Recycling has its merits — it helps save resources, energy, and it helps to protect the environment. It’s also labor-intensive, monotonous, and can be hazardous for the workers who sort consumables by hand. Researchers from MIT’s Computer Science & Artificial Intelligence Lab (CSAIL) have developed a robot that can perform the same task and uses a ‘sense of touch’ to accomplish the job.
The RoCycle robot uses sensors in its gripper to identify an objects makeup and then sort it accordingly. An elastic strain sensor provides proprioceptive feedback to gauge the object’s size, while a pair of pressure sensors grants the robot that sense of touch based on how hard or soft an object is — be it a paper carton or aluminum can. Since the sensors are conductive, the robot can identify metal objects rather quickly.
RoCycle’s hands were designed using custom auxetics — materials that have a negative Poisson’s ratio, getting wider when stretched, and twist when cut. Both of the robot’s right and left fingers use those auxetics, which counter each other’s rotation, providing increased dynamic movement without the need for valves or fluidic systems.
Data garnered from those sensors are analyzed using custom software, which compares it to a variety of known paper, plastic, and metal items. When a match is found, it then places the object into a designated receptacle. The platform is 85% accurate when objects are stationary, and only 63% correct when identifying objects on a moving conveyor belt, so it’s unlikely we will see RoCycle at our local recycling centers any time soon.