Country: United Kingdom
Year Submitted: 2017
University: University of Leeds
List of Team Members (with year of graduation): James Dickinson (2017), Joey Suburt (2017), James Ash (2017), Chris Hayley (2017), Amirah Nabilah (2017)
Faculty Advisers: Dr Andrew Jackson, Professor Robert Richardson
Main Contact Email Address: A.E.Jackson@leeds.ac.uk
The Perfect Pint project has recently featured in:
The robot in action
Competing with Humans
Title: The Perfect Pint: A Beer Pouring Robot
Description: Who can pull the perfect pint of ale? A robot, or a human? We want to find out.
LabVIEW 2015, LabVIEW Real-Time Module, LabVIEW FPGA Module, NI LabVIEW Control and Simulation Module, LabVIEW SoftMotion Module, NI Vision 2015, LabVIEW compactRIO 16
NI cRIO-9024, NI cRIO-9114, 2x NI 9505, NI 9215, NI 9403
2x Industrial Devices IDM3 300mm 45mm/s Linear Actuators, Bourns 100mm Linear Slide Potentiometer, TE Connectivity 1.5Kg Compression Load Cell, Logitech C920 HD Camera
Millions of pints of beer are poured across the globe every single day. Bar staff are rushed off their feet satisfying the needs and increasingly pretentious requests of the beer drinking general public who accept nothing less than perfection in modern day society.
In 2016, however, a bearing distributor from the U.K, known as Quality Bearings Online, realised that an increasing number of robots are replicating and improving upon such tedious human tasks every day. They sought assistance from experts at The University of Leeds and Saltaire Brewery and made a connection which would alter the future of ale pouring forever...
A robot, capable of consistently pouring a perfect pint of ale, could reduce the effort of bar staff worldwide whilst satisfying the needs of every ale connoisseur and beer drinker. However, can a machine really beat the humans when it comes to their beloved beer? The only way to find out is to introduce some competition to the scenario and battle head to head. Therefore, in the middle of June, man and machine came together to see who could pour the greatest pint of ale!
The Task: Consistently pour the best pint of Saltaire Blonde ale.
The Robot: Built by a group of 5 students at the University of Leeds and powered by National Instruments technology.
The Human: One of Saltaire Brewery's finest staff members.
From this, The Perfect Pint project was born, set out with the task of designing and building a robot capable of pouring the perfect pint of ale. Whilst this appears to be a very specific task or problem, the project itself has been incredibly varied involving aspects of: mechanical design, dynamics and kinematics calculations and modelling, fluid dynamics (including computational fluid dynamics simulations), motion capture, and system control.
Final Robot Design
The robot was built using two linear actuators; one pulling an ale tap handle and the other raising and lowering a pint glass. In addition to this, it uses a linear potentiometer for closed-loop feedback control of the handle actuator as well as a load cell for sensing the presence of pint glasses and a high definition camera for measuring how much beer has been poured and the thickness of the foam head. It uses a touch screen user interface built into the back of the robot and LED indicators offering an attractive status update on the pint pouring process.
The robot was initially designed by imitating a human bartender using modern motion capture equipment to gather kinematic data. A computational fluid dynamics (CFD) package was then used to assess how varying parameters within the pint pouring process affected the formation of foam head. These results helped to optimise the robot, ensuring that every pint was poured perfectly with no wastage.
The camera used for motion capture analysis
A computational fluid dynamics (CFD) simulation of the pint pouring process
At the heart of the project was an array of National Instruments technology; particularly LabVIEW. The simple graphical-based development environment allowed the robot’s control structure and user interface to be developed quickly, allowing more time to be spent developing and optimising the rest of the robot. It interfaced easily with the National Instruments hardware using pre-built functions for many tasks such as Input/Output (I/O) and proportional, integral, derivative (PID) control. The control structure itself was split up into 3 sections, each performing a different task and utilising a different processing platform:
Field programmable gate array (FPGA)
The NI compactRIO’s FPGA chip was used for I/O such as receiving sensor data and controlling the actuators, performing the robot’s kinematic calculations, and carrying out closed-loop PID control of the pump handle; all with incredible speed.
Real-Time Host (RT)
The compactRIO’s real-time processor was used to house the robot’s main control loop (a state-machine) which performed the handle extension and retraction procedures and raised/lowered the pint glass when necessary. The RT host was also used as a bridge between the Graphical user interface (GUI) and the FPGA program
Finally, a Windows laptop hosted the GUI as well as machine vision used to measure the amount of beer poured. The machine vision used LabVIEW’s vision toolkit to take a snapshot near the end of the beer pouring process and adjust the pour accordingly.
System connection diagram
Laptop Graphical User Interface
Touch Screen Graphical User Interface
On competition day the robot performed excellently; consistently pouring perfect pint after perfect pint in front of the cameras and audience. The barman put up a good fight, beating the robot in speed, but was no match for the zero wastage and pristine head that machine could deliver.