Banner 5 steps for getting started with collaborative robots

Step 2: Identify Potential Processes for Automation

At this point, we need to discuss the strengths and limitations of process automation with collaborative robots.

In fact, you probably already have a bunch of ideas as to what processes you would like to have automated I am afraid to say that most of the tasks you thought were good candidates for automation are probably not the best choices to begin with. We need to be realistic about what the technology can do today. Some applications are easier to automate than others and as a first integration you should start small and simple.

Ideal tasks for a first collaborative robot are tasks that are highly predictable and repeatable. These are easiest to automate with a collaborative robot.

Easy to automate	Harder to automate
Repetitive tasks in which the robot does not apply force Pick and place Dispensing	Complex logic or decision making tasks Projects that require sensors Vision Force-torque sensor
Parts Little part differentiation or changeovers between similar parts Part presentation: Ordered part, structured on a table, matrix or tray	Parts High changeover rates for parts with divergent properties (size, shape) Deformable parts Part presentation: Moving conveyors Unstructured presentation (bins)
Interacting with fixed objects, such as boxes.	Integrating communications with other machines
Location controlled processes: Always pick in the same place Always place in the same place	Force controlled processes: Grinding, polishing, drilling, Precision assembly Processes requiring specialized or human know-how Welding Painting

Tip: To create a quick simulation, perform the task and think about whether you could do the same task without seeing or feeling the part, its weight or the force applied by the part. In other words, could you do it blind? This will give you an indication of what kind of sensors you will need.

What’s easy to automate ?

For a first automation project, the options below are recommended to enure an early success.

Repetitive tasks: Robots are good at repetitive tasks and endurance applications. If you have an application that repeats the same motion all day, it is a good candidate for automation by robots. One place to look for these types of applications are places where you have repetitive stress injuries in your workers, or places that have been identified as high risk for these kinds of injuries.

Part presentation in pick and place:

Ordered parts: A part or a series of parts that are always in the very same position during the whole process are easy to automate. The robot will basically repeat again and again the same motion without asking any questions - or getting injured.

Similar parts: Parts with similar properties: dimension, weight range or physical appearance allow for easier transitions between production changeovers and will ensure that the robot can handle these parts without tool changes.

Dispensing: The part is going to be placed somewhere that requires precision, so that the part will drop in the same spot each and every time. This is an easy dispensing process. Requiring more precision is tougher, but not impossible.

Matrices and trays: If you have a matrix of parts (i.e. 10 x 10 parts), or if they come in trays (for example, if you’re handling small parts), you can teach the robot how to pick each part in the matrix or tray by simply programming the initial position, ending position and the number of parts in each row and column. Fixed matrices or trays are the easiest processes to automate, but in the case of a tray, you can also program the robot to remove and bring a new tray to the spot where it will pick the parts from the tray.

Keep in mind that Universal Robots offers wizards (an on-screen dialog) to automate these types of picking applications. Their wizard is embedded in the robot controller logic and can be engaged very quickly.

What’s complex to automate ?

For a first automation project, we don’t recommend choosing the applications below. While they can be automated, the more complex your robotic cell, the more time, energy and money you will spend on it. It’s better to start small and increase the level of automation difficulty at the same time as you increase your automation experience.

Part presentation:

Unstructured part presentation: If it’s hard to find a part with your hands, it will also surely be difficult for the robot. In fact, if the parts are presented in different positions and orientations each and every time, you will need a little bit more intelligence than just a robot arm. This means more programming, and possibly sensors.

Widely divergent parts: Parts that are unalike in dimension, weight range, physical geometry or that are deformable often require tool changes as well between production changeovers. While this certainly can be done, it will add a greater level of complexity to your automation project and is probably not a good place to start for a first robot integration project.

Conveyors: With a little more complexity, it is possible to automate conveyors too. However, you will need to figure out details such as: speed, relative position and other specifications related to time / movement / displacement. This is another case where you will need programming to make it work

Sensors: If you need to integrate any type of sensor, such as vision or force torque, the level of complexity rises. You will need to plan for more integration time and a higher cost to automate the cell overall.

Integration with machines: You might need the robot to interact with a machine so they can communicate with each other. This is a relatively simple process, but can be difficult if you have never worked with automation. If you can skip this step, you are one step closer to simplicity. One trick is to have the robot use the machine interface that was built for humans, such as a manual door and a start button for example. Alternatively, a simple digital I/O wired between the machine and the robot can be used.
Logic: If the process is complicated in terms of logic or if it involves decision making, this will add complexity to the robot program. If you are new to robotics, try building your programming skills with simpler tasks before attacking these types of problems.
Force control related tasks: If you want the robot to grind, polish, drill, weld or paint; these processes can certainly be done by robots. But processes that require a specialized or human touch are much more complex. For the robot to go from A to B is easy. Having a robot follow a trajectory while applying a specific amount of force in a specific process is more difficult, both in terms of sensor integration and programming.

In conclusion, collaborative robots are really good at pick-and-place, material handling, and material dispensing; but applications that require applying force and momentum are going to be more complex. Again starting simple and building up expertise is the approach we find has worked best. If you’d like to explore more on applications for collaborative robots, here are a series of videos that might prove useful: Collaborative Robots Case Studies

Get the PDF version for this step here.