India’s automotive industry stands at a critical juncture as it transitions from hardware-driven engineering to software-defined, AI-optimised mobility. While features like Advanced Driver Assistance Systems (ADAS) mark a vital starting point, OEMs face complex technical, regulatory, and infrastructural challenges.
The path to intelligent mobility is paved with complexities—from a severe shortage of specialised talent in embedded systems and automotive software to the daunting task of blending modern digital architectures with legacy vehicle systems. In a market where affordability dictates adoption, the high costs of LiDARs, radars, AI processors, and high-performance computing infrastructure hinder scalability. Regulatory ambiguity, particularly around data privacy and cybersecurity, further compounds the challenge as global and local standards continue to evolve.
To begin with, “improving road safety through advanced technologies is essential,” even though ADAS adoption in India is still nascent, said Biswajit Bhattacharya, Lead Client Partner & Automotive Industry Leader, IBM Consulting India & South Asia.
Sharing his thoughts with Mobility Outlook, he elaborated that infrastructure inconsistencies—such as missing lane markings and congested roads—impair ADAS functionality, while public perception still sees these technologies as luxury features rather than safety essentials. Moreover, the absence of clear regulatory frameworks for ADAS and autonomous vehicle deployment, coupled with the threat of cyberattacks, raises the stakes for industry players.
According to Bhattacharya, sustained R&D and strong policy intervention are essential. “We need a collaborative approach between automakers, policymakers, infrastructure agencies, and tech providers to create a viable ecosystem for these technologies to thrive,” he noted.
Rishi Aurora and Biswajit Bhattacharya
Elusive Challenges
Replicating human common sense—especially in chaotic, unpredictable traffic—remains one of the most elusive challenges in AI. While people instinctively make split-second decisions based on context, experience, and judgement, machines struggle to emulate such reasoning in real time.
Giving his perspective, Rishi Aurora, Managing Partner at IBM Consulting India & South Asia, explained that the company is tackling this by combining powerful AI models with deep industry knowledge to enable systems that can interpret real-world complexity. These solutions are designed not just to process massive data streams but to mimic human-like understanding—adapting to ambiguous scenarios and making decisions that align with real-world expectations, especially in mobility environments.
Aurora said that IBM places equal weight on responsible innovation. “By combining cutting-edge AI research with practical applications and a focus on explainability and fairness, the company is helping its customers navigate the complexities of common sense reasoning in dynamic environments,” he added. With this holistic approach, the company helps its automotive clients embed AI that is not only intelligent and scalable, but also accountable—laying the foundation for safer and more reliable AI-led mobility.
Bringing Human Reasoning To AI
In high-risk environments like autonomous driving, replicating the nuances of human judgement is essential yet exceptionally complex. On addressing these issues, Bhattacharya mentioned that autonomous vehicles “rely on a layered intelligence stack—combining rule-based systems to follow traffic laws, machine learning to interpret sensor data, and human-in-the-loop frameworks” allowing remote operators to intervene when algorithms hit a limit.
The company integrates human-like reasoning into AI workflows through practical, high-impact applications. In customer service, AI chatbots equipped with “natural language processing and understanding” simulate human interactions to resolve queries swiftly. In supply chain optimisation, AI analyses data on demand, logistics, and vendor capacity—mimicking strategic decisions a manager might make, but at scale and speed. On the factory floor, AI-driven vision systems handle quality control by identifying defects with precision beyond the human eye, trained on vast datasets of real-world anomalies.
Through such applications, “IBM uses such adaptive approaches to AI to push the boundaries of machine reasoning, perception and decision-making in autonomous vehicles,” he said.
Smarter AI At The Edge
Running advanced AI on edge devices like those in vehicles is a challenge marked by tight processing power, energy limitations, and the need for real-time responses. According to Aurora, the company addresses these constraints through a suite of innovations. By using ‘model compression’ techniques, the company shrinks machine learning models without sacrificing accuracy. “We also design efficient model architectures to optimise the balance between model performance, computational resources and deployment constraints. Additionally, we use federated learning (a type of machine learning where the model is trained across multiple devices or servers), which enables devices to learn collaboratively while keeping the data on local servers, decreasing bandwidth usage for edge devices. By combining these methods, IBM helps automotive companies use advanced AI capabilities on edge devices,” Aurora added.
This focus on edge optimisation allows automotive firms to deploy intelligent features in real time, even in resource-constrained environments. One such application is predictive maintenance, as explained by Bhattacharya. By analysing historical data on vehicle usage, component performance, and environmental conditions, “AI systems can identify patterns that indicate potential failures,” mirroring a mechanic’s intuition but with far greater precision. This shift from reactive to proactive maintenance reduces downtime and costs, while enhancing vehicle reliability.
Together, these capabilities bring AI closer to the vehicle and the customer—enabling a new era of responsive, intelligent, and efficient mobility solutions, he added.
Also Read: