AI, Deep Learning Imperative For Improved Autonomous Vehicle Development
atr

Knowledge Hub
Automotive Performance Engineering

Powered by
SIMENS-LOGO1.png

AI, Deep Learning Imperative For Improved Autonomous Vehicle Development

Artificial Intelligence Deep Learning Imperative Autonomous Vehicle Development Tata Elxsi
{data-social}

The autonomous vehicle will form a major part of the future of mobility, and while there are five levels of autonomy, only time will tell what level will be adopted across different markets of the world. However, it is imperative to carry out testing and validation of all levels of autonomous driving solutions for the vehicle, since it is highly critical in ensuring the safety of vehicle occupants as well as other road users. Nevertheless, the time-to-market for all automotive technologies, including autonomous driving is getting shorter, for which digital tools are required for faster validation.

AUTONOMOUS DRIVING DEVELOPMENT USING AI

A company that has been offering solutions for autonomous driving technologies development is Tata Elxsi. The company’s Autonomai solution is its autonomous vehicle middleware platform, which is infused with deep learning and artificial intelligence (AI) capabilities. This product can be used by OEMs to begin development of their autonomous architectures, and can be used in environments like farm equipment, where there is less human obstruction.

CLICK HERE FOR A WEBINAR ON PERFORMANCE ENGINEERING FOR AUTONOMOUS VEHICLES

Advanced Driver Assistance Systems (ADAS), which has been long associated with Autonomous Driving, is hard-wired, where algorithms are taught to behave in one way alone. Meanwhile, autonomous driving will not need to have this feature built in, but also needs to be capable of learning over time. This is said to be the difference between ADAS and autonomous driving, for which Deep Learning and AI have been built into each algorithm. The inputs received from various sensors are processed by the Autonomai Sensor Fusion Engine and then go to the drive-by-wire system carrying out the actual autonomous driving.

In the case of component suppliers looking at developing products for autonomous vehicles, there is also a full-autonomous platform, which is supplied with an adaptation kit to drive different components. The kit provides the hardware to perform drive-by-wire-functions, while the software platform gives the components the intelligence to drive. This is the first time that Tata Elxsi has supplied component manufacturers with such a technology, and had showcased it at CES 2017.

CLICK HERE TO WATCH A VIDEO ON DEVELOPING AUTONOMOUS VEHICLES AT SCALE

CONNECTING THE AUTONOMOUS VEHICLE

In addition to this, there is a requirement for Autonomous vehicles to be connected to everything (V2X) around, including other vehicles, traffic monitoring systems, and other infrastructure forms. The company noted that V2X is complementary to autonomous driving and ADAS, and is a key part of the overall development of connected, self-driving vehicles. In line with V2X, the company is also offering simulation and validation tools that cover the entire portfolio of requirements of a vehicle to sense its surroundings, gain information from various infrastructure sources, and subsequently take on-road decisions.

Tata Elxsi is also carrying out advanced R&D for future vehicle models, working in the areas of autonomous vehicles, integrated cockpits, multi-modal interfaces, and algorithms, for electric, hybrid and ADAS. The company believes that the automotive industry needs to mature and move to more open sourcing, which would subsequently accelerate the speed of development. The entire gamut of technologies, offered as a platform solution can be utilised to build complete vehicles, or components for vehicles to attain a higher level of autonomy.

TEXT: Naveen Arul