Access Official Technical PDF

Share Summary:LinkedInEmail

The autonomous driving market is shaped by a systems challenge familiar to industrial automation: coordinating sensors, computing hardware, software, communications and human oversight under demanding operating conditions. A 2018 market outlook identified several connected priorities, including sensor fusion, artificial intelligence (AI), centralized vehicle computing, driver monitoring and fleet services. For engineers and technical decision-makers, the key lesson is that higher automation depends on integrating these elements rather than advancing any single component in isolation.

The outlook’s market forecasts and timelines are historical projections made in 2018, not current estimates. Its technical themes, however, show how the industry was thinking about the path from driver assistance toward more automated operation.

Sensor fusion: combining different views of the road

Automated driving systems use sensors to detect objects, estimate their movement and support decisions such as braking, steering or route planning. Cameras, radar, LiDAR, ultrasonic sensors and positioning systems provide different kinds of information. Combining their outputs—sensor fusion—can help a vehicle build a more complete picture of its surroundings than relying on one sensor type alone.

The 2018 analysis described camera and long-range radar combinations as a common approach for existing driver-assistance applications, in part because of their cost advantages. It also anticipated larger sensor suites as automation levels increased. Its illustrative configurations included multiple cameras and radars, with LiDAR, thermal sensing and dead reckoning appearing in more advanced configurations. These counts were estimates, not universal design requirements.

Sensor selection is only part of the engineering task. Systems must interpret what sensors detect, distinguish relevant objects and support path planning. The outlook highlighted depth perception, object classification and performance in poor weather or at night as continuing development priorities. That makes validation across varied conditions important when assessing an automated-driving system.

AI and domain controllers reshape vehicle computing

Rule-based software can be difficult to scale across the many situations a vehicle may encounter. The report pointed to AI and machine-learning methods as tools for perception, behavior prediction, mapping, simulation and testing. It discussed reinforcement learning as one approach under investigation, where a system learns from repeated feedback. This should not be confused with proof that a vehicle can handle every real-world situation; the report treated AI development and validation as ongoing work.

Computing architecture was another central issue. Adding a separate electronic control unit (ECU) for every new feature can increase wiring, cost and system complexity. The proposed direction was to consolidate functions through domain controllers: computing units that bring related vehicle functions together. The report presented this as an intermediate step toward more centralized architectures, with the potential to support faster communication for time-critical processing.

For system integrators, consolidation changes the integration problem. Functions that once operated in separate controllers must work within a shared architecture, while safety-related tasks still need appropriate monitoring. The Audi zFAS example in the outlook illustrated a multi-supplier platform combining processing components, sensors, communications and safety supervision. It underscored the coordination required to bring hardware and software from different providers into one vehicle system.

Driver monitoring remains part of automated operation

Automation does not immediately remove the human operator from the control loop. The report identified driver monitoring systems (DMS) as important for transitions between automated and manual control, particularly for Level 3 systems where a driver may be required to resume control. Camera-based systems can monitor gaze, eye movement and attention; other approaches use sensors in components such as the steering wheel or seat.

Each approach presents integration considerations. Cameras must function under different lighting conditions and accommodate driver variation. Biological sensors bring their own placement and cost challenges. The operational requirement is to assess whether the driver is available to take over and to communicate when attention is needed. Human-machine interface design therefore belongs alongside perception and vehicle control in the system design.

OTA updates and cybersecurity extend the system boundary

The 2018 outlook anticipated vehicles equipped with hardware intended to support future functions, with some capabilities delivered or adapted through over-the-air (OTA) software updates. It described a potential shift from a one-time hardware sale toward offering features and services over a vehicle’s lifetime. The report also noted that options could vary by region and legislation, making configuration and update management relevant to deployment.

Connectivity also expands the security boundary. The analysis identified risks involving vehicle networks, telematics data, software updates and links between vehicles and infrastructure. A weakness in a connected system could affect more than one component, particularly where vehicles are operated as a fleet. Security planning therefore needs to account for the vehicle’s electronic architecture and the data paths used to communicate with external services.

Shared mobility makes fleet management a core capability

Autonomous shared mobility was a major commercial theme in the outlook. Rather than focusing only on privately owned vehicles, it examined taxis and shuttles operating in defined areas, such as campuses or other private sites. These applications depend on more than an automated vehicle: operators also need ways to receive service requests, assign vehicles, plan routes and monitor operations.

The report described fleet management as a cycle of tracking, optimizing and controlling vehicles. A cloud platform could combine vehicle telemetry with schedules, traffic information, weather and demand data to coordinate a fleet. This creates an integration task across vehicle hardware, software, communications and operating procedures. It also explains why the report expected partnerships between vehicle manufacturers and technology or mobility-service providers.

Practical implications for technical decision-makers

The 2018 outlook points to several practical considerations for organizations evaluating automated systems:

  • Assess the whole sensing chain: consider sensor coverage, data fusion and performance across operating conditions.
  • Plan computing architecture deliberately: evaluate how domain controllers and centralized processing affect integration and system complexity.
  • Include the human interface: define how driver attention is monitored and how control transitions are communicated.
  • Design for lifecycle management: account for software updates, cybersecurity and region-specific feature configurations.
  • Connect vehicle capability to operations: fleet applications require dispatch, tracking, routing and service-management functions alongside vehicle automation.

Autonomous driving is not a standalone vehicle feature; it is an integrated automation ecosystem. The historical outlook emphasized that progress depends on aligning sensing, AI, computing architecture, driver oversight, cybersecurity and fleet operations. For technical decision-makers, evaluating those interfaces—and the requirements they create across the system lifecycle—is as important as evaluating any individual sensor or algorithm.

Access Official Technical PDF

About The Author