Overview of Automotive Artificial Intelligence: Potential of Adapting Deep Thinking and Quick Learning Paradigm from Gaming Domain

Artificial intelligence (AI) has witnessed significant attention from both research and industry in recent years. AI is not a new area of research, and research in this field has been reported over decades since 1950 on a continuous basis. Renewed interest in this stream of computing has been primarily due to development of pathbreaking methodologies which have potential application in various industries including automotive. The buildup of research in the area of AI over the past 60 years and the general viability of application of the past and ongoing research especially in the area of automotive are the motivation behind this work. However there are still important gaps that need to the bridged to make it possible to develop truly general as well as adaptive intelligent machines with application to the automotive sector. The article is an effort to point out that a meaningfully useful general learning machine must not only be “general” that it is able to learn and solve various types of problems encountered in the automotive domain but also be able to learn quickly in an constantly changing, chaotic environment, for example, general traffic in India. The article reviews the present limitations of industry-accepted AI-based learning methodologies which have the potential to fill the gaps. A paradigm related to deep thinking and deep learning is discussed that has the potential to give a future direction to research to put forth truly general and adaptive intelligent machines.