Tata Elxsi Focuses on ADAS as Bridge to Autonomy

March 11, 2014

Terry Costlow

Image processing is a central aspect of Tata Elxsi’s efforts in safety technologies.

Auto industry watchers are giving plenty of attention to the advanced driver assistance systems (ADASs) that will one day be linked together to provide autonomous driving. As design teams around the globe strive to improve safety with these standalone systems, Tata Elxsi is leveraging its digital signal processing to build a presence in ADAS development.

The independent public company is focusing on ADAS as executives strive to take advantage of the new growth opportunities in automotive electronics. Though the design-services company has broad skills in electronics, one of its strengths fits well with ADAS and autonomy, which will use cameras to monitor roadways, other vehicles, and even drivers.

“One of our main focuses is image processing,” said Anil Sondur, Vice President at Tata Elxsi’s Transportation Business Unit. “A lot of changes are happening in DSPs; they are coming with built in features. We’re benefiting from these advances.”

Currently, the company is focusing on three areas of ADAS that all use cameras. These systems will soon begin migrating from luxury vehicles to more mainstream vehicles, Sondur predicted.

“Lane departure warning, pedestrian detection, and 360-degree sensing are primary areas for us. In model year 2014-2016, the transition to these features will accelerate dramatically,” he said.

As engineers and customers alike become more comfortable with these standalone safety technologies, design teams will focus on pulling them together to provide limited autonomy. Most observers predict that, in initial offerings, cars will only employ brake and steering automation in limited types of driving. Minimal levels of autonomy may first show up in very slow driving conditions.

“The car will take care of driving when you’re in bumper-to-bumper traffic,” Sondur said.

Though some consumer-oriented pundits predict that autonomous features won’t be a major cost adder, that view isn’t typically shared by many in the automotive industry. The price of standalone systems may well come down significantly, but the myriad challenges of linking them together will not come without cost.

“Going forward, we’ll be building autonomous vehicles using ADAS,” Sondur said. “Autonomy will be restricted to high-end luxury vehicles, it takes a lot of extra money to get limited autonomous driving.”

Pulling together information from multiple sensors located around the vehicle and making intelligent decisions is foremost among the many challenges. Design teams will have to determine where data will be processed and how it will be transferred.

“You can either combine functions into one ECU or move to higher bandwidth networks,” Sondur said. “People are starting with multiple CAN networks, but there’s a lot of work going on with FlexRay and Ethernet.”

Automotive OEMs are taking many steps designed to pave the way for autonomy. One of these is to adopt standards that make it easier to link together modules developed by a range of manufacturers. After many years, AUTOSAR (AUTomotive Open System ARchitecture) is finally gaining acceptance.

“AUTOSAR is coming in in a big way, it will absolutely grow. A lot of OEMs now insist on AUTOSAR compliance,” Sondur said.