In a rapidly evolving automotive landscape, Mercedes-Benz is pioneering the next era of intelligent mobility with the Mercedes-Benz Operating System (MB.OS). This keynote explores how MB.OS fundamentally transforms the vehicle into a software-defined platform—designed in-house, scalable across global markets, and fully updateable over-the-air. By bundling intelligence into four central domains—Infotainment, Automated Driving, Body & Comfort, and Driving & Charging—MB.OS delivers a cohesive chip-to-cloud architecture that enables unprecedented levels of personalization, adaptability, and AI-enhanced performance.

Key innovations include a multi-agent virtual assistant powered by large language models, immersive passenger experiences, and high-performance computing infrastructure ready for the next generation of autonomous functions. A spotlight is set on including AI Technology into customer functions. This presentation will outline the strategic and technical foundations of MB.OS, highlighting its role as a catalyst for innovation across the entire value chain — from semiconductor to customer experience.

The Neue Klasse is BMW's new model generation, heralding a new era of software-defined vehicles (SDV) and showcasing the future of driving and user experience. The new BMW Panoramic iDrive is a revolutionary human-machine interface that delivers unprecedented simplicity, personalization, and driver orientation for a perfect user experience. The BMW Operating System X is the “Infotainment Operating System”, which will be scalable across all BMW Group brands and segments. It evolves from BMW Operating System 9 and is based on an Android Open-Source Project (AOSP) software stack. The system signifies a paradigm shift towards software continuity, offering enhanced update and upgrade capabilities, ensuring it is both future-proof and backwards compatible. Designed for additional functions, it keeps vehicles that are equipped with the new BMW Panoramic iDrive at the cutting edge of technology for an extended period. The overall concept of the new BMW Panoramic iDrive, combined with Operating System X, has been made possible by a significant technological leap forward and perfectly demonstrates the potential of a software-defined vehicle. Developed in-house, this system further evidences our technological leadership in digital systems and functions. Third-party content remains fully and seamlessly integrated, continuing to offer a wide range of streaming, gaming, and infotainment offerings in the future. Our software development teams are working globally, enabling the rapid implementation of versions for local markets like China, thereby enhancing flexibility. The system offers top-notch innovative functions based on software, is continuously updated to remain “fresh” over generations, knows, assists, and supports the driver, integrates into leading digital ecosystems, and is tailored to market requirements and environments, among many other capabilities. So stay tuned for more in 2025, and join my presentation to gain insights into how we are implementing these groundbreaking technologies and enhancing the user experience.

Software-defined vehicles (SDV) and advanced electronics centralized / zonal electronics architectures are leading trends in the automotive industry which are fueled by the rapidly growing SoC (System-on-Chip) performance. OEMs are challenged to manage this major architectural transition while delivering customer relevant features that differentiate their cars regarding functionality and cost.
As this SDV transition is not only technically highly complex but also very resource and investment intensive, it becomes more and more critical to allocate investments in differentiating areas. With the skyrocketing electronics complexity an approach to do most development in-house becomes too expensive and too slow. Hence, all players need to decide where to differentiate and compete and where to cooperate and partner. 

Based on these considerations, QNX, Vector and TTTech Auto partner to deliver a pre-integrated and tested SDV software platform based on their expertise in safe and secure operating environments as well as safe and secure automotive approved middleware solutions.
In this keynote, the industry heavyweights will elaborate on their cross-industry collaboration. They will present their joint, pre-integrated and certified out-of-the-box solution that meets the highest functional safety (ISO 26262 ASIL D) and security (ISO 21434) standards. It is designed to accelerate and enhance integration of software through a continuous integration test environment provided by the parties. This new collaboration aims to eliminate the complexities and high costs associated with software integration, allowing automakers to focus on creating innovative, consumer-facing applications that enhance brand loyalty and differentiation. The new platform will be simple by design, performant by nature and able to scale vehicle-wide. To ensure best and easy customer service and access, the new platform concept of the parties will ensure a single point of contact for customers.

This partnership also demonstrates very clearly the transformation of the collaboration model from the traditional value chain to value network and enables the parties to act as a central value network platform. It is open to include contributions from additional ecosystem players as well as from open-source initiatives.

In the rapidly evolving landscape of automated driving technology, Bosch and Volkswagen´s software company CARIAD have formed a strong strategic partnership to drive the development of scalable systems for assisted and automated driving. 

This collaboration focuses on harnessing driving data to enhance AI models, significantly improving the performance of Level 2+ solutions and paving the way for higher levels of automation. 

As we transition from Level 2+ driver assistance systems to Level 3 automated driving, our efforts ensure a seamless evolution of automated capabilities. Central to this initiative is the utilization of cutting-edge end-to-end AI technologies, , which enable the development of robust software solutions that meet both performance and safety standards.

In a global market characterized by high safety standards, Bosch and CARIAD are committed to delivering competitive solutions for volume production. The collaboration positions Bosch and Volkswagen at the forefront of the European competition in the field of automated driving and underlines their unwavering commitment to innovation and safety in the automotive sector. The future of autonomous driving is now, and together, we are driving towards a safer and more advanced mobility landscape.

SDV: The road taken and associated pitfalls

Different OEMs have pursued their own full stack or Vehicle.OS development strategy. Primary debate of this strategy is about huge efforts already spent on elements of software stack as well as integration, which do not necessarily differentiate the features seen by the end user. These parts of the software, though "non-differentiating", are "critical" for delivering the SDV objectives and are also "complex" in nature to implement. This complexity arises out of the absence of early integration and the validation of software being developed.

This "do it all" approach has led to redundant and large scale spend by OEMs, bloating the overall costs and expanding the timelines to realize SDV-based vehicles.

What’s sustainable?

Various industry initiatives include open-source based approach to achieve a common “Vehicle.OS” platform beyond Adaptive AUTOSAR, to free up bandwidth, focus on innovation and differentiation; as well as reduce development costs. There is one more approach being taken where ready E&E vehicle platform from new OEMs from the west and from China are being used as base and customize or adapt as appropriate.

There are 2 categories of problems emerging from this approach taken

1. Programs that have already decided to develop all the elements of the stack are struggling on the underestimated efforts and complexity of integration.
2. Programs that have not really embarked full scale on the journey are unsure about what approach to be taken.

The way forward

There is an opportunity to come up with a new solution for an ecosystem that leverages such common stack strategy to accelerate cycle time, significantly reduce costs for OEMs, without compromising on the overall SDV objectives. Following are some key tenets of such an ecosystem:

• Vehicle.OS distributor scaling up to provide production-intent integration platform, based on a reference architecture.
• Integration of a 3rd party AD/ADAS stack into this platform with minimum architecture impacting features.
• 3rd party integration partner as a link between OEM feature teams and vehicle-level validation team
• Consortiums like ECLIPSE; where industry-ready open source vehicle software solutions are innovated
• Robust and proven validation ecosystem with fully integrated solution
• Outcome based business model offered by such ecosystem players
• Best practices necessary for development as well as integration to realize such an ecosystem

This presentation explores on how such an ecosystem would look like, what would be key the advantages to the OEMs along with potential challenges and key organization changes that maybe needed to realize the full potential of SDV.

The global automotive market is going through a major transformation with new technologies, changes in consumer needs, and players entering, consolidating or leaving the market. A main driver of this change, China is one of the largest markets and the ground of the world most intense competition, where personal car owners fuel a growth for digitalization and automated features like Highway and Urban NOA (Navigate on Autopilot) or Assisted Parking at a breakneck pace. 

In Europe, ADAS developments are closely following many standards and regulations and are mostly driven by technological requirements. In contrast, China OEMs are focused on offering a seamless digital and automated driving experience tailored to consumer demands. As such, COEMs have to respond rapidly to market changes. They need to more quickly scale-up (improve features) and scale-out (make them broadly available) their ADAS offering to discerning consumers.

Horizon Robotics addresses this challenge of accelerated scale-up and scale-out by following a novel approach of hardware and software co-design. Horizon’s processing hardware features an advanced proprietary accelerator, the BPU™, co-designed with algorithms requirements such as Transformer, LLMs and End-to-End architecture. Horizon driving stack SuperDrive™ is tightly developed with the Horizon’s processing hardware to reach optimal performance and energy efficiency through a combination of End-to-end Transformer-based neural networks and formal rule-based algorithms. This hybrid approach enables a better generalization of the solution to scale-out to more diverse ODDs, while complying with the stringent safety requirements of smart driving applications. As a result, SuperDrive™ delivers a safe, assertive, humanlike smart driving experience across diverse ODDs. 

Horizon Robotics approach is market proven, with over 7 million solutions shipped today. We continuously research and invest in the future of smart driving. Our world class talents, with years of experience in ADAS, AD and smart HMI are now working to create a seamless driver and passenger experience by introducing a VLM-based solution (Visual-Language model) for the passenger to more efficiently interact with the vehicle’s embedded system during their smart driving rides. 

Horizon Robotics will present its novel approach and technology overview to make the world safer and better through smart driving.

With the advent of AI being deployed at the edge, the idea of a personal assistant in your car is not that far off. With AI models running locally in your car’s system, your future car, similar to AI PCs today, could operate independently of the cloud, ready to interact with you, seamlessly. From adjusting settings with voice commands, to making dinner reservations while navigating the roads, to reminding you of an appointment then connecting you to the call, AI will redefine personal comfort, convenience, and productivity in the car.

Beyond that, AI will have a profound impact on safety as well, by being able to calculate object location, distance, and mass in real time and make quick decisions on how to maneuver safely around them in a way that can one day help make autonomous driving ubiquitous.

It all starts with locating AI processing close to the source of data to uncover insights and actions. This provides the low latency that is needed for fast decision making, but also allows for enhanced levels of security and connectivity. It also requires a heterogeneous and scalable computing approach, backed by an open ecosystem, to make AI at the Edge a reality.

During this presentation, attendees will learn about:

• The growing compute and power requirements needed to drive AI at the Edge

• Techniques to drive power-efficient AI with heterogeneous computing solutions 

• How to achieve compute efficiency through workload optimization with adaptive computing

• How AMD technology is propelling AI innovation by advancing central compute capabilities for ADAS and automated driving. 

• How integration topologies, such as chiplets, are being considered for future platforms

As automotive manufacturers increasingly recognize software as a core differentiator, the integration of AI and Generative AI tools into the software development process is a key consideration for advancing the software-defined vehicle (SDV). These tools offer the potential to substantially reduce development timelines and resource costs across the full lifecycle of the SDV. However, successful deployment demands significant customization, not only to meet the stringent quality and safety requirements of the automotive industry but also to align with each manufacturer’s unique development goals, processes, and software organization maturity. 

This session offers a pragmatic approach to addressing these requirements through adaptable, phased AI integration. It explores key challenges that manufacturers and suppliers may encounter in deploying AI-driven software tools, focusing on practical solutions that support industry standards while promoting agility in tool adoption and experimentation. Attendees will gain a grounded framework for leveraging AI effectively, positioning manufacturers to thrive amid the industry’s inevitable shift toward software-centric vehicles.

With ADAS/AD and IVI emerging as the key differentiators for software-defined vehicles (SDV), automotive companies depend on high-performance SoC solutions to support the implementation of an increasing variety of features in shorter time at lower cost. At the same time, the solutions need to be future-proof for fast-evolving algorithms and widely scalable to support software re-use over the whole range of car models.

Current automotive high-performance SoCs, typically from large non-EU based semiconductor vendors, address this increasing variety of algorithms with highly heterogeneous architectures, which in general are not only difficult to orchestrate and scale, but also result in an overall low silicon utilization, increased cost, and long time to market.

In contrast, videantis offers a disruptive homogeneous processing architecture which replaces several distinct specialized engines of a heterogeneous SoC with always the same highly efficient programmable processor core for all key algorithms. This approach results in high silicon utilization over a wide variety of use cases, significantly reduces both chip design and SW development complexity for lower cost and faster time-to-market, and inherently leads to a seamlessly scalable SoC architecture.

Proven in automotive serial production over generations in more than 20 million vehicles to date meeting ASPICE and ISO26262 requirements, a high-performance reference SoC platform based on the latest generation videantis architecture is under finalization, enabling SW stack partners, system integrators and automotive OEMs and Tier 1s to engage hands-on with this disruptive SoC architecture paradigm.

Automotive SoC products based on the videantis homogeneous processing architecture will become available in the form of custom SoC families through our manufacturing partner, stand-alone chiplet offerings, and as silicon-verified subsystems as part of 3rd party off-the-shelf SoC solutions. 

In the era of Software-Defined Vehicles (SDVs), traditional OTA solutions are no longer enough to keep pace with the increasing complexity and dynamic nature of modern vehicle architectures. This session will explore six critical capabilities that next-generation OTA systems must possess to address these challenges and unlock new business opportunities:

1. Full-Vehicle Updates: Seamlessly update any ECU across the entire vehicle at any time, enabling rapid delivery of new features, critical patches, and security updates. This capability reduces software development and deployment costs by streamlining release cycles and minimizing engineering overhead.
2. Differential Updates: Efficiently manage software changes by deploying only the specific updates needed, rather than replacing entire software packages. This reduces bandwidth usage, minimizes downtime, and lowers operational costs for OEMs.
3. Digital Software Twin Modeling: Enable real-time validation and simulation of updates using digital twins to catch issues before they reach the vehicle. This reduces recall risks, improves software quality, and accelerates time-to-market for new features.
4. Universal Software Packaging: Ensure compatibility across diverse hardware platforms, reducing the complexity and costs of managing software for global vehicle programs. This capability simplifies engineering workflows, lowers support costs, and accelerates global software deployments.
5. Multi-Stage Rollouts: Allow precise, phased deployment of updates, providing OEMs with the flexibility to test, monitor, and adjust updates in smaller batches before a full-scale release. This minimizes risk, improves stability, and enhances customer experience by reducing potential downtime.
6. Granular Traceability: Provide detailed insight into every step of the reflash process, including logs, audit trails, and the ability to trace issues back to their origin for rapid resolution. This supports compliance, reduces debugging costs, and improves operational efficiency.

With these six fundamentals established, we'll shift focus to the transformative business outcomes they enable. By connecting these core OTA capabilities with advanced data logging, real-time diagnostics, and secure remote command execution, OEMs can:

• Reduce maintenance costs through predictive diagnostics and automated issue resolution.
• Improve customer experience by minimizing downtime, enabling remote fixes, and delivering continuous feature enhancements post-sale.
• Unlock new revenue streams by monetizing data insights and offering feature upgrades, subscriptions, and personalized services after the vehicle sale.
• Lower development costs by streamlining software workflows, minimizing engineering complexity, and reducing recall risks.

Key Takeaways:

• A real world understanding of the six essential capabilities for next-generation OTA systems.
• Insights into how these capabilities drive measurable business outcomes, including reduced maintenance and development costs, improved customer experience, and new revenue streams.
• Practical strategies for scaling OTA and connected ecosystem services to support global vehicle programs.

Join us to explore how mastering these six fundamentals can transform vehicle software management into a comprehensive, real-time solution that drives safety, efficiency, and innovation—ultimately delivering long-term value for OEMs in the SDV landscape.

Every automotive organization faces challenges in tackling the "Software Defined Vehicle" development changes. They strive to continuously deliver high-quality features and software updates to vehicle platforms with diverse and evolving processor architectures.  Even those with advanced software development capabilities must ensure their systems and software engineering processes are meeting functional safety requirements.  This transformation is proving to be difficult and is taking quite some time. The difficulties are due to technical, cultural, process and organizational challenges.

Specific challenges we have observed include:

• Refactoring software from microcontrollers to microprocessors. The hardware requirements are evolving to new central and zonal E/E architectures and will continue to evolve over carlines and time. This requires software that is adaptable to different hardware and middleware configurations. Decoupled hardware and software development is required to deploy software over a fleet of carlines.
• Updating toolchains, software factories and development platforms. Continuous integration and deployment of Software requires continuous testing. It is key to have an efficient V&V strategy, and that software factories are also designed for automotive safety and quality practices (e.g. ISO26262). This "shift-left" approach requires virtualization/simulation to ensure software compatibility and functionality across the hardware systems in the fleet. How to manage the complexity and deploy it efficiently to the cloud are some of the underlying challenges.
• Including requirement and system engineering in the agile development processes. Historically these processes were essential early steps in the V-Diagram and primarily cascaded through the development process. Now they need to be regularly reviewed and revisited as part of the continuous development.
• Reskilling and reorganizing. Automotive companies are hiring software engineers from other sectors into their organizations, and are working to retrain their existing staff for the future mobility needs. Getting them all aligned, integrated, communicating and collaborating is a challenge. The goal is that both worlds complement each other to get high quality software aligned with good system design.

We will discuss how processes and toolchains are evolving to address these challenges. We will examine examples of how automotive companies around the world are bridging different systems and software practices. How they are leveraging cloud, virtualization, SysML2, and other disruptive technologies to shift left, improve efficiency, and finally accelerate the development of Software Defined Vehicles.  

AI is no longer just an enabler of innovation in automotive — it is the engine behind the evolution of SDVs. From design and manufacturing to in-vehicle experiences, AI is redefining value across the entire ecosystem. But with this evolution comes a new level of systemic complexity and cyber risk. As SDVs become AI-empowered, attackers gain new tools, and traditional security approaches fall short. In this session, we’ll uncover how the growing intelligence of vehicles demands a shift in cybersecurity strategy — from fragmented defenses to whole-system visibility. Learn why the ability to see, predict, and respond across the full digital footprint of the vehicle is now mission-critical.

Future vehicles will be based on multi-featured, more centralized compute platforms as a foundation for AI-based functions to enable highly automated driving. Advances in automotive SOCs and AI accelerators have led to an abundance in the adoption of emerging technology nodes and chiplet packages. We will discuss the resiliency challenges for chiplet based automotive systems, and optimizing health of safety critical systems using prognostics and overall silicon lifecycle (SLM) based solutions which improve quality and yield; and also address aging and degradation challenges for improved overall dependability. This talk will describe technology trends, and the need for Silicon Lifecycle Management for predictive maintenance. It will also describe guidance from IEEE 1856-2017: IEEE Standard Framework for Prognostics and Health Management of Electronic Systems.

More AI based components in future highly automated vehicles impose new risks in the context of automotive safety. Moreover, safety critical software updates over the air require new verification and validation techniques to assess whole lifecycle compliance with safety regulations. An extended V-model based advanced dependability engineering approach is regarded as a powerful systemic solution. There has been a growing effort recently in the development of newer standards for dependable computing. This talk will discuss the new topics being addressed in IEEE P2851 and other recent international standardization efforts in the area of silicon health and automotive functional safety, including -

• IEEE 2851 family of standards which address a known problem related to functional safety interoperability
• IEEE 2851-2023 which defines a dependability lifecycle of products with focus on interoperable activities related to functional safety and its interactions with reliability, security, operational safety and time-determinism. It defines methods, description languages, data models, and databases that have been identified as necessary or critical, to enable the exchange/interoperability of data across all steps of the lifecycle encompassing activities executed at IP, SoC, system and item levels, in a technology independent way across application domains such as automotive, industrial, medical and avionics safety critical systems, and to support developing methodologies such as Artificial Intelligence.