Max Robotics — Global Service Robot Directory + US Certification

About

Ivan Kapelyukh is a robotics researcher whose work sits at the intersection of computer vision, natural language processing, and robot manipulation. His primary research focus is enabling robots to understand and rearrange household objects in a human-like, intuitive manner. Kapelyukh’s most significant contribution is **DALL-E-Bot** (2023, 86 citations), the first system to integrate web-scale diffusion models into a robotics pipeline. This work allows a robot to observe a cluttered scene, infer a text description of the objects, generate a target image of a tidy, natural arrangement using DALL-E, and then physically rearrange the objects to match that image. He extended this concept with **Dream2Real** (2024, 16 citations), which leverages vision-language models (VLMs) to perform zero-shot 3D object rearrangement, enabling robots to plan and execute complex tasks without any task-specific training. Earlier, in **NeatNet** (2021, 7 citations), Kapelyukh tackled the challenge of personalizing robot behavior by using Graph Neural Networks to learn a user’s unique tidying preferences from just a few examples. Through these works, Kapelyukh is pioneering a new paradigm where robots don’t just follow rigid commands but instead leverage large, pre-trained models to understand context, aesthetics, and user intent, making them more adaptable and helpful in real-world homes.