Honda and Audi show the future of driving at Nvidia conference

This digital instrument cluster in the Lamborghini Aventador is powered by an Nvidia chip.

(Credit: Wayne Cunningham/CNET)

SAN JOSE--Over the past decade, new in-car electronics have helped us navigate and made more music easily available while driving. But if the work shown at Nvidia's GPU Technology Conference (GTC) is any indication, bigger and better changes are in store.
Among the many automotive seminars at this year's GTC, Honda showed off its development of a head-up display, while Audi discussed its initiatives to make urban driving safer.
In Honda's seminar, Victor Ng-Thow-Hing, Principal Scientist at the Honda Research Institute in Mountain View, California, showed h
ead-up display technology that makes current production examples look extremely primitive. Instead of simply projecting a speed readout or turn-by-turn directions on the windshield, Ng-Thow-Hing demonstrated work in augmented reality, projecting location sensitive information useful to drivers.

Cars on display at the GTC using Nvidia technology were this Lamboghini Aventador, a McLaren MP4-12C, and a Tesla Model S.

(Credit: Wayne Cunningham/CNET)

In one example, street names were projected onto the windshield in such a way as to appear to be signs on actual buildings. Ng-Thow-Hing explained that merely showing a flat street name on the windshield leads to drivers having difficulty adjusting between the 2D information of the label and the 3D world outside of the car.
This type of projection would be extremely useful in cities, where street signs can often be difficult to find amongst the urban clutter. And a clear indication of street names would lead to less confusion and stress for drivers.
Ng-Thow-Hing also showed two driver assistance projections, the first using a nine-square grid projection high up on the windshield to give drivers better situational awareness. The driver's car is always in the center square, while the other eight squares appear red when another car is in that space. This grid projection would serve the same purpose as current blind spot monitors, but also work as a quick driver reference to see note cars to the front and back.
The second example projected the path of the driver's car when waiting to make a turn through an intersection, coloring the path in red when oncoming traffic made going ahead with the turn unsafe. Ng-Thow-Hing pointed out that drivers cannot always correctly estimate the distance and speed of traffic coming through an intersection, and so make mistakes that can lead to a crash. With this system, sensors would see the oncoming traffic and the car's processor would determine when the driver could safety make a left turn through the intersection.
The urban problem
Giving Audi's presentation was Mario Tippelhofer, an engineer working out of Volkswagen's Electronic Research Lab south of San Francisco. He prefaced Audi's work by noting that urban populations have increased in the U.S. more than 80 percent of the population now lives in an urban area.
Most accidents in urban areas are caused by traffic, passengers distracting the driver, driver stress, and the constant search for parking, according to Audi's research.

Audi uses Nvidia technology to integrate Google Earth imagery in its A7 model.

(Credit: Josh Miller/CNET)

Audi's Urban Intelligent Assist program attempts to address the causes of these accidents through a variety of technologies. One application presented by Tippelhofer was termed Smart Parking. Essentially, instead of a car's navigation routing you to a final destination, it would take you to an available parking space near the final destination. San Francisco has already installed smart parking meters, which detect when cars are parked next to them. This data can let the car know about available spaces before you start on a journey, and can update based on changing parking availability.
Complementing that technology was another application called Seamless Navigation, which would let you effortlessly send a destination from a smartphone to the car's navigation system. The car would then route to available parking, and your phone would provide walking directions to the actual destination.
Audi's parking applications would make it unnecessary to circle block after block looking for parking, and potentially causing an accident while scanning for an open spot.
Other technologies that Audi is looking to implement involve predicting traffic problems, helping drivers merge or make lane changes, and detecting when the driver's attention is away from the road ahead.
Tippelhofer said that Audi would be demonstrating some of these technologies this year.
Under the tech hood
Both the Honda and Audi technology initiatives go beyond current cabin electronics, which generally center around navigation, hands-free phone, and digital audio features. What might make these developments possible is the technology offered by Nvidia in the form of its Tegra line of graphical processing units.
At the GTC, Nvidia's Automotive Applications Manager Dave Anderson offered CNET some context for how automakers could deploy these new technologies. Nvidia's primary automotive product is the Visual Computing Module (VCM), a piece of hardware that includes a Tegra processor and will process sensor inputs from around the car, then output visual information for the driver.
Anderson touted how automakers can design a car's systems around the VCM, and plug the latest version of the module into the car at the time of production. As an example, he cited Tesla, which began designing the Model S to use Nvidia's Tegra 2 GPU. By the time the Model S was ready for production, Nvidia had released its Tegra 3 chip. The architecture let Tesla use the latest chip, keeping the car's electronics current.

This navigation unit in the new VW Golf runs off Nvidia's VCM.

(Credit: Wayne Cunningham/CNET)

The VCM mostly solves the problem of automakers releasing cars with cabin electronics two or three years old.
Nvidia's VCM currently sees use in models from Audi, Lamborghini, and BMW. As Nvidia has only made its VCM available for two years so far, many more automakers could be jumping on the bandwagon.
Addressing the issue of component cost, Anderson said that Nvidia can offer less capable but cheaper versions of the VCM. As an example, he noted that where Audi uses an advanced version that enables such features as Google Earth integration in its navigation, Volkswagen is deploying a simpler version in the new Golf, which we saw unveiled at the Paris auto show last year.
The presentations at the GTC, and Nvidia's own automotive offerings, show a huge technological potential for the automotive industry, with future models that can make driving more convenient and, hopefully, accident-free. In fact, they make today's car technology look very primitive.