aiSim™ — 5.1.0 release notes

Welcome back to another edition of aiSim’s Bi-Monthly Release Highlights! We’re thrilled to continue sharing the latest breakthroughs and enhancements from our cutting-edge simulation platform, keeping you in the loop with the forefront of automotive simulation technology.

What’s Inside aiSim’s Bi-Monthly Releases?

• ✨ New Features: Explore the latest additions to aiSim that will elevate your simulation experience.
• 🔧 Improvements & Fixes: We’re committed to making aiSim better with every release. Check out the enhancements and fixes that keep your simulations running smoothly.
• 📚 Full Release Notes: Subscribe to our newsletter, and let us know if you want even more — we can provide access to the comprehensive release notes, including intricate details of the updates.

We are happy to announce the 5.1.0 release of aiSim! In the new version, we introduced visibility mask ground truth output to simulate sensor blockage in a more realistic way and introduced obstacle avoidance for pedestrians in the scenarios.

We introduced the support of RoadRunner and HERE maps with the aiSim map importer (Atlas) and reworked the vehicle coloring system to achieve a more realistic vehicle appearance. Moreover, we added several new assets, such as speed bumps and crane models for urban maps, Japanese buildings, lamps, traffic signs, and vehicles.

New Features

Visibility mask Ground Truth output: Added new image output for camera sensors called ‘visibility mask’. This output is a grayscale image containing the combined strength of a configurable combination of various environmental effects such as atmospheric fog, precipitation, or sensor blockage. The sensor blockage Ground Truth output is replaced by this new visibility mask, containing continuous values instead of the four strength categories in the sensor blockage Ground Truth.

Color and visibility mask GT output

Feed events into sensor plugins: Added ADAS/AD specific signals to the aimScenario, and extended the global movement log with them to enhance the resimulation capabilities.

Stopping at and avoiding obstacles: Pedestrians can be configured to avoid collision with scenario obstacles either by stopping in front of them or by calculating a new route to avoid them.

aiSim GUI Scenario Editor — Phase 5: Extended the Scenario Editor — AIM workspace with a snap & move-on-lane feature. By selecting the proper tool in the Editor, users can snap objects to the closest lane and move them along the lane center line with the help of a new viewport widget.

OpenSCENARIO lib network capabilities: Added preliminary support for overriding the flow of an OpenSCENARIO from a client application.

OpenSCENARIO in resimulation pipeline: Global movement logs can be converted to valid XOSC files to be replayed with aiSim.

Driver off event: Lateral and longitudinal ‘driver off’ events can be described in aimScenarios. The scenario developer is able to describe a situation where the driver stops driving the vehicle (hands off of the wheel or not using the pedals).

Distributed real-time simulation: Added support to distribute a real-time simulation to multiple aiSim servers.

Import custom parameter matrix: Made possible to upload CSV format as input files in the aiFab workspace to define the parametrization of scenario variables with externally generated data.

Web deployment of aiFab and Large-scale workspace: Added a new web application deployment unit called aisim_gui.html, to have an initial web-based version of the aiFab and the Large Scale workspaces.

Bazel 7 update: The example toolchain now provides an example which is compatible with Bazel 7 build system.

Improvements

Made the following atmospheric rendering improvements:

• Skyglow now properly illuminates fog.
• Sun disk is now visible in overcast and foggy environments.

Improved image quality when using a raytrace camera sensor with HighQuality denoiser configuration.

Added a new configuration parameter (max_denoiser_memory_mb) to the raytrace camera to limit the amount of GPU memory used by the denoiser. This allows avoiding cases of insufficient GPU memory (typically with many high resolution camera sensors), at the cost of performance.

Made the following AimScenario Editor improvements:

• Last working directory is remembered to ease user workflow.
• Revision number is incremented with each save to properly administrate how a scenario file develops.

Parameters of aiFab can now be varied independently with a mixture of strategies in the same test case: randomly, following a grid pattern, or imported from external data sources.

Enabled smart optimization of parametrized test cases created with aiFab. Based on the initial batch, new test samples are iteratively generated and executed to explore the space and find interesting results.

The --ego_vehicle_config and --aimscenario_config CLI arguments for aiSim clients are now optional. The structure of the simulation config JSON files are changed. The “config” keyword is removed, and all the necessary fields are listed.

New 3D content features

aiSim map generator (Atlas) — RoadRunner import: Extended aiSim’s Atlas map importer pipeline to support the import of RoadRunner and HERE maps. With the help of Atlas, you can now bring RoadRunner–, OpenDRIVE–, and HERE map data to aiSim, displaying the 3D assets from RoadRunner and having an aiSim-ready, maneuverable road network in each format.

Atlas — RoadRunner import

Vehicle color rework: Reworked the vehicle coloring system by updating the VehicleDB JSON, removing unused parameters, and prioritizing coloring based on sRGB decimal values. The new colors consider vehicle color statistics and allow for a more descriptive and greater variability of colors. The resimulation workflow benefits from this improvement as the automatic vehicle color assignment becomes more accurate with the wider color scale.

Reworked vehicle colors

New 3D assets in the Asset Library

• Speed bumps
• Japanese pavement markings
• Japanese lanterns
• Japanese building props
• Japanese building facades
• Japanese buildings
• Japanese low-poly buildings
• Budapest suburban building elements
• US-style suburban building procedural elements, advertising boards, and streetlights
• Crane model
• Japanese-style buildings and Japanese traffic signs

New Japanese assets in aiSim 5.1.0

New vehicle models

Added the following vehicle models:

• Nissan Sakura
• Honda N-Box
• Daihatsu Hijet Jumbo

Nissan Sakura, Honda N-Box, and Daihatsu Hijest Jumbo vehicle models in aiSim

Map updates

Created a new map, Parking_US_SanJose map, a digital twin of an open-air parking area in San Jose, US.

Decorated the ZalaZone_SmartCity map, a digital twin of a dedicated ADAS & AD test track, located in Hungary, with sidewalks.

Budapest Urban map updated with road bumps

Improvements

Disabled Shadow casting by default for certain low intensity lights.

Subscribe to our newsletter today on LinkedIn to receive access to the full release notes and unlock the potential of aiSim.