2021.3 Release Features#
With Docker integration, we have extended scenario runners support on SVL Simulator running on Windows. Now it is possible to run end-to-end simulations using Python API and Visual Scenario Editor runtime templates on Windows as well. Please follow our updated installation instructions to get started.
With Local Automation, you can now trigger simulation jobs from scripts without needing to manually interact with the web user interface. This empowers you to begin integrating SVL Simulator with existing automated testing or CI workflows (like Gitlab or Jenkins).
We have several features to extend support for robotics simulation use cases. We support multi-robot simulations by using namespaces on sensor topics. Please refer to our multi-robot sensor configuration for an example. We also added new sensors (Lidar2DSensor, DestinationSensor, IntelRealSense) and new indoor environments (Warehouse, WarehouseDT) with costmaps for robotics development. We have created an SDF importer to help import robot and environments from SDF world files. The Python API has been extended to enable setting a NavOrigin and destination with ROS2 Navigation stack. Finally, we have added support for indoor lighting and rain effects for creating improved indoor environments.
There are some updates on our user uploaded content license terms. Please review the updated license terms at our website.
We will be deprecating official Simulator support for Ubuntu 18.04 starting with our next 2021.4 release.
Perception sensor 2D and 3D will now use image data to create tight bounding boxes to increase coverage and accuracy. If any pixel is visible, entity will be detected by the ground truth sensor. Vehicles no longer require box colliders, bounds are calculated based on mesh.
You can now create more complex and improved scenarios using our visual scenario editor. It supports BezierSpline paths, acceleration for agents’ waypoints, improved controllables policies, support for NavOrigin for robotics, and improved scenario playback to validate before running in simulation.
You can now use custom NPC Managers in NPC behavior plugins to create complex behaviors (like Parking).
We have moved simulator input actions to sensor bundles to make it easy to add new input controllers like gamepads and other wheel controllers. You no longer need to make changes to the simulator source code for adding new input devices.
For a full list of bug fixes, improvements, and all other changes since the last release, please see the release notes.