Skip to main content

Faster and cheaper robots for object recognition in Waste sorting through AI and external robot control

Periodic Reporting for period 1 - WEXOBOT (Faster and cheaper robots for object recognition in Waste sorting through AI and external robot control)

Berichtszeitraum: 2019-12-01 bis 2020-03-31

The European problem we are addressing with WEXOBOT is the low ratio of recycling, in the broad sense. In the narrow sense, our mission is to improve the sorting process within downstream waste management activities. Although waste management in the EU has grown considerably in recent decades, over a quarter of municipal waste is still landfilled, and less than half is recycled or composted, with wide variations between the Member States. According to the Eurostat, in terms of household waste, only 40% of it is reused or recycled, and in some countries, more than 80% still goes to landfill. In Waste Directive, the EU defines 50% recycling target for 2020 to promote the circular economy (recently revised to include new and more ambitious targets: 70% to be achieved by 2030). In this respect, to make the goals mentioned earlier, it is necessary to improve the waste management process. Waste management companies need more efficient and affordable technologies that maximise the recovery of pure material to increase the recycling of secondary raw, decrease the burden on landfills, decrease the carbon footprint and improve safety and health in the work environment in waste management. The potential of automating and especially using robots to perform waste sorting is untapped in Europe. The two main drawbacks of the existing robots for waste sorting are related to their high price, and their lack of customisation and reprogramming to increase object recognition. The most significant challenge within waste management is placed in the downstream activities, precisely within the sorting process. Waste sorting robots have emerged as disruptive intelligent automated machines. These machines are proficient at multitasking, autonomous, scalable and have integrated learning systems that can function tirelessly 24/7. The challenge is that the sorting robots are expensive because of programming activities, and they are inflexible to implement in the existing plant, required additional investment costs. Also, most of the waste management companies are using NIR technology. According to TOMRA's, the sensor-based sorting technology is the modern method for recycling plant in the waste industry. Still, there are some weaknesses of this technology regarding recognition of items such as black plastic, glass and rubber. As a guarantee of a good quality of sorting materials, manual labour is still needed. When it comes to the batteries, most of them during the recycling process goes with other metal components because of a magnetic way of sorting garbage and stays there unused or even dangerous. Robot programming should be simple to be effective and invite as many people as possible into the robots' ecosystems. Byte Motion has realised that robots controlled by humans do a much better job than robots that controls themselves. Byte Motion aims to support as many robots and brands as possible, making them faster and easier to reprogram. Byte Motion's technical solution is a fusion of the latest VR/AR solutions for human-like robotic movements with existing robotic technologies. Byte Motion is raising the bar and setting new standards for telerobotic and telepresence. As of January 2019, Byte Motion's beta version of AI for object recognition is capable of recognising glass bottles, PET and Tetra Pak. We have expanded this solution to batteries and other materials and shapes during this project. Our overall objective is to improve the waste sorting process by making waste sorting robots more flexible, available and precise. And on the broader sense, we aim to contribute to the circular economy sustention, as well as making working places within the waste sorting process more human.
"Within these fourth months, we have done what was determined within work packages. By conducting primary market research and interviewed several MRFs in Norway, we have discovered following barriers with robotics: progress of the robot technology is too slow; the technology today isn't good enough; a robot is not fast enough to replace NIR-machine because of the turnover capacity; robot's software capacity and sensor functions in regards to not recognising the waste through the sensor; size per hour is not good enough, and the costs of implementation are too high. According to their needs and our customers' requirements, we see the need for technology to evolve further in the direction of one complete smart solution that operates within the space as human. Further activities we plan are to improve the teleoperation, 3D model extraction, new lasers integration, ArUco Boards integration, and to merge it in one flexible ""all in one"" platform. The best model for local marketing penetration is through a strategic alliance with robot integrators. For now, we have a strong and fruit-bearing strategic partnership with Robot Norge, and along with them, we are approaching the final customers on the Norway market. International marketing strategy is marketing development – using the same solution to geographically different markets (Denmark, Sweden and further in Europe). In the long-term meaning, there is an open space to collaborate with robot producer directly (e.g. ABB). We have developed a technical road map to be clear with parts of technology that should be improved, as well as the time needed and the budget. Within this project, we already have developed a better version of AI while conveyor moving. Regarding IPR strategy, the procedure is complicated, long-lasting (minimum one month) and expensive. We believe it is better to focus our strength and time on product development. Software solutions are not easy to patent and more importantly, harder to protect from copying. The business plan showed that WEXOBOT is a profitable idea. We aim to have a positive state in 2020, 7 people employed in five years. At the end of 2024, we expect around 4 million EUR profit and 70 clients."
"Progress beyond state of the art is the fact we already have three customers. However, one of them is coming from the waste management industry - IVAR (waste management industry). The other two are coming from the construction industry and the seafood industry. This leads to the fact that we see the application of our solution outside the sector of waste management which is pretty encouraging and promising. Also, during the marketing research, we are kind of amazed at the fact that some segments of robotics are indeed far behind. At the same time, robot producers and robot integrators keep doing pretty much traditionally things. Everything relies on robot software, classical robot programming (for simple repetitive actions) which is slow and expensive. Some producers create complete new smart ""boxes"" with sensors and AI, but they are not flexible for different purposes and environments. We have found our solution even more needed and useful generally for robotic industry, and as well for improving waste sorting process."
Byte Motion AS Owners
Wexobot Solution Picture