aiData | Automated Data Pipeline for ADAS & AD

aiData™

Road-ready automotive AI starts with high-precision data

Transform raw recordings into fully annotated training and validation-ready datasets

Automated driving requires millions of kilometers of real world driving data for training, testing and validating AI-based Advanced Driver-Assistance Systems (ADAS) and Autonomous Driving (AD) features. Creating high-quality recordings, annotation and data curation is a very resource-intensive process and prone to errors due to the extensive manual work involved in the process. By minimizing manual effort and maximizing efficiency, aiData helps teams accelerate the development of ADAS/AD features by transforming raw recordings into fully annotated training and validation-ready datasets - fast, accurate, and effortlessly scalable.

Features & benefits

aiData is a streamlined, automated data pipeline designed to optimize the development of automated driving technology. By automating key stages in the data workflow – from data collection and preparation to curation, annotation, and evaluation – aiData ensures a smooth transition of data between data scientists and developers.

aiData Recorder

Upload fast, process immediately

• aiData Recorder ensures accurate calibration and precise synchronization - with real-time sensor diagnostics for live feedback to ensure quality and reduce data waste.
• All sensor recordings are automatically uploaded and registered and are available for processing within the hour.
• Automatically generated metadata for all recordings, enabling searchability.
• The aiRec framework can run the Software Under Development to flag relevant scenarios.

aiData Auto Annotator

Accurate and low-cost automatic annotation – faster than ever

• 3D bounding boxes for dynamic traffic participants – such as vehicles, pedestrians, cyclists, traffic signs, and traffic lights – and automatically labeling static elements like lanes and road markings.
• Superhuman performance with an industry-leading 95% precision and recall for all ODDs.
• Instant effect on the development budget - 95%+ cost reduction only after 500 hours of annotated data, compared to manual labeling.
• Return of initial investment within 30 hours of annotated data.
• Fast and scalable performance with a 400-time speed gain compared to manual annotation.

aiData Versioning System

Find the data that matters with AI-powered data curation

• Keep track of the whole data journey from recording through annotation, adding to a dataset, and validation of the resulting software.
• Improve model performance with automatically curated and growing datasets.
• AI-based searching, such as scenario similarity, text, image, or video.
• SQL-based queries for advanced queries and machine-to-machine use cases.
• Full traceability in the data lifecycle.

aiData Metrics

Evaluate to unlock peak AI performance

• Completely automatic V&V of new software releases produced overnight

• Evaluate algorithms and detection software using 150+ built-in metrics, such as calculating NCAP scores and metrics for regulatory features, in simulation, readback, and resimulation.

• Create custom metrics easily to track progress against AI, software module, or vehicle level KPIs

Are you ready to utilize the aiData pipeline or certain components to ease your data workflow?