The new Light Propagation Engine, also referred to as LPE, released as a beta feature, is a powerful rendering engine that you can now use for camera simulations instead of the default rendering engine.

The Light Propagation Engine not only offers a better and more realistic rendering, it also increases the simulation speed.

If you want to use the Light Propagation Engine for camera simulations, you need the AVxcelerate Sensors Library for LPE, which provides you with resources (assets and tracks) specifically prepared for this rendering engine. To do so, you must select this option when installing the AVxcelerate Sensors solution.

You will then have to configure the LPE Parameters in the camera simulation parameters. You will find three examples of simulation parameters files in the dedicated library.

The DDMA feature, which stands for Direct Data Memory Access, is a new method to get the output data produced by sensors.

Particularly suited for HiL and SiL use cases, DDMA offers better performance compared to data access from the shared memory since it provides direct access to raw data in CPU.

Thanks to the DDMA Software Development Kit (SDK) delivered in the product, you can develop your own dynamic link library (dll) customized to your needs.

The RDMA SDK is composed of:

For more information, refer to the readme file provided in the sample folder or to the Raw Data Output for DDMA page of the User's Guide.

In order to perform a radar simulation in a context which more closely resembles real word scenarios, in the 2025 R2 release of AVxcelerate Sensors Simulator, it is now possible to simulate phase noise in radar modes with a System FMCW or Performance FMCW waveform. To create a radar model with phase noise, use the Sensor Labs application, or the Sensors REST API to model phase noise in one of three ways:

To simulate phase noise, simply add your radar to a sensor layout, configure the radar simulation parameters, and run the simulation. For more information on this parameter, refer to the Mode Configuration Parameters.

To counteract the aliasing phenomenon and improve the signal-to-noise ratio (SNR), you can now add range filters to your sensor model using the Sensor Labs application or the Sensors REST API. Set at the mode level, you can configure up to two range filters per mode in a radar model. You can use one of three ways to define attenuation and create:

For more information, refer to Range Filter or to the Mode Configuration Parameters in the AVxcelerate Sensors Simulator User's Guide.

In the 2025 R2 release of AVxcelerate Sensors Simulator, you can optimize performance when executing consecutive simulation runs thanks to the ability to perform a simulation reset. Using the simulation control procedures, you can now call an Initialize command directly after a Stop command. This triggers the resetting of the scenario using the resources that were loaded from the initial simulation for each subsequent simulation run, thereby optimizing on load time. In the current release:

When a simulation reset is performed and depending on your data access settings, the output files are organized accordingly. Refer to the Results section of the AVxcelerate Sensors Simulator User's Guide for more information.

Leveraging National Instruments RDMA technology, the RDMA feature of AVxcelerate Sensors Simulator allows you to transfer raw data generated by physics-based camera sensors to an RDMA receiver.

In the previous release, you could transfer camera data via RDMA only by using the pre-built RDMA sender available on demand.

In the 2025 R2 release, you now have two options to transfer raw data via RDMA:

Using a custom dll has the many advantages over using the pre-built RDMA sender.

First, when developing a custom dll, you can add any data post-processing or conversion steps you need, for example to convert RGBA data to aRGB, or to turn the spectral irradiance map data produced by a camera with the Lens Output (Light) output type into an RDMA compatible format.

Also, the sensor identifier being provided as a metadata, along with all the information needed to decode the data, it is possible to get the output data of several camera sensors. When using the pre-built RDMA sender, the sensor identifier is not part of the metadata, data identification is based on the timestamp only, which is error-prone when retrieving the output data of several sensors.

Finally, the connection between the receiver and the sender can be established faster.

In order to ease the configuration of sensor layouts including thermal camera models, the Sensor Labs applications and REST APIs now partially support thermal cameras.

You can now:

Layout containing a thermal camera	Thermal camera in Sensors Library

Important notes: The thermal camera sensor models cannot be opened or edited using the Sensor Modeler application, nor be created, updated or read using Sensors REST API v7.

The simulation of a thermal camera sensor now produces temperature maps only, available as text files or as serialized byte arrays in the shared memory. Grayscale images in png format are no longer generated.

This is a breaking change that requires you to update the simulation parameters of your thermal camera sensor so that the camera recording format must be set to TEMPERATURE_MAP instead of PNG if you want to save the thermal camera output files to disk.

3D bounding boxes are now part of the ground truth data that camera sensors can generate.

The 3D bounding box data are generated along with the 2D bounding box data when the generate2dBoundingBoxes camera simulation parameter is set to true. The 2dboundingboxes.xml files now includes both the 2D and 3D bounding boxes.

While a 2D bounding box is defined with pixel coordinates in the image frame, a 3D bounding box is defined by:

For more details, refer to the Bounding Box section of the Camera Ground Truth Output Data page.

The following sensors simulation improvements have been made in the AVxcelerate Sensors 2025 R2 release:

Ubuntu 24.04 is now a Linux distribution compatible with AVxcelerate Sensors Simulator. Ubuntu 20.04 is no longer supported. For the complete list of compatible operating systems and distributions, refer to the Platform Support by Application / Product document published here on the Ansys Website.

When installing the AVxcelerate Sensors solution, its main software components are no longer installed in Optical Products/VRXPERIENCE on Windows and OpticalProducts/VRXPERIENCE on Linux, the subfolders are now named Autonomy/AVxcelerateSensors. The same changes have been made to the folders in %appdata%, %localappdata% and %programdata%.

The below-listed executable, log and configuration files have also been renamed.