Adaptive Robotic Systems for Orbital Infrastructure

The construction and maintenance of sustainable orbital and lunar infrastructure including persistent platforms, space stations, space based solar power, space stations, large aperture observatories, and large fuel depots requires flexible, autonomous robotic assembly systems. PickNik Robotics proposes to develop an AI-driven robotic assembly system to revolutionize construction in space. Our solution addresses key limitations in current space-based robotic assembly by leveraging Large Language Models (LLMs) to autonomously generate behavior tree-based task plans from high-level procedural descriptions, removing the need for manual programming. Our system integrates real-time task success verification, allowing robots to self-assess and recover from errors autonomously, reducing reliance on human intervention. With funding from this SBIR Phase I, we will develop and validate our technology using MoveIt Pro, our advanced robotics development platform. We will use our simulation platform to generate training data and simulate orbital assembly conditions. We will train and test LLM-based task planning and verification models, equipping them with anomaly detection and autonomous correction capabilities. By demonstrating feasibility in high-fidelity simulations, we will lay the groundwork for future hardware integration in Phase II. Our target markets include NASA’s In-Space Servicing, Assembly, and Manufacturing (ISAM) and Moon to Mars initiatives, commercial satellite servicing, and future deep-space infrastructure projects. Beyond space, our technology extends to terrestrial automation in advanced construction, manufacturing, and logistics. By enabling intelligent, adaptable, and scalable robotic assembly, we are positioning this innovation to drive down costs, increase efficiency, and make long-term space infrastructure viable.