Archivio di tutti i clip:
(Notebook di Evernote).
While it may be a long time before self-driving cars become common, it’s not too soon to anticipate the need to meet them at least half-way as smarter pedestrians.
By Ross Stapleton-Gray
Sep 27, 2017—
The world of smart things is coming in bits and pieces, and at varying paces. Putting “smarts” in a thermostat is a pretty obvious feat; that we’ve seen as much progress as we have in getting fully autonomous vehicles on the road is pretty astonishing. But smarts have tended to be in pockets, and particular applications are being fielded even as the landscape is largely still mostly “dumb.”
At some point, however, we’re going to start seeing tipping points, at which enough smarts in the objects and the infrastructure will make those unsmart things that remain more the anomaly than the norm. I wonder if it isn’t too soon to be thinking about a need to enroll people in the so-called Internet of Things, to put people on a par with the toasters and the Teslas.
Autonomy in vehicles now mostly means non-cooperatively sensing surroundings, using radar, LiDAR and machine vision. Standards setting and systems for car-car and car-curb communication have lagged the deployment of systems that can cope with an un-augmented environment. Autonomous vehicles will, over time, communicate more cooperatively—that is, they will alert other cars (and be alerted by them) to better coordinate as they merge, pass or turn in front of each other.
The fatal Tesla accident earlier this year—in which a Tesla failed to discern a white truck broadside against a light sky, killing its driver (who ought to have been watching the road as well)—could likely have been avoided if the Tesla had a richer means to interact with its environment than just visual cameras. In other words, the outcome could have been different had the road been alerting it to other vehicles, or had the truck been reporting its own location and speed. State transportation authorities are exploring ways to make roads and obstacles more easily seen, such as widening the lane striping or finding ways to make it more visible in inclement weather.
In addition, cars could be designed to interpret what they see when they encounter humans, beyond detecting and recognizing them as such. Some engineer out there is probably, right this moment, designing those “middle finger detection” models to go from “I see a human” to “I see a human who’s rather upset by what we just did.” More practical, perhaps, would be the detection of “No, you go ahead,” and “Hello, I’m trying to merge here” gestures from human drivers.
Better and safer, though, would be allowing for the same digital communication (no, not that middle-finger digit, but ones and zeroes) that cars will be using between themselves and other vehicles—and the roadway, and other objects—to be accessible to humans, both to communicate and anticipate intent.
Most people carry communication devices—their smartphones—that are increasingly being used for applications other than making merely phone calls. They are being used as beacons, authentication tokens, access passes and digital wallets. And one could imagine a “smart pedestrian” app to negotiate with all the smart vehicles that will be entering the landscape—to extrapolate the owner’s trajectory and intent, using an accelerometer, the history of prior activity, a contact address list, and so forth.
Longer-term, it won’t be an app on a phone, but rather “how I relate to the world of smart things” (cars and everything else). It won’t necessarily be a device, but rather a set of protocols that humans will use for interfacing with that world, whatever the digital interconnect it is they’re wearing, carrying or implanted with.
The U.S. House of Representatives just passed a bill that would bar the states from imposing restrictions on self-driving cars, so the pace of their introduction won’t be hobbled by states needing to come up to speed, so to speak. While there will still be some time (perhaps a long time) before self-driving cars become commonplace, let alone the norm, it’s probably not too soon to anticipate the need to meet them at least half-way as smarter pedestrians.
Ross Stapleton-Gray, Ph.D., is the president of Stapleton-Gray & Associates, a San Francisco Bay-area technology and policy consultancy. He is a former CIA intelligence analyst and technology entrepreneur with interests in cybersecurity, the Internet of Things, surveillance and privacy.