Idea: technology to improve road safety

[Veloman]

I'm sure the self driving vehicles already have the tech to accomplish this, but it should be packaged for human piloted vehicles as well.

Sensors to detect when the vehicle is on path for hitting a vulnerable road user such as a cyclist, pedestrian or motorcycle (could apply for other vehicles too). The system will then engage the vehicle's horn (and maybe brakes) to alert the pending victim that this vehicle is being piloted towards them. We have too many incompetent drivers, even commercial drivers such as city bus drivers.

On Monday at UT Austin we had a bus driver murder a cyclist, after being shouted at by passengers to stop, who saw the whole thing happen.

I believe this tech could be effective without causing many false alarms. With a combination of location, speed, and trajectory data for both the vehicle and the pending victim, this would save lives.

 

Don't like ads? Remove them while supporting the forum: Subscribe to Fastlane Insiders.

[ApparentHorizon]

Sounds like a patentable idea, and you'll have a couple of decades to sell as many as you can before self driving is the norm.

It might even be one of those things mandatory in new cars. Installed on the front bumper.

Or maybe something plugged into the 12v or USB and stick it on your windshield.

Better get to work

 

[TKDTyler]

Unfortunately, manufacturers will not take on the cost of a Lidar+Radar system for human detection due to cost. Collision warning is feasible due to the amount of sensors and types required to achieve that solution, but human detection is much harder and requires many more sensors for the field of vision required with high accuracy.

Different sensors are stronger for different applications and must be built and optimized accordingly for that application. This has many factors such as resolution, field of view, distance - all of which can never account or help with human reaction time. Add on to the fact that there inevitably is constant obstruction from pedestrian view. I don't believe any company would take on the liability of death if their product didn't work as advertised, especially with the R&D + legal costs associated with automation.

What you are talking about would be classified as a level 3 autonomous vehicle (there are 5 classes in all, level 5 being fully autonomous where the vehicle performs on par or better to a human in any situation presented including the extremes).

In an ideal world, it would be great to have in any human operated vehicle.

 

[ApparentHorizon]

Unfortunately, manufacturers will not take on the cost of a Lidar+Radar system for human detection due to cost. Collision warning is feasible due to the amount of sensors and types required to achieve that solution, but human detection is much harder and requires many more sensors for the field of vision required with high accuracy.

Different sensors are stronger for different applications and must be built and optimized accordingly for that application. This has many factors such as resolution, field of view, distance - all of which can never account or help with human reaction time. Add on to the fact that there inevitably is constant obstruction from pedestrian view. I don't believe any company would take on the liability of death if their product didn't work as advertised, especially with the R&D + legal costs associated with automation.

What you are talking about would be classified as a level 3 autonomous vehicle (there are 5 classes in all, level 5 being fully autonomous where the vehicle performs on par or better to a human in any situation presented including the extremes).

In an ideal world, it would be great to have in any human operated vehicle.

They'll have to do it if they ever want to compete in the self driving market.

And you can make a system like that in your own garage for under 1k. If you know how to program.

You start out selling them to bus operators and gov't orgs. Maybe to mechanics and dealerships so they can resell.

Then luxury cars. And once at scale, reduce costs for all.

This guy plans on making plug and play autonomous devices for cars, for under $1k. If true, then OPs idea is more than feasible.

"We ship a camera and use the sensors that are already built into the car"

View: https://www.youtube.com/watch?v=aqdYbwY9vPU


His plans are open source and wants people to take it and develop their own products.

comma.ai

 

Don't like ads? Remove them while supporting the forum: Subscribe to Fastlane Insiders.

[Deleted50669]

Very solid idea, and I doubt many consumers would disagree.

As others have said, would need a way to motivate manufacturers to eat the cost.

An alternative would be to develop the patent and sell the patent. Let a company with lobbiests / deep pockets worry about penetrating the auto market.

 

[Guest61362]

Another way to begin - start with having it on vehicles on a set and predictable path like rails- trams, trains. Railway drivers (operators?) tend to loose attention due to not steering that much if any, perhaps?

• Rails stay on the map until reconstruction.
• Whole route can be scanned from A to Z and back.
• Probably implemented already. Either monitoring driver's awareness or scanning the route or both.

 

[Deleted50669]

Another way to begin - start with having it on vehicles on a set and predictable path like rails- trams, trains. Railway drivers (operators?) tend to loose attention due to not steering that much if any, perhaps?

• Rails stay on the map until reconstruction.
• Whole route can be scanned from A to Z and back.
• Probably implemented already. Either monitoring driver's awareness or scanning the route or both.

I know in the case of DC's metro system they have highly sensitive sensors and collision projections that emit a loud beeping sound to the operator if they need to slow down - even if it's well-before contact would be made. That being said, the DC metro is on fire half the time, so there's room for improvement in other areas

 

Don't like ads? Remove them while supporting the forum: Subscribe to Fastlane Insiders.