Embedded software for runtime validation of ADAS/AD driving functions
The Trajectory Checker is solving the issue of real-time monitoring and runtime validation
Real-time monitoring & runtime validation
Is the system seeing properly?
Perception level
✓ No false negatives
✓ No false positives
Is the system behaving properly?
Planning level
✓ Follow intended behavior
✓ Minimise risk
Is the system functioning properly?
Functional level
✓ Hardware signals
✓ Sub-components signals
✓ Anomalies detection
What is it?
The Trajectory Checker is an embedded software component that checks and quantifies the risk of a planned trajectory. The checks are performed based on a prediction of the environment and conformance of the plan to the safety policies (e.g. OEM rulebook, IVEX formal model, RSS, etc.).
Applications
Autonomous driving L4 technology
The Trajectory Checker L4 supervises the main AD system, (1) performing a safety analysis of the planned trajectory, and (2) checking if the system is still in a sane operating state
ADAS (L2+/3) for commercial vehicles
The Trajectory Checker L2+/3 supervises the existing ADAS system, performing a safety analysis of the planned trajectory at runtime
ADAS (L1/2) for Busses & trucks
The Trajectory Checker L1/2 is supervising the driving, making sure he will not be involved in safety critical situations.
Uniqueness?
IVEX combines knowledge from experience and from formal methods into a model for safe driving. The experience comes from the Data Analytics Platform that gathers knowledge from all the most challenging and edge-case situations. The formal model is powered by the IVEX patented process to create safety models.
IVEX patented an innovative AI-powered process to create a explainable safety models. This process automatically transforms safety rules for motion into formally verified software. The process highlights limitations of the safety rules for the motion controller and formally guarantees that the motion rules are consistent
The Data Analytics Platform collected petabytes of safety critical and edge-case scenarios. Those situations are re-used to improve the safety model.
