Katrix.AI

What they're building: Katrix appears to be commercializing this technology in two forms. The first is a Perception SDK, designed to integrate into existing robotics stacks through C++, Python, and ROS2 interfaces. The second is a hardware “Black Box” edge module, built on NVIDIA-class computing platforms and capable of ingesting multiple sensor streams for real-time processing.

Target market: Autonomous drones, smart infrastructure, industrial automation, maritime systems, energy monitoring, and defense-related applications, all sectors where reliable edge perception remains a major bottleneck.

Technical thesis: The system ingests inputs from multiple sensors, including LiDAR, radar, cameras, IMU, GPS, thermal sensors, audio, and ultrasound. It combines them through a fusion architecture that integrates machine learning models with deterministic processing pipelines. Rather than relying solely on deep learning perception models, the company emphasizes a traceable, explainable architecture for environments where safety and reliability are critical. The platform produces outputs such as: object tracking and environmental mapping, vehicle localization and pose estimation, contextual interpretation of surroundings, risk assessment, and safety signals for decision systems

Current strategy: The company is positioning itself as a provider of mission-critical perception systems for autonomous machines, combining AI with deterministic safety frameworks. The company’s technology narrative, founder background, and corporate timing suggest that Katrix may represent a strategic reboot of earlier robotics work rather than a completely new startup.