Use WorldGenerator™ to edit 3D agriculture simulation environments for synthetic dataset generation, and validation and verification of autonomous systems.
Perception models can be joint-trained with real world and synthetic data. Virtual 3D worlds can be used to gather synthetic datasets to train these models.
Pixel-perfect annotation is a benefit of synthetic data. Our datasets can therefore reduce the cost of training data, by providing a cost-efficient alternative to real world, manually annotated datasets.
WorldGenerator™ is our proprietary platform that creates 3D simulation environments. From road signs to potholes, our worlds are highly customisable to suit your specific data needs.
Create urban simulation environments that can be used for synthetic dataset generation to train perception systems.
Create fictitious sites with path networks to simulate in. Define drivable path for agricultural vehicles.
Auto-annotated assets including crops, people, animals, trees, vehicles, obstacles and more.
Create deformations, such as potholes, on the path to replicate the surface you’d drive in reality. This way, your perception systems are truly being tested to identify the drivable path in an environment closer to reality.
Export your custom virtual world into your own simulator. We support industry standard file formats.
We create the tools for you to build custom 3D simulation environments for autonomous vehicles.
© Repli5 AB. All rights reserved.