软件定义汽车-几何伙伴

Self-Developed Middleware

Multi-Platform Support

Compatible with QNX / Linux / Android / Windows / seL4 systems and supports multi-domain controller management (including cockpit-drive fusion platform).

High Performance & Low Latency

Advanced QoS strategies with zero-copy architecture, optimized for high-load, low-latency application scenarios.

Simplified API & Unified Interface

General interfaces with minimalist design, masking protocol discrepancies to minimize learning/implementation cost.

Multi-Protocol & Serialization Support

Supporting transport protocols including FastDDS, CycloneDDS, RTIDDS, SomeIP, Iceoryx and Zenoh, and serialization including Protobuf, OMG IDL, Flatbuffers, Capnproto and FIDL.

Advanced Message Scheduling

Supporting deterministic scheduling, taskflow orchestration, nanosecond-precision timers and coroutines.

Comprehensive Debugging Toolchain

Supporting data recording/refilling, real-time monitoring, data instrumentation, gateway forwarding, analytic and visualization tools.

Simulation platform

Software and Hardware System Testing

Hardware in loop simulation is used for software deployment and software/hardware integration phase verification, which can cover software and hardware testing related to the interaction of domain controllers and their associated controllers.

Closed-Loop Algorithm Testing

Supporting basic software functional testing (communication, diagnostics, fault injection, etc.) and closed-loop algorithm logic testing (perception-prediction-decision-planning&control, etc.).

Accelerated Development Iteration

Software in loop simulation, combined with cloud-based CI-CD-CT, can quickly verify logical strategies during algorithm development and accelerate development, debugging, and iteration speed.

Cloud Based SIL

Using the advantages of high concurrency and large computing power in the cloud, the daily simulation mileage exceeds 5 million kilometers, and the scene category exceeds 1000 categories.

Data Platform

AI-accelerated Data Closed Loop

Enhanced Data Closed Loop 2.0 platform utilizes AI large models for accelerating data mining, automatic annotation, model training and simulation testing, enhancing the scale and throughput efficiency of data.

Full-process Automation

Enhancements: NeRF-based 3D-scene reconstruction; automated data lebelling and classification; data searching engine with large model (image to image/text to image/text to video clip); automatic annotation with large model.