A UK recycling company has introduced a Chinese-manufactured humanoid robot in response to a workforce turnover rate of 40% per year and an 8-fold fatality rate in the waste sorting industry.

      A family-operated recycling business in East London is training a humanoid robot, manufactured in China, to sort waste on its conveyor belts, amidst a staffing crisis where employee turnover reaches 40% and the fatality rate is significantly higher than the national average. Although the robot is not yet functional, the pressing labor issues in the industry necessitate the move toward automation.

      The recycling sector faces a labor challenge that recruitment efforts alone cannot remedy. Annual employee turnover at waste sorting facilities stands at 40%, while the fatality rate is eight times that of the overall workforce average. Furthermore, work-related injuries and health issues are 45% more prevalent than in other industries. Employees must stand alongside rapidly moving conveyor belts, extracting items like shoes, concrete blocks, VHS tapes, and sometimes firearms from a mixed waste stream, all in a dusty, noisy environment that often leads to swift departures before workers can become proficient. The industry has attempted strategies like higher wages, shift rotations, and the use of temporary staff, but these measures have not altered the situation: the work is hazardous, unpleasant, and physically taxing, prompting workers to leave for better opportunities. In an East London waste yard, a family-owned business has determined that the solution lies not in improved hiring tactics but in a humanoid robot trained by the very employees it is intended to replace.

      The robot, named Alpha, is part of the Sharp Group, which processes 280,000 tonnes of mixed recycling annually at its Rainham facility, with 24 agency employees working on fast-moving conveyor belts. Founded by Tom Sharp, the company is now managed by the third generation of the family. Alpha, developed by RealMan Robotics in China and modified for recycling tasks by British startup TeknTrash Robotics, is designed to resemble a human worker. TeknTrash's founder, Al Costa, believes this humanoid design allows for seamless integration into existing plant layouts, avoiding the need for costly facility redesigns. In contrast, companies like AMP and Glacier, based in Colorado and California respectively, have created specialized sorting systems using robotic arms, air jets, and AI vision, which, while effective, require new facilities or expensive retrofitting. A humanoid robot that can occupy the same space and perform the same tasks as a human presents a more cost-effective and quicker path to automation for smaller recycling operations that cannot afford extensive upgrades.

      As of the BBC's visit, Alpha was still undergoing training, learning to execute arm movements while a human worker, equipped with a Meta Quest 3 VR headset, demonstrated sorting motions. TeknTrash's HoloLab system uses data from multiple cameras to train the robot on two main tasks: identifying items on the conveyor belt and physically lifting them. Each day, thousands of items generate millions of data points for training. Costa acknowledges the challenges, stating, "The market thinks these robots are ready to use, but they require a significant amount of data to be operationally effective." Training will span several months, and TeknTrash aims to implement the system in 1,000 European plants, all cloud-connected, contingent on Alpha mastering sorting reliably at one location initially.

      Unlike the humanoid approach, the recycling automation market has many competitors that have chosen different strategies. Sereact secured $110 million in April to enhance AI that adapts industrial robots for logistics and manufacturing, reflecting a wider trend that values software capabilities over physical form. AMP has raised $91 million in its Series D round and operates three facilities while providing AI-driven sorting technology to over 100 sites worldwide, using air jets to direct items to chutes at speeds eight to ten times faster than a human worker. CEO Tim Stuart describes this method as fundamentally different from replicating human movements, emphasizing the integration of sorting intelligence in the system's design.

      Glacier, which has the backing of Amazon and was co-founded by Rebecca Hu-Thrams, has opted for a mixed approach with robotic arms guided by AI vision systems, allowing installation in existing structures without complete overhauls. The company raised $16 million in 2025, processes recycling for nearly 10% of Americans, and earned a spot on TIME’s Best Inventions list. Hu-Thrams points out that Glacier's technology caters to semi-rural facilities on limited budgets rather than just large urban centers. The AI continuously learns from over a billion sorted items, addressing the inherent variability of waste. “Sometimes a beer can will be spraying liquid everywhere, damaging machinery,” she notes, highlighting the range of hazards her clients have encountered, including hand grenades and firearms on sorting lines.

      A recent trial by Siemens showcased a humanoid robot powered by Nvidia in a factory setting, demonstrating its ability to retrieve storage totes and place them on conveyor belts over two weeks. While it indicated that humanoids could work in actual industrial spaces, it also illuminated the gap between controlled tests and consistent production usage. Recycling environments are more challenging due to their chaotic nature; while factory floors are structured and predictable, recycling belts carry random assortments of wet