Enabling Autonomous Cars
Written by Bob Leigh
January 27, 2016
An autonomous car is a great example of a highly distributed dynamic system, where component objects continuously make real-time local decisions based on system-wide constraints and approximate global state. DDS evolved to specifically address this type of system, and RTI has become a trusted expert assisting the innovators of future autonomous cars.
The ease of integration and flexible, reliable, and fast publish-subscribe data model of the RTI Connext DDS middleware are uniquely suited to addressing many of the toughest challenges posed by autonomous cars:
Vehicle subsystem integration and control, spanning driving control, safety, infotainment, and diagnostic functions
Inter-vehicle interactions, for collision avoidance and optimized travel experiences
Tracking and control functions, for fleet management, traffic monitoring and management, crisis management, and government agency coordination
Sensor and camera data aggregation at millisecond speeds
Local and remote feedback loops
Reliable communications over unreliable channels (for example, wireless, cellular)
Ability to operate within redundant environments (intelligently delivering only one copy of data)
Rapid time to market for safety-certifiable infrastructure, using RTI Connext DDS Cert DDS within connected vehicle architecture
DDS Map to Autonomous Car Requirements
Unlike other connectivity middleware, DDS emerged more than 10 years ago to address physics-speed connectivity requirements. Today, DDS remains the only middleware capable of satisfying the most stringent requirements including:
- Reliability. Within an autonomous car, even five milliseconds of downtime can be a disaster. DDS implements natural redundancy to ensure continued operation.
- Performance. For the system components that need millisecond or microsecond response, DDS provides fast peer-to-peer
- Integration at scale. Autonomous cars integrate many applications and deal with thousands of addressable data items during normal operation. Data-centric DDS eases complex data flow within these types of large-scale systems.
To minimize overhead, the DDS publish-subscribe model delivers:
- Fine control of quality of service (QoS) parameters including reliability, bandwidth control, delivery deadlines, liveliness status, resource limits, and security
- Explicitly managed communications data model, with a choice of connection types
- Data centricity, with inherent understanding about the contents of the information being managed and shared
- Inherent automation (no hard-coded interactions between applications and devices)
- Device discovery (easy add-on of new devices without any configuration changes required)
Compared to traditional point-to-point communications, DDS offers a superior databus with plug-and-play simplicity, scalability, and an architecture that can evolve while maintaining exceptional performance levels. Scalability and integration capacity of DDS are also instrumental in enabling a car’s connections with other vehicles and their own environments, including external systems such as traffic monitoring.