Sensor Fusion Algorithm Development

BASELABS Create is the tool for the development of environment perception applications for automated vehicles.

Data fusion designer

  • Supports relevant driving functions like AEB, ACC, lane keeping and piloted driving in highway and urban scenarios
  • For all automotive sensors like cameras, radars and lidars
  • Seamless workflow from the graphical configuration of the data fusion to an executable application in a middleware
  • Source code access to the data fusion application
  • Available for Windows and Linux, x86, ARM and PowerPC architectures
  • Example data fusion applications including sample sensor data available for ROS, RTMaps and vADASdeveloper
  • Complementary data fusion expert support is available

Configuration

Loading the player ...

BASELABS Create reduces the workload for the data fusion developer. As a user, you provide system information like the sensor (mounting position), model and sensor parameters (like the field of view and the detection rate as well as the false alarm rate). Existing models can be re-used as well. Then, you select different models with the GUI or via a programmatic call into the SDK. In most cases, the selection of a system model, a sensor measurement model and a sensor detection model is required. In addition, the selection of a track management strategy is necessary. With that information, BASELABS Create handles the 'calculation aspects' of the system: the sensor data fusion, the hand-over between sensors, all probabilistic calculations and the synchronization of the sensors.

Customization

Loading the player ...

The data fusion designer of BASELABS Create allows to setup and generate complex data fusion systems. To characterize each sensor in the setup, you select from either built-in or custom models, adjust parameters and then generate a data fusion algorithm using pre-defined algorithm templates. While this approach quickly provides results, some setups and use cases require to fine-tune the standard algorithm templates which BASELABS Create contains. BASELABS Create allows to completely customize data fusion algorithms, e.g. to use a specific algorithm for one sensor from the configured sensor setup while using standard algorithms for the other sensors. By this, fast algorithm generation by the data fusion designer can be combined with custom approaches.

Visualization

Loading the player ...

To analyze behavior and performance of a data fusion application, developers require a rich and customizable visualization that shows all relevant information in an integrated way. This includes the field of view of all sensors, the measurements from the sensors and the actual result as well as the reliability of the data fusion.

BASELABS Create visualizes data fusion results, the sensor fields of view and sensor data in a bird’s-eye-view according to the configured sensor setup. From that, data fusion developers can quickly analyze the behavior and performance of the data fusion system. In BASELABS Create, the visualization is auto-generated from the configured system and thus automatically adapts each time the configuration is changed, e.g. if the mounting position of a sensor is calibrated or a new sensor is added. The generated source code of the visualization is extensible and customizable so that additional information can be easily displayed. The visualization runs side-by-side in all supported middleware systems on Windows and Linux.

Diagnostics

Loading the player ...

In addition to the actual data fusion system, BASELABS Create generates a visualization from the configured sensor setup to diagnose the resulting data fusion system. Among others, this allows to verify the sensors mounting position, field of view and sensor characteristics like measurement noises, detection probability and false alarm rate. As this visualization is auto-generated from the sensor setup, it is automatically adapted when the setup is changed, e.g. if a new sensor is added or sensor parameters like the mounting position are changed.

Integrated plausibility checks

Loading the player ...

The integration of a data fusion algorithm to a middleware is a demanding task that sometimes introduces unintentional errors such as incorrectly converted time stamps or sensor data, invalid ego motion data or repetitive sensor measurements. Often, these errors are hard to find and thus require additional development time. BASELABS Create provides an optional and configurable plausibility check to detect typical errors in sensor data and its time stamps. This decreases the number of potential error sources and accelerates the integration.

Sensor data fusion

Arbitrary sensors and data levels

BASELABS Create allows to combine information from arbitrary sensors like cameras, radars and lidars. The data from these sensors can be combined on different data levels, for example tracks from smart sensors, raw data or the result of a data preprocessing using artificial intelligence techniques such as deep learning. The result is a stable environment model with a unified output - regardless of the sensors or preprocessing you plug into the system.

Sensors

Data fusion in the layer model

BASELABS Create closes the gap between middleware and driving function and provides all relevant building blocks for data fusion algorithms. The middleware provides the infrastructure of the system like sensor interfaces to acquire sensor data. BASELABS Create complements the middleware with the data fusion functionality. Most relevant to demanding development projects like a highway-pilot is the scalable approach of BASELABS Create. Sensors can be added easily, while the output that is used by the driving functions stays the same.

Challenges

The more complex the system under development, the more demanding gets the implementation process. BASELABS Create supports the developer in tackling the challenges of data fusion.

Cross Traffic

360° perception

Handover

False Negatives

False Positives

True Positives with Noise

Latency

Middleware integration

BASELABS Create is middleware-agnostic and supports the integration in systems like vADASdeveloper, ROS, AutoSar, RTMaps, Simulink and ADTF.

You can use the prototyping platform of your choice for a data fusion developed with BASELABS Create. Besides using your development PC for running the data fusion, it can also be used on all relevant target platforms like the Nvidia DRIVE PX 2, dspace MicroAutobox, Renesas R-Car H3 and others. Runtime licenses are available for the usage in test fleets or in-the-loop setups.

Middleware integration

Robot Operating System (ROS)

BASELABS Create provides a seamless workflow from the configuration of the data fusion to an executable application in ROS.

The Robot Operating System is a middleware which is commonly used in the industry for system integration. BASELABS Create adds the data fusion code that is required for implementing environment perception systems for automated vehicles.

Learn more about the data fusion development with ROS and BASELABS Create.

Middleware integration

ADTF

Middleware integration

vADASdeveloper by Vector

Middleware integration

RTMaps by Intempora

RTMaps is a middleware which is widely used in different industries for system development and prototyping. BASELABS Create automatically generates the source code for an RTMaps component that executes the configured data fusion.

Learn more about the data fusion development with RTMaps and BASELABS Create.

Data Fusion SDK

Modular Architecture

Parameters

Inline formula documentation

Besides the graphical configuration with the Data Fusion Designer, BASELABS Create can be used as a software development kit (SDK) to access expert features and increase the implementation flexibility. The user can call numerous building blocks of a data fusion system from the SDK and configure the data fusion in a programmatic way.

Use the operating system of your choice

Compatibility

Windows and Linux

Visual Studio

Mono Develop

You can use BASELABS Create either under Windows or under Linux. Under Windows, BASELABS Create is integrated into Microsoft Visual Studio and allows the development with the Data Fusion Designer as well as the programmatic usage of the SDK. Under Linux, BASELABS Create is used with Mono and allows the programmatic usage of the SDK.

Releases

Product improvements, new software features and bug fixing
BASELABS is working on permanent product improvements including new features or bug fixing. We inform our users about software updates and releases regularly.

BASELABS CREATE Releases

BASELABS Create 7.1

Fixed: The maintenance date of licenses is incorrectly handled.

BASELABS CREATE 7

New Features

  • Customizable data fusion: Generated data fusion systems can now be completely customized, e.g. to use a specific algorithm for one sensor from the configured sensor setup while using standard algorithms for the other sensors. By this, fast algorithm generation by the data fusion designer can be combined with custom approaches.
  • EKF based data fusion: The data fusion designer now generates both Unscented Kalman filter (UKF) and extended Kalman filter (EKF) based data fusion systems.
  • RTMaps integration: The data fusion designer now generates the code for an RTMaps component that executes a generated data fusion system. By that, data fusion systems can be easily integrated in RTMaps. Additionally, an RTMaps example project is available.
  • ROS integration: For ROS, the data fusion designer now generates the code for a ROS node that runs a generated data fusion system. An example project for ROS is available.
  • More visualization: The auto-generated visualization now additionally shows which measurements have been associated to an object including their association weights. This allows further diagnostics of data fusion systems.

Improved

  • On Linux, the data fusion designer now generates code from templates.
  • The time to generate a data fusion system has been significantly reduced.

Fixed

  • Fixed: When a sensor is deactivated in the data fusion designer, the generated API of the next sensors changes.
  • Fixed: The method 'GetTracks()' sometimes returns tracks with ambiguous IDs.
  • Fixed: The CASpace is not implemented as AngularSpace.
  • Fixed: The InflationSigmaHeading value in template projects is too large.
  • Fixed: The data fusion designer cannot be opened when project references have dependencies that are not present in the output folder.
  • Fixed: Template items cannot be added if the project contains NuGet packages whose dependencies cannot be resolved.
  • Fixed: The data fusion designer cannot be opened when a project references assembles with a different platform target.
  • Fixed: The visualization does not allow zooming when run on Windows and Mono.
  • Fixed: In the UKF data fusion template, the default alpha value for sigma points scaling has been reset to 0.1.
  • Fixed: When processing a template for a configuration that does not contain any sensor, an error message is shown in the error window of Visual Studio.

BASELABS CREATE 6

New Features

  • Data fusion visualization: BASELABS Create now visualizes data fusion results, the sensor fields of view and sensor data in a bird’s-eye-view. This helps data fusion developers to analyze the behavior and performance of the configured data fusion system. The customizable visualization is auto-generated and runs in all supported middleware systems on Windows and Linux.
  • Data fusion input plausibility check: Generated data fusion systems now optionally check for typical errors and inconsistencies in the sensor data, e.g. time stamp errors, incorrect sensor parameters like mounting position and data repetition.
  • Custom templates: BASELABS Create offers to adapt existing templates or to provide full custom templates to generate code that depends on the setup that has been configured by the Data Fusion Designer, e.g. to generate custom middleware integration, visualizations or diagnostics.

Improved

  • Runtime performance: The execution speed of data fusion systems has been significantly improved by leveraging CPUs with multiple processing cores.
  • Error handling: The error handling in the generated C-interop code has been improved. It now allows to access the full error message stack.
  • False alarm handling: The default value of the false alarm lambda has been decreased to better address smart sensors.

Fixed

  • In the smart camera detection model, no category is set for 'MaxFrontDetectionAngle' and 'MaxRearDetectionAngle'.
  • Access violation in generated C-interop code when running data fusion systems for a long time.
  • SigmaPoints class is not thread safe.
  • Invalid tutorial links.
  • The function 'ExitIfDirDoesNotExist()' cannot be found when compiling the generated C-interop code with Visual Studio 2010.
  • The ROS examples now contains the ROS message files.

BASELABS Create 5

This software release includes the following new features and product improvements:

New:

  • Data Fusion Designer for Linux: The Data Fusion Designer can now be used as stand-alone application in Linux. By that the complete workflow from the configuration of the sensor setup to the generated data fusion system becomes available in Linux. Please find further information about the Data Fusion Designer on our website.
  • Cross traffic support: The Data Fusion Designer now allows to select the new ‘Cross traffic radar track proposer’ and the `Smart camera detection model´ to better address traffic at intersections, e.g. for Automated Emergency Brake (AEB) systems in urban environments.

  • Extended object tracking: The new ‘Radar measurement model for extended objects’ can be selected in the Data Fusion Designer to improve the handling of objects in adjacent lanes, e.g. for Highway Pilot systems.

  • Middleware integration: The Data Fusion Designer now generates the code that allows to integrate the generated data fusion system into C/C++ environments, e.g. ROS and ADTF.

  • Data fusion diagnostics: The generated data fusion code now provides a diagnostics interface to support further analysis tasks.

  • New tutorials on data fusion and on ROS integration have been added to the user manual.

  • The user manual is available in Japanese.

 

Improved:

  • Data Fusion Designer: The names and the values of referenced properties are displayed.

  • Data Fusion Designer: The units of properties are displayed.

  • Data Fusion Designer: The ‘Tab’ key can now be used to switch between properties.

  • The SPRT algorithm can now be transformed by BASELABS Code.

  • Smart sensor track proposers: The smart sensor track proposers can be used for both moving and stationary objects.

  • The default alpha value for sigma points transformation is set to 0.9 to improve the overall stability.

     

Fixed:

  • The License Manager crashes on Linux when pressing the "prepare email" button.

  • Data Fusion Designer: Model runtime errors are not shown.

  • Data Fusion Designer: Invalid model errors are shown.

  • Data Fusion Designer: Misleading error message the configuration contains no sensor.

  • Data Fusion Designer: Multiple dml files lead to compiler error.

  • Data Fusion Designer: The Data Fusion Designer cannot be opened if the Baselabs.Statistics nuget package has not been restored.

  • Incorrect radar track proposer documentation.

  • Data Fusion Designer: Detection models with custom spaces cannot be displayed.

  • Data Fusion Designer: GUI blocks when a model contains virtual properties.

  • Data Fusion Designer: Obsolete model properties are used.

  • Data Fusion Designer: If a detection model uses a space that is derived from the measurement space, the field of view cannot be shown in the editor.

BASELABS CREATE 4.3

This software release includes the following product improvements:

Improved:

  • Improved: A generic radar measurement model has been added which can be extended by using custom measurement and system spaces.
  • Improved: A generic radar track proposer has been added that serves as base class for custom track proposers.
  • Improved: The code generated by the data fusion designer can be transformed by BASELABS Code. Requires code regeneration.

Fixed issues:

  • Fixed: If a space is defined over an angular space and the angle of this space is set close to +/-Pi, the SigmaPoints.ToGaussian() method may calculate incorrect results.
  • Fixed: The CTRV model may lead to unexpected results when used in the Extended Kalman filter (EKF). Fixed: The radar measurement model cannot be used in the EKF in scenarios where the x-position of a track equals the x-position of the radar.
  • Fixed: The GetTracks() method of the generated data fusion class does not support retrodiction and throws an exception if called after an immediate preceding call to 'ProcessVelocity()' or 'ProcessYawRate()'. Requires code regeneration.
  • Fixed: Smart sensor track proposers determine incorrect velocity and heading values for large yaw rates and large track x/y positions respectively.
  • Fixed: The smart sensor measurement model uses an incorrect measurement noise. Requires code regeneration.

BASELABS CREATE 4.2

  • Fixed: The Data Fusion Designer cannot be opened in Visual Studio 2013.

BASELABS CREATE 4.1

  • Fixed: Compiler error if project path contains symbols such as "#"
  • Fixed: Stability and consistency issues in the user interface of the data fusion designer
  • Fixed: Code generated by data fusion designer contains error when namespace and file name are identical
  • Fixed: Error during code generation in data fusion designer
  • Fixed: Initialization code is not generated w/o vADASdeveloper reference
  • Fixed: Save dml file does not work
  • Fixed: Settings of a generated vADASdeveloper component are not grouped
  • Fixed: Items are missing in the list of linkable settings in the data fusion designer
  • Fixed: Generated code mixes spaces and tabs
  • Fixed: Generated code does not check for negative time difference
  • Fixed: Reference to System.Drawing is missing
  • Fixed: Message boxes of the data fusion designer are not visible
  • Fixed: Inconsistent project saving state after code generation
  • Fixed: Visual Studio crashes if a BASELABS Create component is added in the Connection Designer of vADASdeveloper 2.5
  • Fixed: Invalid code generated for 'Smart sensor detection model'
  • New: The data fusion designer and the item templates can be used in Visual Studio 2012 and 2013

BASELABS CREATE 4

This is a major release of BASELABS Create which contains the following new features and fixed issues:

New and improved:

  • Configurable data fusion: Data fusion systems can now be configured using a rich user interface that is integrated in Visual Studio.
  • Camera measurement and detection models
  • Radar detection model
  • Smart sensor models
  • Track management
  • Constant acceleration model
  • Visual Studio item templates for creating new measurement models, detection models, track proposers, track removers and system models.
  • Improved accuracy of radar measurement model

Fixed issues:

  • Fixed: Biased calculation of the covariance in the SampleSet.ToGaussian() method
  • Fixed: BayesFilter.Predict() returns wrong results
  • Fixed: Race condition in licensing
  • Fixed: PersitenceModel.Evaluate() returns negative values for retrodiction
  • Fixed: Call for Evaluate() on system models leads to numeric exception
  • Fixed: License Exception in vADASdeveloper connection designer when GhostDoc is installed
  • Fixed: Usage of BASELABS Create with folder path including symbols such as "#"

BASELABS CREATE 3.5

This is a bug fix release of BASELABS Create which contains the following improvements:

  • Fixed: If a space is defined over an angular space and the angle of this space is set close to +/-Pi, the SigmaPoints.ToGaussian() method may calculate incorrect results.
  • Fixed: The CTRV model may lead to unexpected results when used in the Extended Kalman filter (EKF).
  • Fixed: The radar measurement model cannot be used in the EKF in scenarios where the x-position of a track equals the x-position of the radar.

BASELABS CREATE 3.3

This software release includes the following product improvements:

  • Linux development support. Developers can now develop on Linux, e.g. Ubuntu with MonoDevelop.
  • Fixed a bug in the vADASdeveloper connection designer when GhostDoc is installed which caused a license exception.

BASELABS CREATE 3.2

This is a bug fix release of BASELABS Create which contains the following improvements:

  • Fixed a bug that could cause the BayesFilter.Predict() method to return wrong results.
  • Fixed a bug that could cause a biased calculation of the covariance in the SampleSet.ToGaussian() method.
  • Fixed a bug that could cause the PersistenceModel.Evaluate() method to return negative likelihood values in retrodiction scenarios.
  • Fixed a bug that could cause licensing issues on Linux.

BASELABS CREATE 3.1

This is a bug fix release of BASELABS Create which contains the following improvements:

  • Fixed a bug in the example code of the 'BASELABS Create Data Fusion Application' template project which is part of the Visual Studio integration of BASELABS Create.
  • Fixed a bug in the Visual Studio extension which caused Visual Studio to crash if multiple versions of BASELABS Create are used.
  • Fixed a bug in the 'Baselabs.Statistics' NuGet package which could lead to inconsistent assembly references after an update of BASELABS Create.
  • Removed the contract assemblies from the 'Baselabs.Statistics' NuGet package.
Top of page
Contact us!
Close

Contact us!

Get in touch directly
(+49 (0)371 3371 51 51) or leave your contact data below.

Submit
* Required fields
This website uses cookies to ensure you get the best experience on our website. By visiting www.baselabs.de you agree to the use of cookies.