| HN Mirror

Here's an odd, tiny, somewhat controversial/dangerous-sounding yet possibly interesting idea I thought of a little while ago that you might like to play with: a road-crossing simulator/trainer (and related concept areas).

My house fronts onto a small but fairly active 4-lane regional/suburban highway which I need to cross whenever I get the bus home, and also sometimes when I leave depending on which direction I'm headed. There are complete traffic breaks every 1-5 minutes or so, and it never gets jammed (there are no traffic lights nearby and it's a long stretch of road), so for a highway it's reasonably tame. My main goal is always trying to take advantage of the "near-breaks" that sometimes happen where the road almost completely clears and I can cross if I'm willing to dodge traffic. I especially try to do this when there's a bus approaching the stop across the road!

I've slowly gained confidence and experience over the past 13 years I've lived where I do (I'm 26 now, FWIW), and I now know when, how and why I can safely begin to cross even when cars are on the still road, so I often don't have to wait for complete breaks. That's been a fairly recent development; my progress hasn't been instantaneous.

I'm at the point where I'm trying to improve my ability to break the road down into lanes and actively track the activity in all the lanes simultaneously, so I can properly "leap-frog" across the road even more quickly. I am (perhaps understandably) not very good at this bit at all: I've found that taking (opposing!) traffic motion across multiple lanes and turning that a precise, realtime and confident/low-doubt go/no-go actually requires a fair bit of neurological development. Problem is, road-crossing has no/few common analogues from other life-skills situations that relate to spatial awareness, gross motor coordination, etc, so it's hard to create and iteratively improve this ability.

The main two reasons for this, I think, is that a) road-crossing is potentially life-threatening, so you want to get it right, and b) (important bit) we all seem to be taught to treat crossing roads as almost as dangerous as jumping out of planes - it's something nerve-wracking that must be done as quickly as possible before any damage (which could happen at any moment) is done. I'm guessing this ideology gets rooted in our heads due to our parents' overarching instincts to protect us from harm at all costs, juxtapositioned with the fact that 99.9% of the population does not have a sound understanding of psychology and an idea of the impact of different presentational styles. (In my own case I was simply taught to be extremely careful, but I only had experience with high-traffic roads after 13, and I had a general fear of roads before that point as I didn't need to cross that many, and when I did I was never alone.)

I think that if we can bootstrap ourselves to the point where we can eliminate the FUD and "helpless prey"/deer-in-headlights mentality surrounding crossing roads, we can begin to actually develop mental models that will likely serve us equally well in many different kinds of split-second situations that involve precise timing.

VR would be a way to get to that point: by creating a virtual environment full of various different types of vehicles and environments and simulating those vehicles bearing down on us (using a highly physically accurate 3D engine), we could actually learn through infinite repetition what 60 miles an hour looks like starting half a mile away, or what 20 miles an hour looks like starting a quarter of a mile away, etc etc. And we could slowly get to the point where we can confidently say things like "I know that I'll just make it across this road before that car does if it doesn't change speed" with much greater accuracy than we currently can. Some users may even begin to accurately guess vehicle speed just by watching the vehicle for a few seconds. It would be kind of fun and awesome to make a VR system where kids can be exposed to these kinds of experiences from a young age as an almost standard thing.

Besides a projector system which wouldn't be nearly as realistic, the only alternative to VR I can think of is repeatedly crossing an actual road all day. That would theoretically work, but there are four risk factors: a) is obvious, the fact that each crossing carries discrete risk; b) the fact that exhaustion from running back and forth would raise the stakes of (a); c) the fact that I'd be trying to be adventurous for the sake of learning which would make things worse; and d) the fact that as I gained experience and skill my risk of complacency would go through the roof due to repeated success.

Point (d) is valid for a simulation, too, but could be combated by constantly mixing up the environments - plain road; road with sharp bends; road with car speeding at 60 miles an hour around sharp turn or behind hill; etc - and maybe weird things like only allowing you to end the game when you failed, etc.

The huge controversy with this (there is a catch) is that young minds would latch onto this new kind of information instantly and turn kids into absolute ninjas capable of crossing complex roads routinely leaving just inches to spare. I see the average retiree driver heart attack rate going through the roof, to say the very least.

Because of this, I sadly don't see a school curriculum supporting something like this, and trying to make a company out of it would quite likely fail too because of the constant stream of negative press it would inevitably attract.

All the ingredients are there - you can repeat as much as you want with no cost, there's the element of competition and winning, and there's nothing stopping you from being adventurous and moving at the absolute last minute. Of course kids (full of energy, no idea what to do with it) are going to game that to the hilt to impress their friends. I have doubts that a game engine would be able to competently prevent that - I'm thinking of a "minimum winning crossing distance" metric, but I'm not sure if that would cover everything.

My crazy argument is to let it happen anyway: _let them_ scrape through the levels with inches to spare - because it might mean someone can save a life one day because they have the confidence to know they'll be able to do it in time. I've seen crazy internet videos of things like people dashing onto train tracks to rescue others at the last moment, and I'm not sure if I'd be able to manage that quickly enough because I'm missing precisely the information I describe here. (These are the related concept areas I mentioned at the start of this post.)

I think something like this would likely be best done as an open source project, in a framework where artists and modelers can easily collaborate and feed back art assets for new environments. The whole thing would need to stand on its own to gain traction, I think.

This is definitely not the kind of thing that looks awesome on paper, although I can see it being a lot of fun to work on, and something where you know you'd be teaching some really cool and liberating skills.

FWIW, I have absolutely no hope of getting my hands on any VR hardware anytime soon - due to circumstances entirely outside my control I've been stuck on hand-me-down PCs that average 10 years old for the past 2 decades - so I just thought I'd share it in case you (or anyone else) wants to play with it.