Unity’s SimViz Solution Template gives you everything you need to get started and accelerate the building of your simulation environments, so you can focus on autonomous simulation and scaling your program.
Many transportation and robotics companies are actively building toward a future where machines – whether automobiles or robots – will be deployed to perform specific tasks while navigating completely on their own. To achieve this safely and effectively – both in terms of costs and development speed – companies need to train algorithms inside virtual environments that represent the real-world as closely as possible in order to capture and visualize simulation data for actionable results.
Currently, there is not yet a mature ecosystem or solution for autonomous simulation and so the industry has yet to come to a consensus regarding the entire simulation pipeline.
As a result, companies are working through solving for:
- The content in which they can run a simulation: cities with cars, pedestrians and vehicles with sensors that see in the same way actual physicals sensors will perform.
- The tools to create the tens of thousands of scenarios, which they want to run in simulation.
- A platform that can integrate into their larger system for simulation, including choosing their own physics models, integrating hardware in the loop, etc.
- Only once the first three are addressed can they then move to the fourth stage, simulating at scale in the cloud.
Get started designing autonomous driving simulation solutions
We are creating SimViz Solution Templates in partnership with CVEDIA and INDAGO to address the first of the areas that companies are currently solving for: the content in which they can run a simulation. The goal is to provide those who just started designing autonomous driving simulation solutions with a quick reference design, so they can get a feel for the ease of use of Unity’s development environment and the richness of Unity’s development ecosystem.
While the assets in these packages do not represent any real-world environment, they do provide a realistic-looking representation of a fictional city and terrain. They are intended to serve as reference designs to jumpstart training and validation of autonomous driving functions, so it is important to note that they do not represent validated simulation models (especially when it comes to sensor logic and dynamic elements), nor were the sensors modeled out of real physical components – or specific products – in existing autonomous systems.
They can, however, be completely modified, and some will find them quite appropriate to support their work on higher level Machine Learning (ML) models applied to autonomous vehicle stacks out of the gate.
Unity SimViz Solution Template: What’s included
The Unity SimViz Solution Template package includes sensor models, referred to as SynCity Lite, scene models, and supporting documentation. The parts can work together or separately, depending on the use case.
Sensor models included:
- RGB-D Camera
* Not included in early access package.
Sensors and scenes from Unity’s SimViz Solution Template in action together. Shown here: two LiDAR sensors and one RGB-D camera from the Sensors package are shown here interacting with the City Streets scene.
Reference scenes include:
- City Streets – A section of a typical city fabric similar in style and layout to metropolitan areas of US, such as Chicago, New York, Seattle, and Los Angeles although not representative of any actual city.
- Open Highway* – Miles of open highway road containing varied terrain variations, entrance/exit, and road junctions.
- Residential Streets* – A section of a typical suburban American development following the superblock model designed to reduce through traffic, including local arterial streets, cul-de-sacs, and non-grid layouts.
* Not included in early access package.
We are currently focused on working with a select number of customers to implement projects using these templates. Early access versions are available today, while full versions will be available for Unity strategic partners by the end of 2018.
The best simulation platform for autonomous driving
We are committed to becoming the best simulation platform for autonomous driving, with our real-time 3D engine at the core. We aim to further enable our ecosystem partners and make building simulation solutions as easy as it can be.
We are developing a more comprehensive solution for SimViz to address the functional areas highlighted below:
Our plans for the coming year include:
- Support for ROS (Robotic Operating System), for integration with automation and robotics functional modules
- Support for OpenDrive and other industry standards
- Sensor-ready contents to standardize the compatibility of scenes/scenarios developed with Unity and sensor models developed independently, to encourage best practices and achieve best performance
- Integration of Unity ML-Agents