Driver assistance swerves for pedestrians but hits sheep: what went wrong?

Let me share something. Recently, I saw discussions about assisted driving systems dodging pedestrians but hitting sheep and road barriers. My first reaction was that marketing had distorted the boundaries.

These days, I'm trying out a breakdown table for smart driving promotions. I start by asking: when they say "tested," did they test for pedestrians, cars, stationary cars, construction zones, animals, or just easy-to-pass scenarios? Public discussions already include manufacturer explanations that AEB (Automatic Emergency Braking) doesn't necessarily respond to cones, water-filled barriers, stones, or animals; vision systems also rely on samples and perform poorly on irregular objects. This isn't surprising. What is surprising is that some rhetoric frames "recognizing pedestrians" as "handling everything on the road."

Those working on cockpits fear this mismatch the most. When drivers see the words "assisted driving," they easily interpret automotive-grade safety boundaries as automatic fallback.

When a sheep, cone, or construction sign suddenly appears, if the system doesn't brake and the driver isn't watching the road, risk emerges. So I don't entirely agree with just blaming the technology. What should change more is product expression. Capability boundaries must be clearly explained during purchase, OTA updates, and interaction prompts. Don't use pretty demos to hide shortcomings.

My suggestion is straightforward: in marketing, say less about "leading by a wide margin" and more about when you cannot drive. Cockpit prompts shouldn't clutter the screen; use one sentence to let the driver know they must take over immediately. That's it.