fbpx
Traffic > Safety

Hyundai cars could soon tell the intent of pedestrians

This will obviously be good for improved road safety

How does a self-driving vehicle react to a chaotic road? PHOTO FROM PIXABAY

Most high-end cars these days are equipped with lots of driver-assist systems. Take the Tesla Model S with its autopilot system, for instance. It’s actually an intelligent piece of kit that allows the car to cruise on the freeway with the driver’s hands and feet off the steering wheel and the pedals. But how the system reacts to this environment is pretty much passive. Autonomous braking, for example, will only kick in if the sensors detect the obstruction. There is no way the car can actually anticipate anything coming its way. Hyundai wants to change this by getting cars to be actively aware of its surroundings.

Hyundai’s Center for Robotic-Augmented Design in Living Experiences (CRADLE) is the Korean carmaker’s innovation and advanced technology arm. In the interest of developing driver-assist systems and autonomous vehicles in the future, it has partnered with Perceptive Automata, a startup based in Somerville, Massachusetts. Perceptive Automata has created software that enables self-driving cars to predict a person’s state of mind, just like how you’d predict your girlfriend’s state of mind when you had to cancel your date at the last minute.

Soon, vehicles will be able to read foot travelers. Hopefully including criminal minds. INFOGRAPHIC FROM HYUNDAI

Perceptive Automata’s goal is to make driver-assist systems and autonomous vehicles better integrated in an environment that is always dominated by humans. Its approach to doing this is to have its software understand human behavior through available visual cues. The software does its wizardry by measuring a pedestrian’s intention and awareness. Intention is quantified by the pedestrian’s desire to act, while awareness is calculated from the pedestrian’s knowledge of his or her surroundings.

Perceptive Automata has created software that enables self-driving cars to predict a person’s state of mind, just like how you’d predict your girlfriend’s state of mind when you had to cancel your date at the last minute

These visual cues are often manifested by how pedestrians around the car behave. Signs of strong intent include running or brisk walking, while looking both ways usually signifies heightened situational awareness. The data from these observations become the basis for what the car should do next according to the software’s logic. It might start priming the brakes so that—just in case someone crosses the street without looking—maximum braking force may be instantly applied. In any case, the car and its driver are always prepared to respond to any situation, rather than acting only when the obstacle becomes a threat.

We think the future of this technology is quite bright. More and more automakers are working toward the development of solutions to autonomous mobility. And with Perceptive Automata’s pioneering research into advanced assist systems, we will probably see the day when human error is almost eliminated from driving.



Miggi Solidum

Miggi is an editor-at-large at VISOR. Professionally speaking, he is a software engineering dude who happens to like cars a lot. And as an automotive enthusiast, he wants a platform from which he can share his motoring thoughts with fellow petrolheads. He writes the 'G-Force' column.



Comments