3 ms·
I think there's a crucial issue with sensoring. Companies are tackling the problem mainly using sensors that are contained within the car. Even with LIDAR (and
by gota 6y ago
I think there's a crucial issue with sensoring. Companies are tackling the problem mainly using sensors that are contained within the car. Even with LIDAR (and not just cameras) it is very hard to "sense" the full scene[1].
What if our roads were built to communicate with the cars? What if there were only AVs on the street and they communicated with each other? What if every bicycle, school backpack, shopping cart did so too, identifying itself in the process?
I think the problem of self-driving is harder right now in the transitional period in which they have to coexist with human drivers and essentially trying to "emulate" how a human driver gathers information (not really, but you get my meaning)
I'm not sure it won't take decades, but I feel like there will be a magical few-months period in which we go from "can it be done?" to "oh, ok,we got it"
[1] shoot, I lost the reference. But Cruise apparently assumes "use whatever is available until it works, we'll make it cost-effective later"
- dmitrygr 6y ago> What if every bicycle, school backpack, shopping cart did so too, identifying itself in the process? That sounds like a nazi-state dream - easily track every backpack, car, bicycle, etc... no thank you
- nicoburns 6y agoThat's a really good idea. You probably wouldn't need too many sensors either. Now that I think about it, a few strategically placed cameras around blind conrners woul be pretty useful for human drivers too!
- stefan_ 6y agoThen we can just build trains instead. Perfectly automated today.
- thu2111 6y agoNo, the physics of that don't work at all. Trains for example cannot handle gradients that cars manage with ease.
- sudosysgen 6y agoNot true. Not every train uses steel wheels, you know.
- slfnflctd 6y agoOh definitely, I've been convinced sensors distributed in the environment are the only way forward for FSD with current tech since about 2017. Do a test run with a dedicated road/lane for a while, gather data, improve the system and then gradually roll it out elsewhere. It's clearer and clearer that self contained 'autopilot' units are going to keep failing in scenarios involving unpredictable people & situations. They don't have enough data and never will (without AGI, anyway). So we need to spoon feed them that data by tagging everything & everyone around them.