| Topic: | Re:Reply | |
| Posted by: | Al Webber | |
| Date/Time: | 24/02/20 10:07:00 |
| Autonomous cars use GPS for navigation, but they have sensors (radar, lidar) to detect objects in their path. One of the big obstacles they are facing is the detection range at motorway speeds. Given that stopping distances are a LOT longer for trains, and that these systems can't see around corners in tunnels, these systems are a non-starter for the Underground. Driverless trains are coming eventually. Even now, on 4 Tube lines, the drivers just close the doors, and are essentially there to cope with an emergency. The sub-surface lines are in the process of being switched over. New trains on the Underground will be delivered with cabs, so they can run on the current signalling system, but I believe that there is a desire that they are designed so that the cab could be removed eventually to make more space for passengers. I think current plans would have a DLR style train captain on board. However, GPS and basic WiFi is not the answer. These systems work with dedicated transponders by the side of the track, so they are much more robust. It's a closed system for security reasons. There has to be absolutely zero chance of terrorists being able to hack it. It will not rely on publically available Wifi and GPS. Even then, the problems are not so much with the transponders and knowing where the train is, but with the software. The software has to be absolutely robust and failsafe. Autonomous cars won't be perfect, but if they reduce accident rates by 75%, they will be a huge success. Railways work to a much higher standard of safety. Accidents have to be zero. No-one will take any chances. It's more like the aviation industry used to be before Boeing started cutting corners. It's all that stuff about how the software handles it when a sensor goes down, or gives bad readings. That's not simple to do - and would need doing even with your "solution". It's the software that is delaying Crossrail - not the detection systems. It's the software that is the big issue with any system. Writing software from scratch for what you want to do would be an absolute non-starter - especially with the complications of integrating with ancient system elsewhere on the line. Writing that software would be a much bigger, and more expensive job than the software for the new system will be. At least that's based on technology and systems already in place elsewhere, and you don't have to write it from scratch. |