceability systems; and automated pumped product transfer and cleaning regimes. However, current systems still rely on product being transferred from one bulk storage / processing vessel to another in order for the next production operation to be carried out, relying on a high degree of manual handling and operator interface. Transfer systems not only damage product through use of pumps and pipe work, but generate waste left in the system.
Food manufacturers are now having to seek alternative production methods due to: the highly competitive nature of food processing; continuing downward pressure on costs from retailers and consumers; demand for increased variety, fresher, better and healthier products requiring smaller batch production runs; ever more stringent health and safety and environmental legislation; and the national minimum wage increasing to £9 per hour by 2020. This step change, initially for soup, sauce and liquid based products, will be achieved through the innovative combination of disruptive and proven enabling technologies. With the availability of such technologies, including OAL's revolutionary patented, Steam Infusion process, (number 20140064988), advanced automation capabilities and robotics, automated food production on an industrial scale can now become reality through ARFMS (Automated Robotic Food Manufacturing System). The fully integrated ARFMS system will enable and apply Industry 4.0 (Digital Manufacturing) technologies for the benefit of food manufacturers by connecting physical and digital manufacturing across various sizes of business, ultimately enabling "end to end flexible automation" of a food factory. The system will produce consistent, better quality products, faster, whilst significantly reducing wastage and energy costs, and taking up to 50% less factory space. The flexibility that ARFMS provides overcomes the barriers that automation and the use of robotics for food manufacturing has suffered to date.
Summarised, the key innovative aspects of ARFMS are: i) Automated robotic hygienic food processing system, (UK Patent application no. 1517271.1 and International PCT/GB2016/053005 submitted); ii) Use of smaller mobile (750-1000Kg) vessels, allowing small or repeatable larger scale production, quickly moved between processing stations by advanced robotics; iii) Processing workheads selected and fitted to open vessels by robot; iv) All instruments and control equipment will be located in the processing workheads, robot and installation, not in the open vessels.
Food manufacturers are now having to seek alternative production methods due to: the highly competitive nature of food processing; continuing downward pressure on costs from retailers and consumers; demand for increased variety, fresher, better and healthier products requiring smaller batch production runs; ever more stringent health and safety and environmental legislation; and the national minimum wage increasing to £9 per hour by 2020. This step change, initially for soup, sauce and liquid based products, will be achieved through the innovative combination of disruptive and proven enabling technologies. With the availability of such technologies, including OAL's revolutionary patented, Steam Infusion process, (number 20140064988), advanced automation capabilities and robotics, automated food production on an industrial scale can now become reality through ARFMS (Automated Robotic Food Manufacturing System). The fully integrated ARFMS system will enable and apply Industry 4.0 (Digital Manufacturing) technologies for the benefit of food manufacturers by connecting physical and digital manufacturing across various sizes of business, ultimately enabling "end to end flexible automation" of a food factory. The system will produce consistent, better quality products, faster, whilst significantly reducing wastage and energy costs, and taking up to 50% less factory space. The flexibility that ARFMS provides overcomes the barriers that automation and the use of robotics for food manufacturing has suffered to date.
Summarised, the key innovative aspects of ARFMS are: i) Automated robotic hygienic food processing system, (UK Patent application no. 1517271.1 and International PCT/GB2016/053005 submitted); ii) Use of smaller mobile (750-1000Kg) vessels, allowing small or repeatable larger scale production, quickly moved between processing stations by advanced robotics; iii) Processing workheads selected and fitted to open vessels by robot; iv) All instruments and control equipment will be located in the processing workheads, robot and installation, not in the open vessels.
Organisations
Publications
De Sousa Ribeiro F
(2019)
Deep Bayesian Self-Training
in Neural Computing and Applications
Moriello L
(2018)
Manipulating liquids with robots: A sloshing-free solution
in Control Engineering Practice
| Description | The key findings recorded so far are instrumental to the development of a robotic autonomous dry ingredient weighing station. Namely they go in the direction of: a) using robots to manipulate powders and liquids; b) using digital scales and/or force sensors to weigh the load handled by the robot with a great accuracy (1g); c) using vision systems to monitor the process. In particular, the following results have been achieved from the beginning of the project until 30/03/2018: - A novel technique to transport liquids and powders with robots avoiding sloshing and spilling have been developed. The method is based on modelling the load as a mechanical pendulum and shaping the accelerations of the robot in order to avoid oscillations; - Digital weighing systems have been interfaced with the robot to feedback the weight of the loaded ingredient. This information is used to close the loop during the ingredient loading/pouring operations. Preliminary results on the use of force sensors instead of a digital scale have already been produced. Our effort during the final phase of the project will be devoted to further explore this possibility; - In terms of vision system we have used a 3D camera to monitor the ingredient behaviour when the robot moves. More specifically, a method to derive a model of the waves generating within the vessel moved by the robot has been produced. In addition, machine vision techniques have been used to progess on-line and downstream label verification activities. During the final period of the project (30/03/2018 - 31/07/2018), the following technical findings have been produced: - The wrist force/torque sensor has been integrated in the system through a PC running Matlab and a client/server architecture based on UDP network communication. In the new configuration the force sensor is attached directly to the robot wrist and the ingredient loading tool is attached to the sensor. Loading (unloading) ingredients will change the sensor readings as the tool will be subject to external forces/torques (load weigh force). If we assume the load (ingredients) having a single point mass, it is possible to use a simple mathematical model to project the weigh force on the wrist and, consequently, to derive the weigh of the load from the sensor force/torque readings. This model will depend on the current pose of the robot, which can be measured from the robot control system. More in details, the robot interrogates the sensor to receive a weighing measure. Following this request, the sensor transmit the raw measurement to the Matlab PC, which elaborates it - also using some information coming from the robot (e.g., the tool pose) - and send the refined weighing measure to the robot. The tests performed to prove the effectiveness of the method have been successful both in terms of measurement precision and repeatability and in terms of communication speed. - The weighing measure has been used to close a control loop and obtain a precise ingredient dosing. More in details the current weigh is compared with the desired setpoint and the obtained error is utilized to compute the robot pouring pose by means of a digital PID (proportional-integral-derivative controller). This algorithm has allowed to move from a rough pouring method - where only three pouring angles where considered - to a more precise, modulated algorithm by means of which an error of less then 1% was obtained. |
| Exploitation Route | The presented findings can be instrumental to any robotics application that is aimed at transporting/manipulating any load that is in a liquid or powder state of matter. This can therefore be utilized in industrial applications for the food industry as this task is particularly important for food and drink manufacturing, but it can also be exploited in the metal industry - where melted metals are in a liquid state and where powder materials are used to mitigate and control the combustion - and in the pharmaceutical sector. In addition, from an academic point of view, the presented findings can be further improved by including the vision system into the control loop, thus using the visual feedback to steer the robot behaviour based on the load dynamics. In addition, the findings about the use of force sensors to weigh the material handled by the robot can be used to develop new fault detection algorithms aimed at improving the quality of the robotic task. |
| Sectors | Agriculture, Food and Drink,Manufacturing, including Industrial Biotechology |
| Description | Economic and Health & Safety Impact: Food & Drink processing businesses involved in the weighing and further processing of foodstuffs such as soups, sauces, dips, dressings, preserves and readymeals, are all currently highly reliant on manual labour to carry out these tasks. The current system is therefore inherently high cost and susceptible to human error and significant health & safety issues (manual handling and dust control are the two priority areas for the Health & Safety Executive (HSE)). The technology developed enables businesses to reduce these costs and improve sector resilience post brexit where traditional labour pools for the food & drink sector are diminishing. Societal Impact: The Food & Drink Manufacturing sector is a significant employer of 15% of the UK workforce. Increased business resilience due to the uptake of these advanced technologies ensure that these major employers will continue to operate, thereby providing jobs and GDP. |
| First Year Of Impact | 2018 |
| Sector | Agriculture, Food and Drink |
| Impact Types | Economic |
| Description | Engagement with the UK Government Industry Strategy Review |
| Geographic Reach | National |
| Policy Influence Type | Influenced training of practitioners or researchers |
| Impact | Researchers also been involved in the project have been engaged with the UK Government Industry Strategy Review, particularly the Digitalisation Review led by Juergen Maier of Siemens. Whilst this was at a very broad level, this project helped ensure that all input was well informed from both academic and industry perspectives. Influence on practice: The development of this robotics materials handling technology will enable the food processing sector to process far faster, more accurately and more efficiently than current sector standard practice. This is expected to advance productivity and also resilience of the food manufacturing sector (which is the UK's largest manufacturing sector). |
| Title | Robotics Weighing Station |
| Description | The developed robotic system is able to automatically weigh dry micro-ingredients in food factories with an accuracy of 1g, while working alongside human operators. |
| Type Of Technology | Systems, Materials & Instrumental Engineering |
| Year Produced | 2018 |
| Impact | This robotic weighing station is able to address and solve many problems concerned with handling and weighing dry ingredients in the food industry. As a matter of fact, whilst it is possible to automate high usage ingredients with big bag and silo systems, at present, an estimated 62% by labour cost are handled manually. In this framework, rising labour costs and health and safety have made the manual handling and weighing of ingredients uneconomical. In addition, manual repetitive tasks such as weighing are highly susceptible to human error: traceability is often paper based with inherent limitations on ensuring quality. The developed system fully automates the handling and weighing of micro ingredients overcoming human error and allows an accuracy of +/- 1g with full traceability. It also overcomes the high labour costs associated with the manual handling and weighing of ingredients with robotics. |
| URL | https://www.oalgroup.com/robotic-ingredient-weighing/ |