Catalog
concept#Architecture#Software Engineering#Platform#Reliability

Autonomous Systems

Architectural concept for systems that make decisions and act autonomously, e.g., in robotics, vehicles, or distributed control systems.

Autonomous systems are technical systems capable of making decisions and acting without continuous human control.
Emerging
High

Classification

  • High
  • Technical
  • Architectural
  • Intermediate

Technical context

Vehicle control systems and CAN busCloud backends for fleet management and updatesMonitoring and incident management tools

Principles & goals

Clear separation of perception, planning, and actuationExplicit safety and fail-safe strategiesVerifiability and explainability of decisions
Build
Enterprise, Domain, Team

Use cases & scenarios

Compromises

  • System failure with safety-critical consequences
  • Misclassification in perception components
  • Insufficient robustness to anomalies or attacks
  • Define operational design domain (ODD) and test cases early
  • Redundancy in critical sensors and pathways
  • Continuous monitoring and structured telemetry

I/O & resources

  • Sensor data (camera, lidar, radar, IMU)
  • Environmental and map data
  • System and operational limits (safety rules)
  • Actuator commands and control directives
  • Status and health reports
  • Telemetry and log data for analysis

Description

Autonomous systems are technical systems capable of making decisions and acting without continuous human control. They combine perception, planning, and actuation to achieve goals in dynamic environments. The concept covers robotics, autonomous vehicles, and distributed control architectures. They impose specific requirements on safety, reliability, and system design.

✔

  • Enable autonomous operation without continuous human control
  • Scale deployments in hazardous or inaccessible environments
  • Continuous optimization via data collection and feedback

✖

  • Dependence on sensor quality and environmental conditions
  • High effort for verification and validation in safety-critical domains
  • Complex integration into existing operational and governance models

  • Mean time to recovery

    Time to restore functionality after a failure.

  • Detection rate of critical events

    Share of correctly detected safety-critical situations.

  • End-to-end system latency

    Delay from sensor input to actuator action.

Waymo

Autonomous driving platform focused on perception, planning and scaled field testing.

Autoware (open source)

Open-source stack for autonomous vehicles used for research and prototypes.

Industrial assembly robots with adaptive control

Robot cells that adapt to part variations and integrate inspection processes.

1

Perform requirements analysis and define operational domain.

2

Design modular architecture with clear interfaces.

3

Build prototypes for core functions and test in controlled scenarios.

4

Establish verification, validation and certification processes.

⚠️ Technical debt & bottlenecks

  • Insufficiently documented interfaces between perception and planning
  • Legacy hardware constrains future upgrades
  • Lack of automated test environment for distributed scenarios
Sensor resolution and detection rangeCompute capacity and power supplyVerification and validation effort
  • Deploying autonomous operation outside specified ODD
  • Extensive automation without emergency fallbacks
  • Minimal testing effort before broad field release
  • Overestimating perception capability under changing conditions
  • Complexity growth due to tight coupling of subsystems
  • Underestimating regulatory and liability requirements
System architecture and real-time systemsSensing, signal processing and perceptionSafety engineering and functional safety
Safety and functional fault toleranceLatency and real-time capabilityScalability and distributability
  • • Regulatory requirements and certifications
  • • Physical limits of sensors and actuators
  • • Network bandwidth and latency-critical communication