Robotic assembly is challenging, but it's hard to innovate when we're always looking for reliability. The solution to this challenge? More challenges!
Assembly is a classic application, both for collaborative robots and for traditional industrial robots. But, although robotic assembly is popular, it's not easy. It involves more precision movements than simper tasks like pick and place or machine tending. Also, it often relies on difficult part presentation, fixturing, and detailed positioning.
These challenges can lead to a problem when it comes to innovation. Most of us want to set up our robotic cells as quickly as possible with the maximum reliability, which is understandable and certainly achievable. However, sometimes it can mean that we don't try new things in the ways we implement robot assembly — we just go for the tried and tested methods.
How can we balance the productivity needs of production environments with the need to advance and innovate?
To solve these challenges we need… another type of challenge!
Of course, lab-based robotics research does explore innovative techniques for robotic assembly, but it often lacks practical application. Once researchers have validated their latest development, they move on to the next thing having never reliably demonstrated how their application can be used by industrial users.
As a result, there is a disconnect between the innovative assembly techniques in robotics research and the tried-and-tested robotic assembly used in industry.
Robotic challenges have become increasingly popular to solve this disconnect. And by "challenges" I mean "competitions."
There are many robot challenges around the world, the most famous being DARPA, RoboCup and the Amazon Picking Challenge.
Each challenge has an element of competition to it and focuses on a particular application area. For example, the Amazon Picking Challenge in 2015-2017 challenged participating teams to recognize, grasp, and pack a variety of different objects. Teams worked towards reliable but innovative solutions to this tough bin-picking application. Amazon actually dropped their robotics challenge in 2018, deciding instead to fund research projects through their AMA program, but it contributed a lot to the robotics community in the three years it ran.
Robot challenges are not just a fun way for research groups to compete with each other. Researchers from the University of Alabama argue that challenges are basically special experiments which allow that "research and integrated development can be pursued in tandem as a catalyst for innovation." Other researchers have argued that challenges are ideal benchmarks for robotics research, as they provide consistent tasks with which to compare the performance of different robotic systems.
2018 was a pretty good year for assembly-based robotics challenges.
Here are 3 of the best assembly challenges from the year:
In May, one particularly difficult assembly challenge was set at ICRA 2018 (the International Conference on Robotics and Automation).
There were four challenge areas at the ICRA conference: DJI RoboMaster AI Challenge; Soft Material Robot Challenge, Tidy Up My Room Challenge; and Mobile Microrobotics Challenge.
Part of the microrobotics area was a Microassembly Challenge. In it, microrobots had to assemble components inside a narrow channel that was only the thickness of a human blood vessel.
Thankfully, most of us don't have to use robots at such a small scale!
A more familiar assembly application was seen at our own Robotiq User Conference (RUC 2018) back in September.
As part of the Tech Challenge, our participants had to assemble promotional kits for a fictional company. This involved a tricky electronic assembly task of assembling a USB drive.
Unlike many robotics challenges, where the teams have months to develop their solutions, our Tech Challenge participants had just 24 hours to design, integrate, and get their robot cells in operation.
In October, the World Robot Summit was held in Tokyo, Japan. The event is a combined "challenge and expo" where robot manufacturers come from all around the world to show off their products and compete in a series of challenges.
The Industrial Assembly Challenge required participants to achieve quick and accurate assembly of products. It involved three distinct phases:
The challenge had some similarities with our RUC challenge — it was a complex assembly task with various interdependent stages. The biggest differences were that the robots used were different for each team and that they were not (necessarily) collaborative robots.
Challenges and competitions certainly seem to be a good way of moving forward robotic developments while focusing on robust, practical implementation. They include defined tasks, with strict success criteria, which makes teams develop real, working solutions.
Challenges achieve this improvement in two ways:
When an assembly task is too complex for current robotic systems, the only way to achieve it is to make the technology better. By gradually increasing the difficultly of robot challenges year-on-year, the technology has to keep developing. This was demonstrated at this year's World Robot Summit, where the gear assembly from 2017 was replaced by a more challenging flexible belt.
The purpose of our RUC Tech Challenge was less about improving the technology and more about improving the skills of our participants. The teams developed invaluable skills which they can now take back to their own assembly applications to improve their production.
And, after all, isn't improved production the point of assembly robots in the first place?
Which assembly task would you like to give to a robot? Tell us in the comments below or join the discussion on LinkedIn, Twitter, Facebook or the DoF professional robotics community.