p71924506
/
carla
mirror of https://github.com/carla-simulator/carla.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

added details from james

This commit is contained in:
MattRoweEAIF 2023-09-26 15:57:00 +02:00 committed by Blyron
parent 53a7a15fa0
commit 1510f7522d
2 changed files with 38 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
Docs/ecosys_synkrotron.md
View File

@ -1,13 +1,46 @@
# Synkrotron simulation solutions
<img src="../img/synkrotron.jpg" alt= “synkrotron_logo” style="display: block; margin-left: auto; margin-right: auto; width: 70%;">
Synkrotron provides advanced solutions for autonomous driving simulation, built on top of CARLA's existing toolset to streamline scenario authoring, simulation data pipelining, sensor modelling and cloud based compute scaling tools for streamlining the AD research and development process.
# Synkrotron simulation solutions
Synkrotron provides advanced solutions for autonomous driving simulation built on top of CARLA. Its product suite Oasis supports ​a broad range of applications including scenario generation, sensor modelling, ​traffic simulation and data management. Based on a flexible architecture Oasis ​can be deployed on the cloud at scale, or in a developer's local environment for prototyping.
---
# __OASIS platform__
<img src="../img/oasis_logo.png" alt= “synkrotron_logo” style="display: block; margin-left: auto; margin-right: auto; width: 70%;">
## [__OASIS simulation platform__](https://www.synkrotron.ai/sim.html)
The OASIS simulation platform provides a set of tools to augment CARLA's functionality with a graphical scenario authoring tool, a built in scenario library, a scenario description language, a REST API for integration with DevOps and CI/CD piptines along with cloud computing tools for scaling AD simulation with massive concurrency.
Oasis Sim is a full-fledged scalable simulation platform with CARLA at its core. It supports the complete life-cycle of AD simulation: scenario import & editing with a graphical user interface, sensor configuration, distributed task management as well as diagnosis through rich simulation data and logs. Both cloud and local deployment are available through containerized ​packaging. Comprehensive APIs are exposed for integration with DevOps too. You can [request a trial](https://synkrotron.ai/contact.html) from Synkrotron.
## [__OASIS data platform__](https://www.synkrotron.ai/data.html)
The OASIS data platform provides a data management system for managing the high volume of data required throughout the autonomous driving R&D pipeline. The OASIS data management solution helps managing data pipelines to streamline the transferral of data through data acquisition, storage, filtering, annotation, scene extraction and finally into algorithm training, testing and deployment. OASIS provides a user friendly management interface to control all aspects of the data pipeline.
Oasis Data is a platform for managing the high volume of data flowing through the autonomous driving R&D pipeline. Data-driven development of AD systems are made possible with Oasis Data via the following functionalities,:
* data acquisition and anonymization
* multi-stage filtering based on structured (CAN bus signals, active-safety triggering, etc.) and unstructured (sensor readings)
* information mapping and environment reconstruction with lidar and/or vision-only strategy
* auto-labeling of data with pre-trained perception models
* scenario tagging and reconstruction with OpenX format output
The processed data from the platform supports downstream applications such as scenario re-simulation with Oasis Sim, retraining of perception models with new labeled data and operational fleet management.
## __Synkrotron Tools and Services__
In addition to the complete solutions introduced above, Synkrotron also offers the following developer tools and services to facilitate AD simulation development.
| Product/service | Description |
| -----------| ------ |
| __Sensor model__: fish-eye camera | Fisheye camera with configurable distortion parameters |
| __Sensor model__: LIDAR | Advanced lidar model with configurable shutter mode and material reflections |
| __SOTIF__ scenario generation tool | Given the ODD description from user, identify important scenario elements and generate critical scenarios to help uncover unknown unsafe domains of the autonomous driving systems, achieved by using an ontology-based method combined with iterative optimizations against evaluations in Carla |
| __Map creation__ | Given lidar scan data (can also provide road-test data collection equipments & service), create HD maps for the user and output OpenDrive files which can be loaded in Carla for PnC tests or logsim |
| __Sensor model__: physics models for camera | Supports CMOS simulations and outputs 12bit raw sensor data, for customers who develop ISP algorithms or who has ECUs that need raw data from camera instead of RGBs |
| __Static scene creation__ | Combined HD mapping, 3D asset development and procedural modeling to create 3D static scenes/digital twins for users |
| __Dynamic scenario reconstruction__ | On top of static scenes, also detect and track various traffic participants from user's road-test data; the recovered trajectories/behaviors are turned into OpenScenario 1.0 files and can be re-simulated in CARLA |

BIN
Docs/img/oasis_logo.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 66 KiB