Eco-friendly Robotics: New Recycling Technologies
Scientists at the MIT have designed a new sensor-equipped robotic arm that can identify and classify waste for recycling.
When picturing industrial robotics, we tend to think about robots in car factories or carrying out construction tasks. There is a new generation of robots, however, that is poised to reduce the environmental impact of their industrial counterparts. The solution has been devised at the MIT's Computer Science and Artificial Intelligence Lab (CSAIL) and, basically, it is an innovative robotic arm able to identify, grip and move waste materials intended for recycling.
The RoCycle robotic arms will perform their work beside the conveyor belts that carry thousands of discarded materials every day. Thanks to their articulated design, their soft grippers, and their tactile sensors, they can sort out paper, metal and plastic waste. Their high deformation capacitive pressure and strain sensors can then allow them to grab boxes, cups, and cans. Moreover, this new technology can even distinguish waste that would go unrecognized for a human operator.
Although many of the recycling plants use magnets to sort metals and air filter to retrieve paper and plastic, most of the sorting out of waste is done manually. As the researchers of this technological project have pointed out in an article published in the MIT Technology Review, this can prove a considerable bottleneck for waste management. Even though people at their homes are aware of the colors of each recycling bin, the lack of efficient recycling technologies in the final stage of the process leaves a weak link in the chain. Mix-ups in the selection process entail a tangible cost: just in the USA, 25% of the waste is so contaminated that must be sent to landfills. Furthermore, organic waste mixed with plastics, paper, and metal poses health risks for human workers.
The challenge ahead is to improve the precision of this new technology, as the tests carried out so far have delivered an 83% success rate for static objects and a 63% rate for simulated conveyor belts. The next step will be adding sensors that provide a video feed to hone the skills of the RoCycle robot.
The robotic arm that can shoot a 3-pointer
MIT’s recycling robot has a close relative. Teams from Google, the universities of Columbia and Princeton, and the MIT itself have developed another robotic arm with an interesting feature. As anyone who has played basketball can attest, scoring a basket requires a complex calculation process, where the weight, distance, and trajectory of the ball need to be factored in. For a robot, that can be an enormous challenge. Thus, the researchers aimed to develop a robotic prototype that could scan objects and, after selecting them, toss them into a bin.
Besides the ability to recognize objects, this innovative technology makes use of a deep learning artificial intelligence system that can identify the collected object and establish the projected trajectory. The tests carried out by the researchers with this robotic arm achieved an 87% accuracy in the lifting of objects and an 83% of successful throws. Perhaps the researchers were not professional basket players, but their success rate was lower than that.