aiData
Automated data collection
aiData Recorder is focusing on gaps and edge cases: reference sensor design, calibration toolkit, recording and data upload software solution.
Automated annotation
Multi sensor AI-based automatic annotation for dynamic and static objects with industry-leading precision
Synthetic data generation
High-fidelity sensor simulation enabling outputs indistinguishable from the real world, combined with domain randomization to ensure variability effortlessly
Integrated metrics evaluation
Tracks development progress against requirements and provides real-time insights and data gap analysis.
Data Versioning System
aiData Versioning System enables precise measurement of the effect of adding new data to fill gaps & tracks usefulness of collected data
Data-driven pipeline for AD
Proprietary data pipeline reduces the complexity of the processing with high automatization while still ensuring automotive quality
aiData Recorder
- Synchronized, smart data collection on scale
- Designed to be sensor agnostic
- Internal reference implementation on various vehicle models with various sensor setups
aiData Auto Annotator
- Automated edge-case mining and active learning to reduce storage and compute costs
- Automated annotation and knowledge transfer to reduce annotation costs
- Real-world to model-space scenario extraction
- Simultaneous annotation of all sensors with a consistent 4D (space + time) environment model
- 100% precision for static object annotation
- 90%+ precision for dynamic object annotation
aiFab
- Highest fidelity sensor data generation
- Automated generation of scenario variations based on parameter sweep, randomization or Monte Carlo methods
- Replicates variability in real-world data with domain randomization
aiData Metrics
- Comprehensive tool for evaluation of NN algorithms and detection SW
- Interfaceable with requirement management tools
- Built-in visualization for quick assessment
- Flexible output format for post-processing and integration
aiData Versioning System
- Reproducibility of the network structure, training data sets and training method that led to a specific neural network
- Understanding the strengths and gaps in the data
- Transparency of algorithm performance in different conditions
- Precise measurement of the effect of adding new data to fill gaps
- Track usefulness of collected data
