With great strides, Bosch Mobility, the largest business unit of the Stuttgart-based Bosch Group, together with the software specialist and VW subsidiary Cariad, is advancing the perfection of automated driving functions. The driver assistance solutions controlled by Artificial Intelligence (AI) are now being tested on electric VW transporters and are "systematically trained with large data sets." In just under a year, the solution for the "software-defined vehicle" is expected to be production-ready for the Wolfsburg-based car manufacturer.
The expanded electronic control is the result of a deepened collaboration within the Automated Driving Alliance, which both companies established in 2022. The purpose of this cooperation is to develop Level 2 systems with the possibility of hands-free driving in certain areas of the city, on country roads, and highways using AI technologies. Another field of work involves developments at the Level 3 stage for the autonomous driving of the system on highways. Test vehicles are operating in Europe, the United States, and Japan.
Cars Stay Up-to-Date Longer
A year ago, at its Bosch Tech Day 2024 in Renningen, west of Stuttgart, Bosch announced plans to generate "billion-dollar revenues" with software by the end of the decade, in light of the "era of the software-defined vehicle" (Bosch Executive Dr. Markus Heyn). According to Bosch, the company employs 48,000 staff in software development — the Mobility business unit alone accounts for 42,000. Heyn predicted:
“In the future, cars will seamlessly integrate into the digital world, making them primarily one thing: capable of updates. With Bosch technology, cars age more slowly.”
More Safety for Everyone
The current software stack serves assisted and automated driving at Levels 2 and 3. The goal, according to a Bosch press release, is to completely replace these driving stages with Artificial Intelligence. For this purpose, existing approaches are being expanded with state-of-the-art AI methods. Bosch:
“This is expected to lead to more powerful and intelligent driver assistance systems that act as naturally as a human driver — thereby elevating the driving experience to a new level and making it even safer.”
The software package promises providers the implementation of all essential rational human functions like the interpretation of perceptions, decision-making, and actions. The goal is to make automated driving accessible not only for shuttle purposes in public transport or transport services — VW has been involved in such projects with partners like Mobileye or its Hamburg subsidiary Moia for some time — but also for all private drivers: "from volume to premium segment." This allows drivers to take their hands off the wheel in certain
driving situations and engage in entertainment, for example.
Implementations in test fleets have already been made. These are now being "systematically trained and further developed with large amounts of data," according to Bosch. It is anticipated that a software stack will be available for series projects by mid-2026.
From Assistance to Steering Wheel-less Autonomy
Many vehicles already contain Level 1 of the overall 5-stage automated driving: here the driver still controls and is responsible for the operation and is merely supported by certain assistance systems, such as a cruise control to maintain speed or lane-keeping assist systems, which can be turned off. This stage is also called assisted driving.
Semi-automated driving at Level 2 includes Advanced Driver Assistance Systems (ADAS), meaning that two assistance systems are combined, for example, lane-keeping assist and distance control. In this situation, the vehicle can autonomously complete some tasks without human input, such as automatic parking. Variations exist here. The Friedrichshafen-based supplier and mobility expert ZF, for instance, offers Level 2+ systems where various driver assistance systems are "intelligently interconnected," as they put it. Even though this level allows for the temporary release of the steering wheel, responsibility remains with the driver, who must be ready to intervene if necessary. ZF states on its website about this topic: "Writing text messages on the smartphone while driving remains taboo."
With the next Level 3, highly automated driving is achieved: with automatic control and monitoring of traffic, "drivers" may under certain conditions also read the newspaper, as constant monitoring of the traffic situation is no longer required. The driver only needs to intervene when the system signals it in the event of a problem. Currently, the few vehicles of this type — for instance, from Mercedes-Benz — are only allowed to drive up to 60 kilometers per hour in this mode. And the driver can freely use his position for entertainment: movies, video meetings, mobile phone use — all allowed. However, sleeping is not yet allowed, and the driver must still be able to intervene after a warning period.
Only with fully automated driving at Level 4 can the driver transfer full responsibility to the vehicle in certain situations, which then drives "completely autonomously." However, this condition only applies to defined routes, such as on the highway or in parking garages, as ZF explains. After a fully automated phase, the driver may take back control. Tests on this level have been underway for some time. According to ZF, Germany is the first country worldwide to have created the legal framework for vehicles
of this stage "in fixed operation areas in public road traffic during regular operations." In the logistics traffic sector, truck manufacturer MAN, for example, is working with partners in the USA on autonomous trucks that perform driverless transport services — in confined logistics areas.
With Level 5 autonomous driving, there are no more constraints. The vehicle acts entirely independently in all traffic situations, and human intervention is no longer anticipated. Consequently, pedals or a steering wheel are no longer necessary — the human is merely a passenger.
Breaking Old Patterns
The solutions from Bosch and Cariad aim to advance the "Software-defined Vehicle" architecture at VW. To ensure the technology has a broad impact, Bosch intends to make the technology available globally to other manufacturers as a scalable solution. This is also intended to strengthen Germany's economic position and serve as proof of Europe's power in the field of digital solutions, says Peter Bosch, CEO of Cariad:
“We are showing that the German automotive industry masters the key technologies of Artificial Intelligence and Automated Driving itself. With the expertise of our developers and engineers, we ensure an integral part of Europe’s digital sovereignty.”
The main goal is automated driving systems for as many people as possible — "so they can reclaim valuable time in their cars."
Mathias Pillin, CTO at Bosch Mobility, believes that data-driven AI models are indispensable for widespread deployment of automated driving systems at scale. To achieve this, "old mindsets must be broken," and the Automated Driving Alliance demonstrates how such concepts can be successfully implemented.
AI: Doping for the Software
Initially only used for object detection in the "perception" area, AI is now active along the entire software engineering chain, meaning that the system integrates various technical elements such as cameras and radars, assists in decision-making, and implements "the safe automated control of drive, steering, and brakes." These functions are unimaginable without AI, it is said. The provider writes:
“The automated driving functions will in the future be based on a continuous, so-called end-to-end AI architecture, in which all modules become even more powerful and intelligent through the use of AI.”
They rely on cutting-edge technology from adaptive, generative AI applications that infer the likely behavior of road users by analyzing various scenarios. Bosch and Cariad offer an appropriate and modern development environment and comprehensive hardware strategy that allows scalable and future-proof deployment across all vehicle classes.
High Data Protection, Transparency
The development basis is a proprietary source code that is fully technically controlled and combines several functions — from standardized data
protection to security, driving safety — but also transparency, it is stated. Optimizing the source code also allows innovations to be generated quickly and delivered to customers, describes Bosch the development — while also ensuring oversight:
“The developers design the architecture so that decisions and actions of Artificial Intelligence remain safe, traceable, and explainable.”
Imitating Humans
A further function of the software stack is to serve as a basis for integrating multimodal AI approaches: visual and linguistic information is linked together. Vision Language Action (VLA) approaches are used, which can "imitate human logical thinking and action": "Such use enables even more efficient training and a deeper understanding of complex traffic situations," says Bosch. This way, VLA can help detect "hidden risks while driving" and trigger an appropriate reaction.
Large Test Fleet
Bosch assumes that the developed AI stacking can lay a more secure foundation for automated Level 2 and Level 3 driving. Until the series introduction, announced for mid-2026, continuous "enormous amounts of data" will be collected to make the system even safer and more reliable. This can only be ensured with a large test and validation fleet operating in public traffic. For this, the teams from Cariad and Bosch are working on relevant traffic routes in Europe, Japan, and the USA — ultimately, the software stacks are meant to be compatible for worldwide automated use. Bosch describes how this process roughly works: "The development is data-driven, allowing the software to be updated and optimized multiple times a day in the test vehicles with new updates and optimizations of the source code."
The application is currently being used in test vehicles like the ID.Buzz and Audi Q8. The fleet is swiftly being expanded: "this year alone, additional test vehicles in the three-digit range will be equipped with a comprehensive sensor network to collect high-quality data," the statement announces. Based on this foundation, the AI stacks will be further optimized and particularly critical driving situations analyzed, also called "corner cases." The provider expects the AI-supported software solution to result in even safer and more comfortable "assisted and automated driving" for the end customer.
Tests with autonomous ID.Buzz vehicles have been ongoing for several years. The vehicles are packed with computer technology to capture their surroundings seamlessly: according to a report from ADAC about tests with the automated ID.Buzz used in cooperation with Mobileye, for instance, distances are measured using 11 radars based on electromagnetic waves and six Lidars that work with laser light. Additionally, 14 cameras ensure the detection of the environment with objects up to
