Capgemini explores the current state of AI-enabled in-vehicle function and what the future holds for next-generation software-defined cars
Since the introduction of the first commercially sold vehicles, automakers have been competing to maintain market relevance by releasing the latest and greatest models. In the US alone, the number of models on the market is expected to jump from 42 in 2020 to 62 in 2025. This rapid production cycle is not only due to a saturated market, but also partly because of consumers’ increasing demands for enhanced digital experiences and products. In order to win shoppers’ attention and increase sales, automakers are producing vehicles that are more like smart devices on wheels with internet connectivity and interactive dashboards.
AI is largely responsible for enabling these digital in-vehicle capabilities that drivers have come to know and love. But with mounting regulatory scrutiny over the broader AI landscape as well as advanced technology in cars, many automotive companies, as well as auto component and solution providers, must adjust their AI implementation strategies if they hope to continue dominating the market with state-of-the-art cars.
Current AI-enabled capabilities and future features
As a whole, AI maturity across the auto sector is fairly advanced. Most companies have made significant progress integrating AI. This is largely due to drastic shifts in product features as vehicles have become increasingly more software-defined with built-in functionalities.
AI enables a range of features in cars today, but arguably the most well-known and all-encompassing functionality that AI powers is the advanced driver assistance system (ADAS). From safety features such as lane assistance, adaptive cruise control, and driver monitoring to enhancements such as navigation, parking assistance, and autonomous driving, these AI-enabled systems have elevated the driving experience in recent years. Along with ADAS, inertial global positioning systems (GPS) are another example of AI-enabled technology commonly used in automotive applications, particularly in the development of autonomous driving and driver assistance systems. Broadly speaking, AI technologies are crucial for processing sensor data, interpreting the vehicle’s surroundings, making driving decisions, and controlling vehicle dynamics to enable various levels of automation and driver assistance.
AI also plays a crucial role in the flow of data from vehicles to OEMs and automotive suppliers, and vice versa. More specifically, AI models help automakers implement both firmware over-the-air (FOTA) updates as well as software over-the-air (SOTA) updates. And AI enables personalisation algorithms to help tailor vehicle features and settings to individual driver preferences and usage patterns.
Looking ahead, the industry is poised for a boom in AI-enabled vehicles, and McKinsey expects that 95% of all connected cars will be powered by AI by2030. Over the coming years, more complex AI models will enable more advanced iterations of today’s existing vehicle features. What’s more, these capabilities will be even more user friendly and geared towards customer engagement, like smarter and safer infotainment centres. AI will also allow for fully automated parking in areas with no distinct parking lines or boundaries, which is needed for the current parking assistance capabilities in today’s cars. AI advancements should also drive higher levels of data fusions from multiple sensors as well as dynamic environmental modelling, improved anomaly detection and fault tolerance algorithms.
Bumps in the road to advancing AI
Despite the current state of AI maturity and the promising future of more advanced AI-enabled features, there is still room for broader AI adoption across the automotive industry. This delay is not necessarily due to technology or enterprise-level limitations, but rather regulatory roadblocks that are impeding AI implementations.
The automotive industry faces unique challenges and regulatory requirements related to safety, reliability, and liability when deploying AI-driven technologies in vehicles, such as parameters concerning emissions, cyber security, and data privacy. This inevitably influences the pace of adoption compared to other sectors for both auto industry stalwarts and digital-native start-ups. Currently, all automakers are practicing rigorous testing and validation processes that are necessary to verify the safety and performance of AI-driven features, including simulation testing, real-world testing, and compliance with safety standards, such as ISO26262.
Despite the challenges that regulations can bring, OEMs and auto suppliers are increasingly finding the lack of standards specifications to be a barrier. This is particularly apparent across the industry as companies are begging to create their own SDV and ADAS solutions using AI. However, with no governmental guidance across geographies and use cases, automakers and suppliers are often delayed in deploying new functionalities.
Enhancing AI implementations
Rather than trying to boil the ocean, OEMs and their network of component and digital suppliers should start small when rethinking their AI implementation strategies. This hyper-focused approach begins by selecting specific AI-enabled features that align to their business goals and set realistic targets accordingly. Many automakers and Tier 1 suppliers that have implemented AI-enabled features find that certain capabilities are not showing a return on their investments, and that is largely because enterprises have not accurately accounted for changing consumer preferences and behaviour. According to automakers, car buyers are carefully evaluating vehicle features vs cost in today’s economy.
AI technologies are crucial for processing sensor data, interpreting the vehicle’s surroundings, making driving decisions, and controlling vehicle dynamics
It’s also important for auto enterprises of all sizes to remember that AI implementations on any scale and for any use case present considerable change management challenges. And that is exactly why this transformation must come from the top as a CEO-level priority. From a more tactical perspective, enterprises should take more control of their software development functions and processes by bringing them in-house. This will allow auto companies to better manage the software that is going into their vehicles and own the intellectual property rights.
It’s also essential for enterprises to embrace AI implementations as an ecosystem play. Traditionally, car manufacturers have shipped their parts to their assembly plants and owned distribution. However, the advanced AI-enabled features that will one day dominate the market will not be solely developed in manufacturing plants. Auto production processes will now occur in partner facilities as they have the technology and talent needed to develop software. Auto companies must therefore find trusted partners and establish an ecosystem framework with clear reporting structures and communication cadences to streamline processes, shorten time to market, and ultimately install best-in-class AI-enabled features in their vehicles year in and year out.
The automotive landscape is rife with competition. Most automakers are releasing cars with comparable capabilities at similar price points. To stand out in today’s market and ensure resilience in the future, companies must produce advanced vehicles with markedly enhanced digital capabilities. And the key to achieving this advancement lies in furthering AI implementations.
About the authors: Vamshi Rachakonda is Vice President, Manufacturing, Automotive and Life Sciences, at Capgemini Americas. Raghu Mocherla is Vice President of Engineering, Head of Industry for Automotive, Aerospace and Defense, and Manufacturing at Capgemini.
