Otto Group and robotics software company Sereact have announced a plan to put AI-controlled robots to work on one of online retail’s hardest jobs: inspecting and sorting returned orders. Ten dual-arm systems will start at an Otto Group logistics centre in Poland, with 15 more planned in a second phase if the first results hold up, taking the programme to 25 robots.
Returns are notoriously difficult to automate. Items come back folded, crumpled or repackaged, and a multi-category retailer’s assortment never stops changing, so traditional automation built around known products and predictable presentation quickly breaks down. Sereact’s answer is Cortex, a physical AI model designed to generalise to objects it has never seen, deciding in real time how to grasp each item, judge its condition and route it onward — without product-specific teaching or a fixed item catalogue.
Cortex is trained and validated on Nvidia’s physical AI stack. Sereact uses Isaac Sim and Isaac Lab to turn real picks from the returns line into thousands of simulated variants, and Cosmos world foundation models to re-render them across garment states, packaging, materials and lighting. New policies are post-trained on Blackwell GPUs and tested before reaching the robots, so fixes can arrive within days.
The robots will be introduced in stages under a Robotics-as-a-Service model, with Sereact operating and maintaining the systems. Staff move from repetitive grasping and lifting toward supervision, exception handling and final quality decisions. The project sits within Otto Group’s medium-term commitment to invest €350 million in technology and AI, and a broader rollout will depend on whether the system meets the reliability, flexibility and quality standards of large-scale returns processing.
The same generalisation idea drives assistants such as ChatGPT and Gemini — compare them in our ChatGPT vs Gemini comparison.
Source: GlobeNewswire
