Tin DDS
Predictable, real-time messaging for mission-critical robotics.
Robots can't afford dropped packets, jittery control loops, or creeping memory bloat. Tin DDS is a high-performance C++ publish-subscribe library built for real-time motion control and distributed robotic platforms.
Processes on the same machine communicate automatically over zero-copy shared memory, while distributed nodes run over standard RTPS/UDP. ROS is not required; a native ROS 2 RMW adapter is ready when you need it.
Engineered for mission-critical reliability
- Zero-copy shared memory: Automatic same-host IPC with microsecond latencies, falling back seamlessly to UDP across network boundaries.
- Deterministic memory bounds: Fixed history queues and bounded allocation prevent silent memory growth and watchdog timeouts.
- Robust discovery & recovery: Transparent fragmentation, retransmission, and repair for high-throughput sensor streams and video.
- State replay for late joiners: Durable topics and persistent history allow newly started nodes to catch up instantly.
- Fine-grained QoS & filtering: Content-filtered topics, keyed instances, and lifecycle events put you in complete control of bandwidth.
- Dual-path teleoperation: Authenticated command admission and independent fail-safe supervision for teleoperated robots.
Security
Authenticated identity & encryption
Secure your robot's data plane with x509 certificate validation, signed governance policies, PKI-DH handshake material, and authenticated AES-GCM wire encryption.
Teleoperation
Fail-safe remote control supervision
Accept commands strictly from authenticated operators with active control authority. Stale, out-of-order, or forged commands are rejected, and failures are reported over two independent communication paths.
Architecture
Predictable performance under pressure
Automatic transport switching
Use the exact same clean C++ reader and writer APIs everywhere. Tin DDS automatically routes same-host traffic through shared memory and remote traffic over UDP.
Resilient over lossy networks
Bounded writer histories and NACK-driven repair recover lost samples so delivery stays reliable. Oversized samples are cleanly fragmented and reassembled.
Persistent durability & crash recovery
Preserve critical device state across application restarts. Corrupt or oversized history files are safely rejected, protecting nodes from starting in invalid states.
No silent overload
Every queue, buffer, and discovery cache has an explicit ceiling. If a limit is hit, your code gets an immediate, actionable error instead of creeping memory consumption.
Openly benchmarked performance
Don't take vendor numbers at face value. Our open benchmark suite records topology, payload, delivery counters, and latency so every number can be checked.
Explore the DDS benchmark suite →Ready to test Tin DDS on your hardware? Contact engineering@tinverse.com.