Exploring the far reaches of space requires highly precise mission preparations, and Airbus has developed image simulation software to support all development phases – from preliminary sizing and feasibility studies to validation and testing.
This software – called SurRender – is particularly useful for the development and testing of vision-based navigation techniques.
Continuously developed by Airbus Defence and Space since 2011, SurRender has been used in many projects, including the development and validation of image processing algorithms for the European JUICE mission (JUpiter ICy moons Explorer), which is planned for a 2022 launch to perform detailed observations of the solar system gas giant Jupiter and three of its largest moons. The simulator uniquely models specific physical effects that are crucial for computer vision algorithms. It is widely used by Airbus teams and it is at the heart of the development process for various space projects, from early prototyping, to extensive performance test campaigns and hardware-in-the loop experiments.
Key features of SurRender:
The software handles various space objects such as planets, asteroids, satellites and spacecraft,
It is capable of accommodating solar system-sized scenes without precision loss, and optimises the ray tracing process to explicitly target objects.
Physically-accurate images are produced with quantitative radiometric values, expressed in physical units (irradiance value per pixel in W.m-2, or spectral irradiance in W.m-2.Hz-1), and SurRender performs complete sensor modelling, including PSF, distortion, motion blur, electronic noises, rolling shutter, and other low-level detector effects.
Powerful and flexible simulation software
SurRender is highly flexible. Supported objects include digital elevation models (DEMs), meshed objects, and analytical shapes such as ellipsoids with local DEMs (planets).
All surface properties can be customised with user-specified reflectance models, textures and normal maps. Lambert, Hapke, and Oren-Nayar reflectance models are included by default.
Projective models can be customised as well. Proposed implementation configurations include pinhole model with or without distortions, pushbroom sensors, and scanning sensors. In addition, low-level electronic behaviour of detectors can be specified, including response non-uniformities, noises and acquisition modes (binning, windowing, rolling shutter, etc.)
SurRender is suited to various simulation setups and needs:
As of Feb. 2020, the latest version of the software is SurRender 6.1.
SurRender works with a client-server protocol (TCP/IP). The SurRender client Python interface is open-source. Users are encouraged to share their contributions on SurRender GitHub repositories: https://github.com/SurRenderSoftware
Recommended minimal computing requirements:
OS: Linux or Windows, 64 bit
Processor: Quad-core desktop processor
Graphics card: NVIDIA GPU with 4 GBytes of dedicated video memory
API: Python 3, Matlab, Simulink, C++, Java, lua
The performance will depend on scene complexity and required quality (e.g. real time vs algorithm training). SurRender is parallelized and can be deployed in a cloud platform.
The SurRender software is a product of the Advanced Studies team of the Sensor Processing Chain department at Airbus Defence and Space, in support of Computer Vision activities.