This project seeks to improve how machines understand and navigate our world by combining advanced mapping techniques with cutting-edge localization technology. Visual-Inertial Simultaneous Localization and Mapping, or VI-SLAM, is a technology that allows devices to determine their position while mapping the environment around them, all in real time. Our innovation involves integrating Gaussian splatting, a sophisticated data processing method, to significantly improve the accuracy and efficiency of VI-SLAM.
Goals and Challenges:
Our mission is to make VI-SLAM more adaptable and capable of handling complex environments, from the interiors of buildings to crowded city streets and unpredictable natural landscapes. The main challenge is processing the vast amounts of data these systems generate quickly enough to operate in real-time without losing accuracy.
We are taking a step-by-step approach, starting with theoretical developments, moving on to algorithm creation, and finally testing our innovations in both controlled and real-world scenarios. Our focus is on refining how these systems process spatial data to make them more versatile and effective.
Impact and Collaboration:
The outcome of our work aims to set a new standard for autonomous navigation technologies, offering more reliable and precise mapping solutions that could transform everything from robotics and augmented reality applications to autonomous vehicles. We’re not just advancing a technology; we’re opening up new possibilities for how machines interact with the world.
The project is funded by Carnegie Robotics LLC.