Robots that know where they are
Kritrim means artificial in Sanskrit and Hindi. कृत्रिम-इंटेलिgence is an early stage robotics and AI startup that puts SLAM and deep learning at the centre of everything it builds.
Why SLAM
A robot can only be useful if it knows where it is. Outdoors, GPS usually answers that question. Inside a warehouse, a tunnel, a factory or a dense city block, it does not. Simultaneous localization and mapping lets a robot build its own map from its sensors and track itself within it.
Deep learning is changing what SLAM can do. Large pretrained 3D models now help robots map with plain cameras, handle blank walls and recognise places they have seen before. Reinforcement learning then helps them decide what to do with that map. We work where these three fields meet.
The journey so far
Abhishek Tyagi is a mechatronics engineer and roboticist, and a graduate student in robotics at Purdue University in West Lafayette, Indiana, USA. कृत्रिम-इंटेलिgence is the startup he is building, and it is still in its very early days.
The journey started with Veg, a quadcopter that flies without GPS. He built the first version as his B.Tech project in 2021 and kept improving it, adding SLAM, LQR control, fault recovery and onboard vision. That work was published as a preprint on arXiv in 2025.
Along the way he designed an object detection rover during an internship with the Defence Research and Development Organisation (DRDO), and spent four years at Deloitte as a Data and AI Engineer, building data pipelines and predictive models. Each step added a piece: robotics, AI and data engineering. The goal now is to bring them together into robots that work in the real world, not only in the lab.
