#	Pagina
attuale pagina	/open-h2020/projects/197960/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 2 - Robolution (Robotic Recycling Revolution)

Teaser

Summary

The increasing scarcity for raw materials, growingly strict regulations and social pressure have turned waste into a resource, making recycling highly attractive. This has created a major market opportunity for new technologies that achieve high purity of sorted materials at a low cost. ZenRobotics Oy has developed a robotic waste sorting system ZenRobotics Recycler (ZRR) that has the potential to revolutionise waste sorting, replacing low-performing hazardous manual jobs with highly efficient and fast autonomous robotic pickers. Unlike traditional recycling machinery that is based on mechanic and electric components, ZRR is powered by artificial intelligence. It is also the first commercially available robotic waste sorting system which offers one system for multiple tasks.
The key innovation of ZRR is a unique machine-learning based system, which gathers gigabytes of data of its environment, makes smart decisions and moves a robot arm in an unpredictable environment. Given the novelty of the technology, a paradigm shift is necessary in waste management for wide uptake of ZRR.
The general objective of the Robolution project was to revolutionise waste sorting by developing the ZRR prototype into a reliable and commercially attractive robotic sorting system for Commercial and Industrial waste that outperforms all existing sorting technologies in terms of picking speed, purity of sorted materials and investment requirements.
The project focused on:
Robot motion control to be able sort faster smaller objects and pick larger and heavier objects.
Recognition to detect new fractions (plastics, ferrous/non-ferrous metals) with 95% purity and recovery.
Development of the reporting tool enabling optimization within a waste sorting plant.
Optimization of the ZRR commissioning and service to shorten time-to-market time.
Testing and demonstration of the new functionality in real life conditions.
ZenRobotics targets the global waste sorting equipment market currently worth about â‚¬1.5-3 billion annually, but thanks to technological advances, the market is likely to explode. The expected annual turnover of the Commercial and Industrial waste ZRR reaches â‚¬150 million by 2021.

Work performed

WP1 Robot motion control
The overall goal was to improve the picking performance, and to better understand the causes of pick failures. The goals were reached and the key results are:
1. Significant performance improvements to the current robot;
2. Technology demonstration of a self-learning next generation robot.

WP2 New Fractions and Recognition Performance
The overall goal was to extend the applicability of the sorting robot to new use cases and new waste fractions. The most significant result is the trainable fractions -feature, which enabled the operator to teach the robot to tell apart fractions. This has been the single most valuable new product feature developed in this project.

WP3 Reporting Tool and Object Real-time Weighing
The goal was to measure the performance of the robot better, and provide customers access to the most relevant information gathered by the robot. We have developed several new tools including online reporting tool and prototype online waste stream characterization.

WP4 ZRR System Commissioning and Service and pre-processing of the ZRR Waste Stream
The goal was to make sure that the time and effort of commissioning a robot is not a bottleneck in sales, and that the waste stream pre-processing is adequate for the robots. The most notable result is that the effort for commissioning a robot has been dramatically lowered - from approx. 2 weeks to 2-3 days.

WP5 Testing
The goal was to continuously gather experience of the robot behaviour in production environment, and find out the limits of its capabilities and performance. The work was closely tied to new feature development, since the production environment cannot be replicated in a simulation or laboratory. This means that all development of new features, additions and bug fixes needed to be tested in a real production environment. Testing is an ongoing effort. The most notable result of the testing work is the reliability of the robots.

WP6 Communication, exploitation and dissemination
WP6 has created a substantial awareness of the work performed. The main dissemination was performed through the following core activities:
Visiting and demonstrating ZRR at related waste management and recycling exhibitions, fairs and conferences;
Updating the ZenRobotics commercialisation plan;
Creating promotional tools and channels.
As a result, the new, trainable version of the ZenRobotics Recycler was launched at the IFAT exhibition in Munich, May 2016. The look of the product had also been improved through industrial design. The launch was a great success, and created a lot of interest in the industry. Together with the product launches in Japan and the US in May and June, the company managed to demonstrate on a global level that the ZRR system is a real, viable solution for sorting waste as of today. The three events together had around 320.000 visitors. Four industry trade magazines featured robotic waste sorting as their main theme during the period.

Final results

Robotic recycling offers vast potential in enabling true circular economy. The fundamental problem in applying robotics to waste sorting has been that industrial automation is unable to function in an unstructured environment. In other words, industrial automation requires that the nature, position and behavior of all objects are defined in advance. There are no surprises.
Sorting waste required technology beyond industrial automation. An incoming waste stream cannot be modeled in advance. A sorting robot must deal with inaccurate sensor data, collisions and unexpected objects arriving on the conveyor belt. Robot must be able to deal with the unexpected. In other words, industrial automation is not enough. A sorting robot needs to be smart.
The first step in applying smart robotics is to develop a robot which:
Can operate in an unstructured environment;
Is robust enough to survive in an actual production environment;
Is a financially viable waste sorting method for the customer.
We have now achieved the first step with the ZRR Recycler. Smart robotics has been a promise for decades. Very few such products have generated any actual revenue.
The impact of current-level robotic sorting is:
Higher versatility; Unlile traditional recycling processes, robotic sorting process can be easily adjusted to the waste on the fly;
Smaller plants; Robots enable smaller and cheaper recycling plants;
Less need for waste incineration.
The next level in smart recycling robotics is to develop a robot which is independent of external feeding and pre-processing hardware. The primary purpose of a next-level robot is to serve as a pre-sorting and feeding equipment for existing material recycling processes.
The ultimate goal for robotic recycling is serve as the mirror image of the manufacturing industry. Every manufactured product will eventually be dismantled and recycled.