What's really annoying to me is that we've been sold computer vision as the current solution, when it's obvious that if it is the eventual solution it's a long time out because our understanding of the field just isn't nearly as strong as it needs to be, but it's edged out a lot of interim solutions that could have been a lot safer, even if they have downsides.
For example, we could have embedded special markers (or spikes with RFID, or more active wireless, or any number of things that could have provided far more accurate lane detection as long as we were willing to require some up-front work to deploy it for special routes. Combined with very reliable lane detection and restricted to specific deployed areas where it could be tested, computer vision and/or radar/lidar for vehicle and large object detection (which would be mostly sufficient for most highway/freeway use) could likely provide a very safe system. The lower requirements for achieving safety might mean we could actually get some busses outfitted as well.
But that would require some actual state action, as no private company would (or should, if they were to keep it proprietary at least) deploy along large stretches of highway freeway. Covering I-5 from Northern California to Southern California would provide many opportunities, but be an enormous cost.
For example, we could have embedded special markers (or spikes with RFID, or more active wireless, or any number of things that could have provided far more accurate lane detection as long as we were willing to require some up-front work to deploy it for special routes. Combined with very reliable lane detection and restricted to specific deployed areas where it could be tested, computer vision and/or radar/lidar for vehicle and large object detection (which would be mostly sufficient for most highway/freeway use) could likely provide a very safe system. The lower requirements for achieving safety might mean we could actually get some busses outfitted as well.
But that would require some actual state action, as no private company would (or should, if they were to keep it proprietary at least) deploy along large stretches of highway freeway. Covering I-5 from Northern California to Southern California would provide many opportunities, but be an enormous cost.