ANSYS SCADE Vision

ANSYS SCADE Vision

Reduce costs of automated driving system embedded perception software testing and safety

ANSYS solutions address all key elements of autonomous vehicles (AV): sensors, human-machine interfaces, automated driving system software, control software, computing platform and vehicle platform. ANSYS SCADE Vision completes the ANSYS solutions for AV to reduce drastically the costs of testing and safety activities for AI-based AV embedded perception software.

ANSYS SCADE Vision - AI-based AV embedded perception software

Autonomous Vehicle Sensor Systems Need to Operate Safely in Complex Environments
AV systems need to operate safely in more and more complex environments, incurring a dramatic increase in the cost of safety validation. The future of the AV industry may depend on which grows faster: market size or validation costs. The industry needs a solution to break the “drive-find-fix” barrier for automated driving systems. From a safety perspective, complexity is about how many “edge cases” AV perception systems might encounter (an edge case is a situation that can result in a failure and that has not been foreseen in the design phase).

Get Safer Perception, Accelerate Time to Value
SCADE Vision speeds up the discovery of weaknesses in your AV embedded perception software that may be tied to edge cases, and helps identify their root causes – also called triggering events – by automatically applying augmentations to your input sensor data to identify fragility in your AI-based embedded perception software:

  • While safety drivers manually record surprising behavior and direct autonomy developers to debug edge cases found, SCADE Vision solution finds fragilities in the AI-based autonomous vehicle perception software automatically; no manual oversight is required.
  • While finding weaknesses in autonomous driving system software requires massive amounts of expensive labeled data, SCADE Vision does not require labeled data; instead, it searches through raw sensor data recorded by your autonomous systems.
  • While new versions of autonomous vehicle perception software must be “test driven” on real roads, risking accidents, SCADE Vision assesses the robustness of perception algorithms using software-in-the-loop testing before your software ever hits the road.

SCADE Vision drastically reduces costs of AV embedded perception software testing and safety, making every mile count, while deriving maximum value from your AV data. This enables you to:

  • Get safer perception: Safer perception is just that — helping to identify the fragility in existing AI-based perception systems. SCADE Vision complements traditional simulation and road testing of perception systems.
  • Accelerate time to value: Find the most interesting data more quickly and cost-effectively compared to manual analysis. SCADE Vision can turn petabytes of data into megabytes of actionable scenarios. Early adopters report up to 30X acceleration to uncover edge cases in embedded perception software compared to manual or semi-automated approaches.
  • Maximize engineering resources: Keep your software engineers performing software engineering functions. Automate and allow them to review/tag data to locate the edge cases.
  • Turbocharge simulation: Simulation helps drive the maturity of the AV system, but it can only simulate what is known.
  • Maximize the value in the most expensive testing: It is expensive to collect road data. It is even more expensive to label it all. Find the most relevant and exciting data to focus the labeling efforts.

Capabilities

Autonomous Vehicle Perception Software Robustness Testing

SCADE Vision automatically identifies potential vulnerabilities in AV perception systems. It helps you find and categorize edge cases where these systems might exhibit unsafe behavior.

Autonomous Vehicle Perception Software Robustness Testing
Triggering events identification in autonomous vehicle perception software

This web-based UI helps to categorize probable defects of the AV perception software into triggering events, or root causes, for fragilities in the AV perception systems.

Triggering events identification in autonomous vehicle perception software
Automatic safety report generation

SCADE Vision automatically produces web and printable reports for perception algorithm development and safety teams.

Automatic safety report generation