This is a single stream recycling robot, RoCycle, developed by MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL). RoCycle is able to differentiate if an object is paper, metal, or plastic. Machines that are able to identify a specific material is not new. I have personally seen one in action in an expo in Atlanta almost twenty years ago, but that optical sensing machine could only sort out PET bottles from non-PET bottles. Another age-old machine is the electrically powered magnet that is use to pick out ferrous metals (usually in the early stage of sorting).
So, as you can see, almost all machines to date are specialist machines and they are not small. Usually, these machines take up quite a significant amount of space. However, today, most part of the sorting process has to be done manually with workers picking out the recyclable materials from a fast-moving conveyor belt. Needless to say, this manual work is dirty and unsafe. I used to be in the industry, btw, and so, I know exactly what’s wrong with the industry.
Computer vision can contribute to sorting by machines, but with RoCycle, it acts like a person, complete with tactile sensors on its grippers to detect an object’s size and stiffness, and appropriating them accordingly. RoCycle could be the answer to a job that not many wants to do. But RoCycle is still at its early stage. At this point, it is 85 percent accurate in detecting materials when stationary, and just 63 percent accurate on an actual simulated conveyor belt. It is nevertheless a good start.
If you ask me, I do not believe in touch sensors alone. I’d say it should be use in conjunction with other sorting technologies like optical-based solutions, magnetic sorters and aluminum sorters, and leaving the final line to robots to pick out papers and other recyclable that may have missed the first few filtering processes – just like what the recycling industry is doing now.
But the key here isn’t about what it can do in the future; it is about the breakthrough in development of high-deformation haptic feedback soft robotic gripper and how it can enable automation in recycling separation, and thereby, potentially replace humans and even all other kinds of mechanized sorting solutions altogether. You can read in-depth on this topic on the two papers submitted by MIT CSAIL HERE and HERE. You may also want to see it in action in the video embedded below.
Images: MIT CSAIL.