Carsmos: Open-Source Smart Mobility

DORA - Dataflow Oriented Robotics Architecture

DORA is an advanced open source autonomous driving and robot middleware framework and data flow computing platform. The DORA project takes software-defined cars and intelligent driving as the target scenarios, integrates a large number of cloud-native architecture concepts and ecological tools, and adopts the safe and reliable Rust language development, supports elastic data flow computing framework, diversified middleware selection Cloud collaborative computing provides another more modern platform choice for the software development of smart cars, and also provides a technical route for upgrading, replacing and surpassing ROS1/ROS2.

Technical value

Provides an autonomous, controllable, open-source, next-generation device-edge-cloud collaborative data flow computing platform that can support innovative autonomous driving and robot applications. The technology is ahead of the ROS architecture. Based on Rust language development, it outperforms C/C++ in terms of development speed, development quality, safety and reliability, and meets the future needs of robots and new energy vehicles.

Business value

Various business forms that meet the needs of robots and autonomous driving, including bicycles, fleets, cloud-edge-device collaboration, etc.

Ecological value

Relying on the world's largest new energy vehicle and robotics market, we will build an open source ecosystem with global influence beyond ROS through the joint production, education and research.

OASIS-smart driving simulation platform and scene description language

The simulation test platform and its ecology developed by OASIS in the form of open source are expected to connect the three parties of scientific research, supervision and industry, promote the growth and transformation of intelligent driving R&D enterprises. OASIS Sim is positioned as an open simulation platform, which realizes the key functions of simulation in the form of minimal organization. OASIS Sim and other open source projects in Carsmos form a complete intelligent driving research and development tool chain, which lowers the research and development threshold and improves the freedom and maintainability of research and development. OASIS DSL is a set of formal domain-specific language for dynamic traffic scene description,which has the advantages of the above formal language. The OASIS DSL open source project provides syntax design,domain model and language implementation on the OASIS Sim simulation platform, including an interactive scene editor based on OASIS DSL.

Technical value

Self-designed scene language and compiler, one of only two open source scene languages in the world. Compatible with OpenScenario 1 & 2 semantics, and supports OpenScenario's simulation ecology. It integrates formal rule description and testing, scenario description and testing. The web-based scene library IDE based on the vena scene language and the scene simulation environment deployed in cloud native mode support the closed loop from scene construction to scene test. The scenario language Spec of the Apache 2.0 license is compatible with the OpenScenario 2.0 standard and allows third-party language implementations.

Business value

Based on the implementation of open source scene language, complete open source simulation environment docking, open source scene library design IDE and deployment tools, to meet the characteristics of research, supervision, industry common use and evolution scene language.

Ecological value

Through the national standard scene description language and scene library, an autonomous and controllable robot and automatic driving simulation test ecology is formed.

Etymos:Data Acquisition + Data Flywheel

EtymOS supports the autonomous driving 2.0 project around data and models, including the construction of standardized data collection EtymosKIT, data set sharing and community collaborative development through the automated data flywheel EtymOS Wheel, to quickly improve the performance of autonomous driving models and create standardized self-driving perception Large model EtymOS.

Technical value

Cloud-native low-code toolchains and open source ODD applications can quickly launch innovative applications, lowering the threshold for development and innovation.

Business value

Pull together the research platform, education platform, and industrial cooperation platform of robotics and autonomous driving, cultivate professional talents, and realize the alignment between research and industry.

Ecological value

Open research introduces the funding and participation of enterprises, promotes the business model of the open source ecosystem, and attracts and injects fresh development and research forces into the industry.

ESSA - Elastic Stateful Serverless Computing Architecture

The ESSA (Elastic Stateful Serverless Computing) project, the ESSA elastic stateful serverless computing architecture, is an accelerator technology for edge-end cloud high-performance computing, mainly including two sub-projects:

Anna-RS

This is a low-latency, auto-scaling (Auto Scaling) and efficient distributed key-value (K/V) database that supports geographical distribution. It uses Anna-RS K/V to support the state management required by the ESSA elastic serverless computing architecture.

Essa-RS

This is a serverless function runtime environment based on WebAssembly. Compared with other serverless function runtime environments, WebAssembly has fast speed, multi-language (C/C++, Rust and other languages) support, very safe sandbox isolation, and less Advantages such as function call overhead.

DORA - Dataflow Oriented Robotics Architecture

EDIT EV open source project is a skateboard chassis design for electric vehicles, in line with the European L7E standard (heavy-duty four-wheeled motor vehicles, dry weight of passengers <400kg, dry weight of loads <550kg, if the electric vehicle does not include battery weight, the maximum net power <15kw ) and China's M1 standard (passenger cars with at least 4 or 3 wheels, the maximum total mass determined by the factory does not exceed 1 ton, and no more than 8 passenger seats except the driver's seat).

Technical value

The separated architecture design can continuously optimize computing and software platforms and optimize component costs.

Business value

Through the open source battery compartment design, fast and low-cost battery replacement services are realized. Support vehicle sales model, Battery-As-A-Service model, Mobility-As-A-Service model.

Ecological value

Build a standard hardware platform through open source, and form a transparent and secure software ecosystem with open source software. Through open source, realize component standards and form a global supply chain.

AutoExpress: D2D interconnection interface in Chiplet mode for agile development of chip modularization

The AutoExpress project is a D2D interconnection interface in the Chiplet mode for agile development of chip modularization, which satisfies multi-chip short-distance, high-bandwidth and low-latency interconnection, and has a car-level design for the automotive field.

Technical value

Make up for the lack of a unified and efficient Die-to-Die interconnection interface in the current domestic Chiplet chip development mode, and avoid repeated redundant investment in the industry.

Business value

By realizing the standardization of the D2D interface, it is beneficial to promote the development of Chiplet technology, which will bring direct economic benefits to the upstream and downstream enterprises of China's semiconductor industry chain.

Ecological value

By realizing the standardization of D2D interface, chiplet technology will be more widely used in the entire industry chain; new chiplet technology will give birth to new business models, break the traditional SOC design model, and reconstruct the supply chain market.