> Can you engineer sensors to do this? Absolutely. Machine vision is getting better every year, but these procedures date from a time when digital cameras and c

> Can you engineer sensors to do this? Absolutely. Machine vision is getting better every year, but these procedures date from a time when digital cameras and computers didn't even exist. They work well, don't require maintenance, and are simple. An engineered sensor-based solution would be expensive, would break often in the dirty environment of the tunnels, and would be pretty imperfect. The London Underground has figured this out, and figured it out decades ago. No tube driver gets out of their cab, or puts their head out of a window (they don’t have one that opens). Stations are equipped with cameras and screens that allow drivers to check doors. The doors are designed to detect obstructions, including items of clothing trapped in them. For Crossrail TfL has invested huge amounts of money into their new trains to further improve the door sensors, due to the extreme length of the trains. The doors have been tested for their ability to not just detect trapped items, but also the difference between bad (someone inside with a scarf trapped) from emergency (someone outside with a scarf trapped). [1] All of this is done to not just improve safety, but also capacity. In London at least, the biggest restriction to capacity is the amount of time trains need to spend at platforms for people to get and off. They’ve already optimised the hell out of transit period between stations. And all of this has been done on the oldest underground network in the world, and trust me it’s age shows. No Victorian ever imagined the tube would be so busy (or they wouldn’t have stuck us with tiny 6” high passenger compartments). [1] https://www.londonreconnections.com/2018/crossrail-cutting-fine/ https://www.londonreconnections.com/2018/crossrail-cutting-f...