Search This Blog

Tuesday, March 3, 2015

The Brains Inside Your Car Are About to Get Smarter

As reported by USA Today: With all this talk of co-piloted and autonomous vehicles, one question stands out. If cars gradually are taking over more of our driving chores, can they do so safely and error-free? After all, the dark side of this rise-of-the-machines scenario is a rogue vehicle that catastrophically misreads the data flooding its sensors.

Freescale Semiconductor (FSL) aims to raise the bar on the quality of chips used in increasingly sentient vehicles. On Monday at Mobile World Congress in Barcelona, the Austin-based company, which spun out of Motorola (MSI) in 2004 and is focusing on the booming Internet of Things space, announced a new S32V vision microprocessor. The company describes it as “the first automotive vision system-on-chip with the requisite reliability, safety and security measures to automate and ‘co-pilot’ a self-aware car.”

Translated into English, this simply means that Freescale’s new chip will to help automakers pack a new level of autonomy into future models that will require less of drivers by upping the processing power and reducing the error-rate from its onboard computer systems. 

“You don’t want the silicon (chips) running your Candy Crush game driving your car,” says Matt Johnson, Freescale’s vice president and general manager of microcontrollers. “Right now, the focus is on assisting the driver with things like lane departure and collision avoidance. But soon we’ll have a radical shift to having the car in control. That means automotive-grade silicon that can function with higher temperatures and with zero defects.”

Johnson gives the example of a car loaded with radar, lidar (laser radar) and ultrasonic sensors, information from which all needs to be aggregated by the vehicle’s computers and turned into a split-second decision about whether to act or not. But the human-cost payoff of successfully integrating tech into cars is apparent.

“Roughly 90% of auto fatalities are due to human error,” he says. “It would be great to help reduce that.”

Given long automotive production cycles and inherent regulatory testing requirements, Johnson says the new S32V chip is likely to make its way into production models by 2020.