Smart Table Assembly
This thesis explores how to fuse the strengths of both, industrial robots and human operators, to reduce the difficult challenges for the operators and to improve the general work environment while improving productivity and flexibility of the assembly line.
To get first-hand insights of current assembly, I observed the work environments of four different manufacturers around Sweden. I analyzed their setups, watched the processes, and got to interview the production managers and operatives to get to know the various challenges of today’s assembly lines.
I clustered my impressions and generated new ideas for the assembly work environment. I tested my concepts with cardboard mockups which helped me to roleplay a user scenario. Then I created the archetype overview for my final product, which I then designed and tested with CNC milled prototypes.
The final result is called COOE. It is a smart table-based assembly system combining the strengths of the machine and the human. It takes human motions and behavior into account while boosting productivity as a collaborative team.
Atlas Copco Industrial Technique
Umeå Institute of Design, Sweden
Assembly operators need to perform repetitive movements several times throughout the day. Repetitive tasks can lead to muscle strain and pain.
The workspace, its parts, tasks, and interactions can lead to an overwhelmingly complex information structure.
The operators stay within the same loop and work routine for most of the time and keep doing the same kind of work over and over. This can fatigue the mind and lead to errors in the assembly.
Current industrial robots are working behind fences, decoupled from human operators.
Modern product assembly is a demanding challenge due to the
high number of parts, machines, tools, techniques, and people
involved in the process.
A current car, for instance, is a system of a multitude of components - roughly 30.000 depending on the model and customization options. Therefore, assembly tasks are arranged in sequence and precisely timed. For the operators, this time is known as takt time, a short interval in which all jobs have to be fulfilled.
A modern product line can produce different variants of the same product, due to customization options that include different parts of the exterior or interior. This complexity provides many possibilities for errors. The workers have been trained thoroughly for each step.
Even though many physically intense, harmful or repetitive tasks are being automated and carried out by machines and robots, human operators continue to be at the core of the assembly line for various reasons.
The worker’s flexibility, quality, speed, cost, and reliability can in some instances still outmatch industrial robots and is, therefore, a key to high productivity and performance.
COOE is an automated, collaborative table assembly system. It takes over many repetitive tasks and hence reduces the cognitive and physical workload of the operative. Its design and freedom in motion combined with an artificial intelligence allow for a safe, seamless and collaborative workflow that is build around a human-centered work environment.
The system combines a robotic arm and a robotic product carrier working in collaboration with a human coworker.
The robots can move omnidirectionally on a unique table surface which enables the system to be set up highly individual to the needs of the manufacturer.
The carrier transports the main frame of the assembled product. It can change the position and angle of the product which enables a fast workflow. Positional data sharing, depth sensing paired with an artificial intelligence core allows the robotic arm to manoeuver with precision safely.
The robotic arm carries a tool head that exchanges through a standard interface allowing many different tools and applications such as bolting, gluing and delivering parts.
When COOE is in standby or transitioning to a new workplace the arms fold in on top of the base plate. Then, its small footprint allows to maximizing the use of valuable space on the table.
To strengthen the relationship with the operator, COOE should convey characteristics that we usually do not associate with industrial robots such as friendliness, softness, and behavioral aspects related to natural intelligence.
Therefore COOE’s main forms are of a mixture of clean geometric shapes along dynamic organic shapes. Functional areas are highlighted with metallic colors. Areas with movement are indicated in a white color. The primary color is an anthracite grey with a matte, rubber-like finish. It should resemble a soft product that the operative does not fear to get in contact with.
Combined with the base, COOE can rotate eight axes. These allow for very fluid motion behaviors. Underneath COOE’s shell are precisely fitted electric motors, bearings and angular gears that enable the motion.
Like a camera gimbal, COOE can move its arm to another position while holding the bolting head still. This is an important feature when the operator needs to reach certain areas on the to be assembled product.
In the picture, the tool head is connected to a bolt supply system that automatically delivers the right bolt for the right task reducing the cognitive load for the operative and preventing errors and damages in the process.
The base can move omnidirectionally on the table due to the use of mechanum wheels. To lift heavy load during assembly, it can lock to cylindrical docking stations that are flush integrated into the table. Due to the circular dock, COOE gains stability but does not lose its ability to rotate the base.
COOE is powered by batteries that sit inside the base and can be charged through the docking interface. The arm is integrated seamlessly in the bottom to prevent injuries. The top of the base plate is 3D printed and can be customized. The outer ring of the base has integrated led lights that indicate the direction of movement and docking for the operative to anticipate COOE’s moves.
COOE continually analyses its surroundings with depth-sensing cameras and laser dot-pattern-matrix-projectors built-in on each side of the joints. The underlying computing platform analyses obstacles and motions and calculates ways to move around them, for instance finding a way around a human hand.
Despite COOE's AI learning skills, COOE still features buttons that can be used to teach tasks during the set up of the system. It can also be used if COOE tends to get into the way of the operative. Therefore each operative has a unique COOE experience.
The joints are designed to feature enough space to make sure not to damage the hand of an operative even if it is in the standby position as seen in the picture above. I called this feature the “high five” resembling both the safety aspect and the collaboration between human and robot. It symbolically refers to the act of greeting a person.
GE HC Bio-Sciences AB in Umeå focuses on the development, production, and assembly of products and services used for medical devices.
The engine assembly station for all Volvo Cars worldwide is in Skövde, Sweden.
Volvo Trucks is one of the largest truck manufacturers worldwide and part of the Volvo Group.
Komatsu Forest is one of the worlds largest producer of large forestry machines.
All assembly lines and their processes were unique for each company. The setup depended on a multitude of factors. This is why I broke down the complexity of the workstations and assembly loop layout in easy to grasp graphics. This way I was able to compare them and find overlapping issues and opportunities much quicker.
During my research, I got the chance to talk directly with the operators in their work environment. Looking over their shoulder during their shift, I was able to get a glimpse of their daily struggles and sometimes I even tried some work tasks myself.
based on observed problems
based on resulting opportunity areas
creating visualizations to communicate
To get an idea of the dimensions and scale of a table operated collaborative robot system, I created cardboard mock-ups. I wanted to be able to act out a scenario on an assembly line using the prototypes.
I used the mock-ups to test and develop my concept idea. In the project studio, I created an assembly workflow on a table. Acting out the scenario made me realize where to implement changes to strengthen the concept.
To communicate the concept idea later I used stop-motion techniques to create a short sketch video showing the scenario tested before.
Ideation sketches helped me to develop the AGV and arm parts further. I wanted to create a design for the robotic arm that would not create a harmful “scissor effect” at its joints by design. Instead, I focused on a shape that would not intimidate the operative to get in touch.
The design language should communicate a powerful, professional, and capable machine while also behaving in a natural, human way. The operator should not be scared to touch the cobot since it would stop its movements immediately if it encounters an obstacle.
After the general package of functions and size was set, CAD models were used to define the form. During the sketch phases those varied from basic to more advanced shapes and allowed to check the appearance of different colours and finishes quickly.
Several CNC milled, and 3D printed 1:1 models were created to check the function.
The process of creating all models can be seen in the image on the left. The model was built in many iterations. It was therefore easy to compare older models with younger ideas as well as to show the development process.
The name COOE (phonetic spelling: [ˈkɔiː]) contains the word “co” for collaboration.
The logo symbolically sums up the workflow of the operator together with the robotic system and the part supply. Looking closely one can spot the operator on the left - in the letter C - facing the two cobots in front - the two letters O. The part supply is hidden in the letter E which comes in from the back.
The logo’s geometric construction lets it appear balanced and clean. It is also easy to scale and reproduce in vector graphics.
The logo was created to seamlessly blend in different backgrounds on paper or in digital media. Therefore the logo needed to be flexible and able to handle different shades of grey and color.
While developing the shape of the model, I also made sure that the virtual model translated well into the physical reality. This is why I created rapid prototypes in 1:1 scale to balance function, size, and joint arrangement. The physical model allowed me to test the movements to get a feel for COOE's behaviour.
The final model consisted of 55 individual pieces. The majority was created with the two Roland CNC mills because of the great finish they offer with little post-processing work, their reliability, the 24h access in University, and because I enjoy working with them.
The other parts were 3D printed with an SLS machine, laser cut, or produced with a lathe. I glued, sanded, painted, assembled myself, which was exhausting but exiting the more pieces came together. Finally, the result was exhibited at UID ‘18.
Let’s start a conversation!
Industrial Designer, MFA