It may be worth keeping an eye on Tactile Mobility. They work with companies including Ford, Porsche and BWM to take their existing sensor data (such as steering angle, speed etc.) and apply machine learning to infer the tactile response of the car. This can be used for things like detecting the severity of potholes. The company hopes to in future install this software in millions of cars as they come off the production line. If the software is run inside the vehicles, then the machine learning must be done inside the vehicle. This sounds like a good task for an ultra low power Akida processor.

https://www.techrepublic.com/article/tactile-mobility-uses-data-from-tires-and-cars-to-create-virtual-sensors-and-track-road-conditions/

Tactile Mobility uses data from tires and cars to create virtual sensors and track road conditions

Virtual sensing tech company worked with Detroit to find and rate potholes to help the city prioritize road maintenance.

The Tactile Mobility platform uses data from existing sensors on cars to create virtual sensors that measure road conditions and improve car safety and performance.

Image: Tactile Mobility

Tactile Mobility added $27 million to its bank account to expand its virtual sensing technology designed to improve car performance and safety. The company announced Wednesday that the new money is part of a Series C funding round led by Delek Motors, with strategic investment from Goodyear Ventures and Porsche Ventures.

The company said in a press release that the funds will be used to expand into new segments among vehicle manufacturers, tier 1 automotive suppliers, insurers and tire manufacturers and to build more virtual sensors and expand the company's cloud platform.

Tactile works with road authorities and municipalities in the U.K., Germany, Israel, Singapore and U.S. The company also partners with car manufacturers including BMW to provide their next generation vehicles with the sense of "touch."

Abhijit Ganguly, managing director, Goodyear Ventures, said in a press release that tire data is key to improved safety and efficiency of connected and autonomous driving.

"Tactile Mobility is at the forefront solving this need with their unique, software-only solution and we look forward to working together to bring this technology to vehicles and help to create safer and more efficient travel."

Yagil Tzur, vice president of products at Tactile Mobility, said the company's technology allows a car to understand the surface of the road, including factors such as impact and changes to the surface such as light rain or heavy rain.

"We take signals from existing sensors in the vehicle such as steering angle, wheel speed, wheel angle and engine torque and run it through a machine learning algorithm to create virtual sensors," he said.

The virtual sensors can sense tire health, estimate the grip between the tire and the road and predict the risk of hydroplaning. Tzur said the virtual sensors generate two data layers:

1. Surface DNA: This is a digital twin of road segments and can understand potholes and cracks as well as how the surface will react to weather conditions such as snow and rain
2. Vehicle DNA: This is a digital twin of the vehicle chassis systems, including tires and steering

Understanding and analyzing this data increases the safety of the car as well as the performance and comfort of a vehicle, Tzur said.

BMW uses Tactile's technology in its cars and the company recently announced an investment from Porsche as well. BMW cars use the data from Tactile's virtual sensors to shorten braking distance and tune the car's active suspension system.

Tactile's vision for the future is using data from these virtual sensors in millions of cars to monitor road conditions and alert drivers to hazards.

"Our vision is that the software comes with the vehicle and is installed in the production line as a standard part of the vehicle," he said. "At that point, you'll have millions of cars hitting the road globally and generating data at a high frequency."

Building privacy protections into data collection

The company has locations in the U.S., Germany and Israel and has about 50 employees, mostly engineers, software developers and researchers. This international focus may be part of the reason the company has factored GDPR concerns into the data collection process.

Tzur said the company classifies the data collection into two categories: information that must be GDPR compliant and and other data that does not require consent to be collected.

"We do not disclose anything about you as a driver or as a owner and the data is uploaded without any vehicle ID," he said. "Also we look at data in chunks of 15 kilometers, so you won't be able to see a long drive of 200 miles and be able to associate that with an individual."

Tactile sensors also don't record the first minutes or the last minutes of a trip or the vehicle speed.

Tzur said future types of data may need consent to be collected.

"A lot of this data is relevant to fleets where consent will not be an issue," he said.

Modernizing the monitoring of road conditions

Tactile also has worked with cities to use this data to monitor road conditions in real-time. One project included the city of Detroit and Ford. The city added Tactile's technology to Ford vehicles to create maps of city roads and rate the health of the pavement.

The city used this detailed and up-to-date data to optimize maintenance budgets and understand which areas need the most attention, Tzur said.

Tactile also has worked with cities to grade potholes. The same vehicles used to monitor road conditions mapped the potholes and created a priority list for repairs. Tactile compared its pothole map to one created by the state road department using traditional survey methods.

"This showed a high correlation between the maps and proved the tech was working as expected and could replace outdated tech that there is today," he said.

The company is working on a dashboard for city street departments for real-time monitoring of road conditions.

"For larger cities, our API will allow them to take the data into a GIS system and display it as a data layer," he said.

Tzur said that the company's data analysis goes through a carmaker's backend servers and does not interact directly with the car.