L&T Techology Services develops autonomous vehicle platform

L&T Technology Services (LTTS), a provider of engineering, research and development services, has developed an autonomous vehicle platform VANGEN (Video Annotation Generator).

VANGEN is a platform that helps in creation of ground-truth data that is essential to verify and validate algorithms that go into the autonomous vehicles.

Commenting on the new platform, LTTS Chief Executive Officer & Managing Director Keshab Panda said the company is focusing on connected and hybrid vehicles and a pilot electric vehicle prototype is running in its Bengaluru campus.

"Our endeavour is to develop core technology for the ongoing efforts to create driverless vehicles. We have developed specialised algorithms to detect objects," he said. The algorithms help in sensing and controls for autonomous vehicles. To support the testing of vehicles for regulatory approvals, VANGEN helps in annotating the objects of interest in a semi-automated and automated manner and generates ground truth data against which the algorithms are verified.

LTTS has expertise on application of LiDAR technology (Light Detection and Ranging ) for detection of obstacles and measuring distances and angle of arrival of various objects around the vehicles.

LTTS has also generated 3D annotation that provides the depth of information of the objects that in turn helps significantly in autonomous driving since the depth parameter becomes a key for overtaking purposes.

The company avails NVIDIA graphic processing chips and platforms for sensing and mapping it with time and scale. "Our platform can detect 180 objects and now we are working on the third phase of integrating sound to it," Panda said.

LTTS is also developing advanced driver assistance services (ADAS) as part of their connected vehicles solution. The company successfully reconstructed 3D scenarios at different views using 2D images/ video stream, said LTTS Global Head of Transportation Shailendra Shrivastava.

"We leverage our knowledge in image processing to deliver low-cost ADAS solutions. We utilise multi-core capabilities to develop optimised image processing algorithms and port them across platforms to ensure faster go-to market," Shrivastava said. The ADAS application has been successfully tested on Bengaluru and Shanghai roads under different environmental conditions.

The company has also developed an algorithm for 3D reconstruction of a scene using 'Structure from Motion' technique.

DH News Service