Hacker News new | ask | show | jobs
by delfinom 1038 days ago
>A system that signals the location of any actively flashing vehicles to all nearby automated cars is technically achievable. It should have been one of the first steps in the development of any of this technology.

Doesn't take much audio processing to detect blaring sirens.

3 comments

If the sirens are at your back, you don't want that robocar to stop and hamper the traffic but to choose a side on the road/street to either park or continue its course until there is room to be passed by the emergency vehicule.

Blaring sirens can be easy to detect, doesn't mean the reaction should always be the same.

[1] preferably same as other vehicules which are possibly blocking traffic for emergency vehicles.

Sirens are designed to have an easily discernible doppler effect; a siren coming at you sounds different from a siren going away, which sounds different than a siren keeping pace with you. And furthermore, a siren behind you sounds different from a siren in front of you.

Human drivers generally manage all of this fine. If computers can't, then they aren't ready to drive.

> Blaring sirens can be easy to detect, doesn't mean the reaction should always be the same.

It's ML, so I imagine a decision like this isn't hardcoded to begin with.

Directionality should also be easy to detect. Should just need a handful of mics around the car and some basic audio processing.
What I meant is that detection is one thing, taking good decisions is another.
May not cover this scenario, but lots of emergency vehicles don't use sirens at night. The thought is, the bright flashing lights is plenty - and if the road isn't busy, why wake all the neighborhoods up on a routine service call.
Curious, your thoughts on those that are deaf and driving?
The deaf use devices with a microphone that translates sirens into a visual signal. It is not burdensome to ask self-driving cars to have microphones.