The inconvenient truth about pedestrians is that their flexibility makes them easy to ignore in street design. We have to put thought into vehicles, parking, bikes... but because it's so easy to ask pedestrians to go out of their way, consideration for them has traditionally come last, if at all.
As a result, we get city streets that aren't walkable (a vicious cycle leading to decreased economic viability and variety), and a rift between pedestrians and traffic law (also a vicious cycle, where pedestrian citations often lead to suspended licenses and, subsequently, more severe issues.)
While there are many champions for pedestrian rights, the cause is frequently thwarted by an inability to prove there is enough demand (and political support) to justify street and intersection redesigns that appear to detract from incumbent demand (vehicle throughput.)
What's being proposed here is the development of a "crosswalk projector" device which feeds directly on pedestrian demand. The idea -- and it is just an idea, to be sure -- is that an overhead streetlamp projector (or other lighting device) could dynamically create crosswalks of all shapes and sizes, at corners and mid-block alike, based on a prioritization algorithm taking into account the current vehicle throughput vs. current pedestrian demand (i.e., number of people waiting to cross in a given area.)
Optimally, whatever pedestrian sensors are utilized (computer vision in-device, or a companion sidewalk pressure pad/visual sensor, etc.) would recognize the intended direction of each pedestrian, and once demand reaches critical mass -- be it one person trying to cross the middle of a block with no cars approaching, or 200 people crossing while ten cars approach -- the overhead lighting device would light up a temporary crosswalk pattern encompassing all the pedestrians' desired paths, or regressing closer to the mean if some pedestrian intentions are too far outside the bounds of the core demand. Once demand subsides, the crosswalk disappears, and vehicles start moving again.
In a more connected city, this dynamic drawing of crosswalks would be useful as a method of blocking off entire blocks during lunchtime, or creating unorthodox scrambles without having to permanently paint what is already the most visually bewildering section of any given road.
Again, in a more connected future, V2X technology could easily coordinate with vehicles to ensure safe crossing. But for now, the proposal theorizes that dynamic crosswalks would indirectly make streets safer even when not in use, as drivers would now be responsible for paying better attention and driving at a speed reasonable enough to stop for pedestrians at any given moment. The lighting device could assist with recognition, for instance by lighting up the crosswalk in a simulated 3d pattern:
There are more peripheral safety benefits, such as the ability of a projected crosswalk to work in snowy conditions, and the mitigation of any need for peds to stand on the curb or in the street prior to crossing. The sensors could identify waiting peds at a safer distance from the curb.
Lastly, the concept of building pedestrian demand dynamically through volume or wait time would serve to educate drivers and peds alike about the complex relationship between choice and throughput. While this proposal seeks to improve pedestrian choice, it also aims to teach all road users the importance of prioritization -- making them better-informed to decide in each community just how much to dial up or down the pedestrian demand that triggers these crosswalks.
I've discussed the idea in further detail here, but again, I want to clarify that the funding would go towards building a prototype for one intersection. The technology is the biggest hurdle, and beyond proving it's technically feasible, observing it in action at one major intersection should provide enough commercial validation to expand it across (and beyond) the city.