Technology

AI

We develop and use many deep-learning models to perform such difficult tasks as object detection, traffic light recognition, road marking detection. Our technology allows trucks to see vehicles and pedestrians several hundred meters away. With incredible AI prediction, we are able to react 10 times faster than a human driver.

Localization

Self-driving starts with understanding where your truck is. We should not rely on GNSS because of dead zones, also road markings are often missing or incorrect. Our technology uses lidars and radars with precise HD mapping to localize autonomous trucks with up to 2 cm. accuracy.

Simulation

In Moove we developed Scene Templator which allows us to generate millions of scenarios for our simulation environment. We use Simulation to teach our trucks to drive safely and smoothly and to generate synthetic dataset for our deep-learning models. For every 1 kilometer driven in real life, there are 1000 km in simulation mode.

Mapping

Trucks moving from point A to B use our own semantic map with many layers: road lanes, traffic lights, speed limits, crosswalks, etc. In Moove we use AI and Moove Map Editor to provide the most accurate maps of routes. We detect road signs and markings using recorded lidar and camera data, as well as satellite images, then our editors manually review automatically created maps. Maps are automatically updated whenever changes occur, such as new construction zones or road markings changes, etc

Hardware

Each Moove autonomous truck is equipped with specially designed and built computer modules, set of lidars, radars and cameras. We developed Moove Truck Sensor Kit (TSK) to convert regular truck to self-driving truck just for several hours. Our hardware installation does not interfere in trucks drive-safe features such as ABS, ESP, etc., communicating with vehicle through CAN interface.