Discover more about the topics and technologies to be discussed at this year's conference, via a series of exclusive interviews with a selection of our expert speakers

Speaker Interview: Chidambaram Subramanian, graduate research assistant at Virginia Tech and CenTiRe
Virginia Tech and CenTiRe

Chidambaram Subramanian, graduate research assistant at Virginia Tech and CenTiRe, talks about new research to integrate sensors into tires, thus creating ‘intelligent tires’. Because sensors such as those for wheel speed, light, rain and vision systems remain limited in their scope for sensing road conditions, intelligent tires could be a large step forward in safety, especially for ADAS and autonomous driving features.

What is your research about?
The automobile industry has long been trying to improve road transportation safety, and one important field of development has been active driver aids. This started with the introduction of ABS and has now moved into autonomous transportation. Tires have always been considered a passive element of the vehicle. However, more recently the idea of the ‘tire as a sensor’ has surfaced and become one of the major research thrusts in tire and vehicle companies. The intelligent tire research at the Center for Tire Research (CenTiRe) began in 2010 and has been going strong ever since. As part of this project, we have developed a deep learning algorithm to classify road surfaces based on acceleration measured inside the tire. The information about the road surface would be highly beneficial for the developing automotive sector to design new control strategies.

How can tires be made to accurately sense road conditions?
All forces between the vehicle and the road are transmitted via the tires. Utilising the tires as sensors would therefore be more accurate and beneficial than trying to estimate the road surface using other indirect and non-contact methods such as cameras, radars or lidar sensors. The physical phenomena that occur when a tire is rolling on any surface can be captured using sensors contained within the tires. The data from the sensors would finally provide sufficient information regarding the tire-road interaction. Emerging sensing technologies, compact accelerometers or strain gauges can be instrumented inside the tires to sense the interaction between the road surface and the tires without modifying the tire properties.

Which surfaces were chosen for this project and why?
For this project, we chose the four most common surfaces: dry asphalt, wet asphalt, snow and ice. Any vehicle on the road frequently experiences these four surfaces. Advanced control strategies and self-driving cars tend to struggle to match their dry performance when driving on low-friction surfaces such as snow, ice or wet asphalt. With these four surfaces, we can cover an extreme range of friction surfaces. Successful classification of these four common surfaces would lead to the next phase, which includes classification of additional surfaces. These would usually form a subcategory of one of the four above-mentioned surfaces, like deep standing water, shallow standing water, fresh snow, etc.

What are the challenges for this technology?
The update rate of the sensor would have to be improved to keep up with the sampling rate of existing vehicle sensors. Reducing the features used in the classification models is a challenge, in other words, to build a robust model that is independent of parameters like tire wear, which is difficult to measure using existing sensors. Advanced sensing technologies would provide the pathway for meeting the desired update rates and reaching a production-ready stage for this technology. In addition, efficient wireless transmission technology should be designed in order to transmit data from the innerliner of the tire.

What’s next for this project?
The prediction of road surfaces opens the door to the next generation of control strategies for automobiles, as their performance would be drastically improved on low-friction surfaces. Common control systems such as ABS, electronic brake distribution, electronic traction control and electronic stability control will be improved on all frictional surfaces.

Reinforcement learning models can be developed to build robust models for different tires and different vehicles. Vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication can be employed to share the details of the road surface with other vehicles on the road. In addition, growth in self-powered sensors, such as piezo-electric technology, would aid in getting this system to production sooner.

Chidambaram Subramanian will give a presentation titled Road surface classification using intelligent tires at the Tire Technology Expo Conference. Click here to book your delegate pass.

Back to speaker interviews