The lean hardware footprint and use of low-cost, vehicle-independent sensor pods makes for easily deployable survey equipment that can be used on rental cars anywhere in the world.
Vehicle-independent sensor pods eliminate the need for purpose-built survey vehicles.
atlatec’s approach to 3D mapping relies mainly on stereo vision, using two front-facing cameras as the primary sensor. Our sensor pods place them 40 cm apart from each other, which makes for a self-contained, portable setup that is only about the size of a suitcase – without compromising on depth-image resolution.
Camera-based feature tracking and bundle adjustment allows for exceptional quality of trajectory recreation and spatial referencing, compensating for the larger part of IMU-related drift. Learn more about the accuracy we achieve here.
A look under the hood: atlatec sensor pods contain stereo cameras, an RTK-GNSS receiver and a motion sensor/IMU.
Our data processing pipeline takes care of GNSS correction, sensor fusion and loop closure, resulting a single, consistent 3D model for every route – with consistency in the sub-centimeter range.
Data processing is done in post-production on our servers in Karlsruhe, Germany and almost 100% automated.
Our proprietary annotation software is heavily optimized for our sensor setup and allows us to leverage AI for many tasks, such as identifying lane borders and inferring the topology of intersections. Human annotators manage quality control and more complex tasks. The degree of fidelity/object-richness depends on individual customer requirements.
Our project management software "atlatrack" is purpose-built for mapping work: It allows us to efficiently manage a distributed workforce on projects of any size, reliably enforcing required quality control and feedback loops as well as managing interdependencies between tasks.