The Robotic revolution

By 2025, machines will perform more current work tasks than humans according to reports by the World Economic Forum [link 1]. Currently machines perform 71% of tasks that humans can perform. This is the robotic revolution and it is taking place now.

Manufacturing as we know it now has its roots in the early Twentieth Century. It has evolved from small handcraft shops, to factories, to large assembly lines. In fact, companies who did not adopt the concept of large assembly lines soon became extinct. The landscape of manufacturing today is changing once again with Industry 4.0 automation becoming the new norm. Companies that do not jump on this phenomenon may soon find themselves being left behind. This automation is fueled by a variety of new technologies: cheaper and more capable robots; more advanced sensors; and increased access to Machine Learning algorithms. One thing is common to every industrial factory - an assembly line. Machines are able to perform assembly line tasks faster than ever before. For example, Quality control processes that required workers to manually check items for defects can now be completed at a fraction of the time with machines. These Quality Assurance machines are able to continuously receive parts along a conveyor belt, use cameras to take pictures, and identify defected parts through Computer Vision at half, or less, of the cost for a similar project just 5-7 years back. 

In addition, collaborative robots (cobots) have also changed the way we think of automation. These cobots are able to perform tasks which have been performed by human operators for years. Simple tasks such as pick and place can be automated. The cobots can be configured without having to write a single line of code. The great news is that prices of robots are dropping significantly. In the past a robot have needed to be caged for safety reasons, which took up more space and they would have started at US$100000. Cobots today cost up to US$42000 (about half the price). The price of cobots has been dropping drastically from month to month and can even be found for below US$20,000 with some as low as US$2000. Cobots these days come with an array of features including more degrees of freedom, dual arms, and embedded 3D cameras. Some of the largest industrial robot makers have released cobots in recent years. ABB, Kuka, and Kawasaki Industries have all released their own versions for cobots for industrial use canablizing thier more expensive conventional robots in cages. acoording to CB insight "" 2019 will see robots moving deeper into the production process, a continuation of the trend of robots leaving more confined, structured environments."".  

The TR1 for example is a simple robot but is commercial and can be bought off the shelf to start work immediately. It showcases how technology is maturing to become cheaper and easier which brings entrepreneurs to develop new applications and robotic solutions. Showing how the hardware cost should not be so high: imagine that there are some cars with thousands of parts from different providers that are much more complex than a cobot, yet these cars cost less than a cobot...

The ease of automation is supported by improving 3D printing technology, sophisticated sensors and an abundance of algorithmic solutions, all of which brings a fast-easy prototype. Rapid prototyping means a design can be iteratively tested and improved on in a much smaller time frame. I tested this out recently by allowing an intern to design and automate a pick and place task. The intern was able to design a gripper for the Sawyer and configure the Sawyer to perform the pick and place task without writing a single line of code.

Sensor technology is also becoming more advanced and the costs are decreasing drastically. Sensors have slowly developed from 2D images to 3D point clouds, and the applications of this are plenty. The Sawyer, released in 2015, included an embedded 2D camera while Epson’s more recent Worksense boasts 3D Cameras which are able to identify an object based on CAD models. Moreover, external 3D cameras such as Pick-it have implemented their camera onto a Swiss maker and are able to add their Pick-it software onto new hardware.

[The above is a good example of cooperation between a Start-Up in Swizerland that developed the 3D point cloud camera + Pick-It that developed the software that uses the point cloud to identify the objects and give the ABB robot the XYZ coordinates of the part]. Just like that a normal robotic arm can be transformed into an arm that can make sense of a point cloud and locate the XYZ coordinates of parts.

CPU power & sensors’ capabilities are also improving while their price is going down, allowing autonomous vehicles to navigate their surroundings in a manufacturing environment. This will lead to autonomous robots moving in the factory to perform tasks from cleaning the factory to performing assembly tasks to moving around to multiple machines located at different points of the factory. Showing rise of companies like polygon, MIR and more.

Machine Learning is becoming the new 21st century commodity. Companies can add predictive maintenance using AI and Machine Learning algorithms to improve their product and move their product to a service. AI as a service brings new capabilities to robotics. Google and others are working on the universal pick challenge, bin-picking and interacting with different ‘tasks’ and unexpected scenarios. There are many examples of the partnership between AI and robotics – Google Grasping training and the Amazon pick challenge. Machine Learning exists everywhere and can be used to make automation faster and smarter. The video below shows how Machine Learning helps robots grasp better and continuous grasping helps train the Machine Learning algorithms. The two go hand in hand.

Human roles will change accordingly. Humans will start to manage, maintain and service the robots. Human workers will not be replaced, instead their job will change from manual intensive jobs to jobs which require more creativity, critical thinking skills, and complex problem solving. In any automation project that I was involved in, humans were not let go, instead they were able to be more productive and get better meaning from their work.

All these advancements in technology is going to ease our transition into automation. Automation is becoming cheaper and easier to implement. Industry 4.0 is going to see a smarter and more efficient factory floor. So if you have not taken the first step towards automation, start now! Or you will be left behind.

Written in collaboration with Sivalingam Denesh If you would like to learn more about automation visit www.let-lab.com and contact us!