ADAS and Autonomous Vehicles need to operate safely in all conditions, and it is crucial for “all conditions” to include the most difficult scenarios such as darkness and poor weather, i.e. rain, snow, and fog. Unfortunately, current testing guidelines from NHTSA and EuroNCAP evaluate vehicle safety in good conditions only and many reports show a lack of robustness in the harsh scenarios. Algolux will review the challenges of robust computer vision, describe advanced approaches to improve computer vision accuracy under all scenarios and show direct comparisons and benchmarks against open and commercial perception solutions.