Tech
Why companies don’t share AV crash data, and how they could
Autonomous vehicles (AVs) have been tested as taxis for decades in San Francisco, Pittsburgh and around the world, and trucking companies have enormous incentives to adopt them.
But AV companies rarely share the crash- and safety-related data that is crucial to improving the safety of their vehicles—mostly because they have little incentive to do so.
Is AV safety data an auto company’s intellectual asset or a public good? It can be both—with a little tweaking, according to a team of Cornell researchers.
A new data-sharing roadmap
The team has created a roadmap outlining the barriers and opportunities to encourage AV companies to share the data to make AVs safer, from untangling public versus private data knowledge, to regulations to creating incentive programs.
“The core of AV market competition involves who has that crash data, because once you have that data, it’s much easier for you to train your AI to not make that error. The hope is to first make this data transparent and then use it for the public good, and not just profit,” said Hauke Sandhaus, M.S. ’24, a doctoral candidate at Cornell Tech and co-author of “My Precious Crash Data,” presented Oct. 16 at the ACM on Human-Computer Interaction.
His co-authors are Qian Yang, assistant professor at the Cornell Ann S. Bowers College of Computing and Information Science; Wendy Ju, associate professor of information science and design tech at Cornell Tech, the Cornell Ann S. Bowers College of Computing and Information Science and the Jacobs Technion-Cornell Institute; and Angel Hsing-Chi Hwang, a former postdoctoral associate at Cornell and now assistant professor of communication at the University of Southern California, Annenberg.
Barriers to sharing AV safety data
The team interviewed 12 AV company employees who work on safety in AV design and deployment, to understand how they currently manage and share safety data, the data sharing challenges and concerns they face, and their ideal data-sharing practices.
The interviews revealed the AV companies have a surprising diversity of approaches, Sandhaus said. “Everyone really has some niche, homegrown data set, and there’s really not a lot of shared knowledge between these companies,” he said. “I expected they would be much more commonality.”
The research team discovered two key barriers to sharing data—both underscoring a lack of incentives. First, crash and safety data includes information about the machine-learning models and infrastructure that the company uses to improve safety.
“Data sharing, even within a company, is political and fraught,” the team wrote in the paper. Second, the interviewees believed AV safety knowledge is private and brings their company a competitive edge.
“This perspective leads them to view safety knowledge embedded in data as a contested space rather than public knowledge for social good,” the team wrote.
And U.S. and European regulations are not helping. They require only information such as the month when the crash occurred, the manufacturer and whether there were injuries. That doesn’t capture the underlying unexpected factors that often cause accidents, such as a person suddenly running onto the street, drivers violating traffic rules, extreme weather conditions or lost cargo blocking the road.
Potential solutions for safer autonomous vehicles
To encourage more data-sharing, it’s crucial to untangle safety knowledge from proprietary data, the researchers said. For example, AV companies could share information about the accident, but not raw video footage that would reveal the company’s technical infrastructure.
Companies could also come up with “exam questions” that AVs would have to pass in order to take the road. “If you have pedestrians coming from one side and vehicles from the other side, then you can use that as a test case that other AVs also have to pass,” Sandhaus said.
Academic institutions could act as data intermediaries with which AV companies could leverage strategic collaborations. Independent research institutions and other civic organizations have set precedents working with industry partners’ public knowledge. “There are arrangements, collaboration, patterns for higher ed to contribute to this without necessarily making the entire data set public,” Qian said.
The team also proposes standardizing AV safety assessment via more effective government regulations. For example, a federal policymaking agency could create a virtual city as a testing ground, with busy traffic intersections and pedestrian-heavy roads that every AV algorithm would have to be able to navigate, she said.
Federal regulators could encourage car companies to contribute scenarios to the testing environment. “The AV companies might say, ‘I want to put my test cases there, because my car probably has passed those tests.’ That can be a mechanism for encouraging safer vehicle development,” Yang said. “Proposing policy changes always feels a little bit distant, but I do think there are near-future policy solutions in this space.”
More information:
Hauke Sandhaus et al, My Precious Crash Data: Barriers and Opportunities in Encouraging Autonomous Driving Companies to Share Safety-Critical Data, Proceedings of the ACM on Human-Computer Interaction (2025). DOI: 10.1145/3757493
Citation:
Why companies don’t share AV crash data, and how they could (2025, November 11)
retrieved 11 November 2025
from https://techxplore.com/news/2025-11-companies-dont-av.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
Tech
The Ricoh GR IV, the Cult Favorite Pocket Camera, Just Got Way Better
When I reviewed the GR III, I wrote about how much I liked snap focus mode, which allows you to set a predetermined focus distance regardless of the aperture. I set up my GR III to use autofocus when I half-pressed the shutter and snap when I quickly pressed, so that snap focus fired off the shot at my predetermined focus distance (usually 1.5 meters).
All that remains, but there is also now a dedicated letter, Sn, on the mode dial that sets the camera in Snap Focus mode, which allows you to dial in not only the distance you want focus at, but also the aperture you want to lock in. You can control the depth of field as well. I rather enjoyed this new mode and found myself shooting with it quite a bit.
Should You Get One?
The GR IV debuted at $1,497, which is significantly more than the GR III’s $999 price at launch. Is it worth the extra money? If you have a GR III and are frustrated by the autofocus, I think you will like the upgrade. It’s significant and, if you have the money, well worth it.
If you have any desire to use your pocket camera for video, this is not the one for you. See our guides to pocket cameras and the best travel cameras for some better, hybrid photo- and video-capable cameras. If you want an APS-C sensor that legitimately fits in your pocket, offers amazing one-handed control, and produces excellent images, the the Ricoh GR IV is for you.
Personally, I am holding out for the GR IVx, which will hopefully, like the GR IIIx, be the same camera with a 40mm-equivalent lens. At the time of writing, Ricoh would not comment on whether there will be a GR IVx.
Tech
Could You Use a Rowboat to Walk on the Seafloor Like Jack Sparrow?
But you already know about this, because Fg is what normies call an object’s “weight,” and for a given volume, weight depends only on the density. Now, if you dropped these blocks in a lake, obviously the styrofoam would float and the steel would sink. So clearly it has something to do with density.
What if you had a block of water with the same volume? If you could somehow hold this cube of water, it would feel pretty heavy, about 62.4 pounds. Now, if you place it carefully in a lake, will it sink or bob on the surface like styrofoam? Neither, right? It’s just going to sit there.
Since it doesn’t move up or down, the total force on the block of water must be zero. That means there has to be a force counteracting gravity by pushing up with equal strength. We call this buoyancy, and for any object, the buoyancy force is equal to the weight of the water it displaces.
So let’s think about this. The steel block displaces the same amount of water, so it has the same upward-pushing buoyancy force as the block of water. But because it’s denser and has more mass, down it goes.
In general, an object will sink if the gravitational force exceeds the buoyancy force, and it will float if the buoyancy force exceeds the gravitational force. Another way of saying that is, an object will sink if it’s denser than water and it will float if it’s less dense.
And right in the middle an object will neither sink nor rise to the surface—we call that neutral buoyancy. Humans are pretty close to neutral because our bodies are 60 percent water. That’s why you feel weightless underwater—the buoyancy force pretty much offsets the gravitational force.
Avast! Hold on there, matey. Aircraft carriers are made of steel and weigh 100,000 tons, so why do they float? Can you guess? It’s because of their shape. Unlike a block of steel, a ship’s hull is hollow and filled with air, so it has a large volume relative to its weight.
But what if you start filling it with cargo? The ship gets heavier, which means it must displace more water to reach that equilibrium point. In general, when you launch a boat or ship into the water, it’ll sink down until the weight of the water it pushes aside equals the boat’s total weight.
Tech
Sleep Number’s P6 Smart Bed Takes Customization to a New Level
Screenshots: Julia Forbes
I spoke with Raj Mills, Sleep Number’s senior vice President of partnerships and research. She tells me, “Our AI models take into account foam depth and still maintain the same level of accuracy regardless of how far below the surface of the bed the sensors are.” She shares that they are cohesively performing a ballistocardiograph, which monitors the blood flow generated by the heart and ultimately determines your heart rate score. How effectively they can do so is debatable.
Ultimately, I found there was quite a bit of variance in terms of the nightly score calculated on both ends. On good nights, both pointed to higher scores, but the final number could differ by up to 10 points. On the Sleep Number app, I found it concerning that most of the time, my Sleep Score numbers were not as high as I thought they would be—my average for the three-week test period was a 74.
Matrix Mattress
If you prefer a remote, that’s either a separate cost ($50) or potentially a different bed altogether. The only way to operate this mattress is by creating an account and downloading the app in advance. Security of one’s personal data is top of mind for many, and I wanted to know how the vast quantities of data accumulated by Sleep Number’s customer base were managed. When you first sign up for your account, you can either share your data with Sleep Number or opt out altogether. For those who share their data willingly—about 550,000 individuals, according to Mills—Sleep Number’s data science team performs research and presents findings from its consumer base at major sleep conferences, including SLEEP and World Sleep Congress.
It’s a comprehensive data set to work with, and it continues to evolve through the adaptive AI at play, which includes SleepIQ technology. According to Mills, the company’s AI models are structured to analyze sleepers at an individual level, because every sleeper has a different build, age, pain issues, and various needs. According to Sleep Number’s privacy policy, the company doesn’t sell your information to third parties, and you can withdraw your consent to share information at any time. At the end of the day, Sleep Number is a business that can potentially use your data to develop new products, and it’s up to you whether that’s worth it or not.
Power Couple
An adjustable base is what enables each side of the mattress to independently adjust head and foot angles. I tested the FlexFit 3 adjustable base alongside the P6 mattress, which is Sleep Number’s top-of-the-line offering. It offers timed foot warming (which takes about two minutes to heat up and has a two-hour default setting) to help blood flow away from your core and aid faster sleep. There is also a zero-gravity setting, partner snore (elevates the head and neck slightly to help open up airways), and motion-detect underbed lighting. I loved the gentle light source for late-night bathroom breaks, and the split king adjustability allowed me to partake in late-night reading without disturbing my partner. However, the only feature that separates the FlexFit 2 base from the FlexFit 3 is the inclusion of foot warming, and it’s a $400 upcharge for a queen size.
Photograph: Julia Forbes
-
Fashion1 week agoIndonesia’s thrift surge fuels waste and textile industry woes
-
Tech1 week agoT-Mobile Business Internet and Phone Deals
-
Business1 week agoBP names new boss as current CEO leaves after less than two years
-
Sports1 week agoPKF summons meeting after Pakistani player represents India in kabaddi tournament
-
Entertainment1 week agoIndia streamlines visa rules in boost for Chinese professionals
-
Sports1 week agoUWCL grades for all 18 teams: Leuven get A+; Barça an A-, PSG fail
-
Sports7 days ago
Alabama turned Oklahoma’s College Football Playoff dream into a nightmare
-
Entertainment1 week agoRadiation fears rise after cracks found in $2 billion Chernobyl shield

