The video shared in a subsequent section of the post is the second instance this week where I came across a child walking on a busy road, one without a sidewalk and crazy traffic, and as you can see in the video, without even a decent shoulder area.

While the other cars did not stop, in both instances, I decided to halt, till they passed, to ensure they were safe.

Since I saw two of these instances on two consecutive days, I started pondering about it. My first thought obviously was around why they would have to walk on the road under such dangerous conditions (to me, allowing a child to do this is akin to child endangerment). Also, whenever I drive this route to pick up my wife from the train station, I see multiple cop cars on the streets and additionally, first responder vehicles like fire trucks. I’m not sure if they keep an eye out for such things. Ideally, they should. The safety of these kids really worried me.

The second aspect of my pondering was around how AI can help enhance pedestrian safety on busy city roads.

In cases where parents are not supervising their kids and kids are not making safe decisions, having some form of capabilities in automobiles (in addition to the vanilla sensors used today), becomes imperative.

There are currently two types of brake assist options available in modern, human-driven cars. One is where the onboard system detects panic braking and automatically deploys maximum brake boost. In some advanced models, sensors in the front may lead to the car braking even when the driver is distracted. But these sensors are primarily designed to avoid rear-ending a car in front. The gist is that both of these features use technologies that have been around for decades.

Cars these days have cameras out in the front as well. With technologies that have matured now, thanks to the obsession with autonomous vehicles (which will not be fully autonomous in the production version in our lifetime), the onboard computer can be used to detect a pedestrian and then perform two possible actions.

The vanilla one is an audio warning like “Pedestrian alert.” Based on the training, the onboard algorithm can flag that a pedestrian is within the vehicle’s driving form factor and in close proximity. The second scenario is where the car starts applying some soft braking as well as the warning to engage the driver further.

In the race to replace humans, like developing self-driving cars, we sometimes forget that the same technology can also help save human lives.