# Apollo (autonomous driving)

Apollo is Baidu's open-source autonomous driving platform, providing a comprehensive suite of software, hardware, and cloud services for developing self-driving vehicles, first released in 2017.

Apollo is an open-source autonomous driving platform developed by Baidu, first released in April 2017. It provides a comprehensive suite of software, hardware, and cloud services that enable developers and automakers to build and deploy self-driving vehicles. The platform is designed to be modular and scalable, supporting a range of autonomous driving levels from advanced driver assistance to fully autonomous operation.

Apollo's architecture is built around a cloud-to-car approach, integrating high-definition maps, localization, perception, planning, control, and vehicle-to-everything (V2X) communication. It leverages [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) and [machine-learning](https://www.wikiprompt.org/wiki/machine-learning) techniques, including [deep-learning](https://www.wikiprompt.org/wiki/deep-learning) and [neural-network](https://www.wikiprompt.org/wiki/neural-network) models, to process sensor data and make real-time driving decisions. The platform is continuously updated with contributions from a global community of developers and partners.

## History and Development

Baidu announced Apollo at the 2017 Shanghai Auto Show, positioning it as an "Android of the autonomous driving industry." The first version, Apollo 1.0, was released in July 2017, offering basic capabilities such as high-definition map data and self-localization. Subsequent releases added features like perception, planning, and control modules, with Apollo 2.0 in January 2018 enabling simple urban driving scenarios.

In 2018, Apollo 3.0 introduced the platform's first production-ready autonomous driving solution for commercial use, including automated valet parking and low-speed shuttle services. Apollo 4.0 (2019) focused on urban road driving, while Apollo 5.0 (2020) added advanced safety features and improved simulation tools. Apollo 6.0 (2021) enhanced the platform's scalability and introduced a new vision-based perception system.

Apollo 7.0 (2022) brought significant improvements in [transformer](https://www.wikiprompt.org/wiki/transformer)-based perception models, enabling better handling of complex traffic scenarios. Apollo 8.0 (2023) introduced a new architecture that separates the software stack into modular components, making it easier for developers to customize and extend. As of 2024, Apollo continues to evolve with a focus on [generative-ai](https://www.wikiprompt.org/wiki/generative-ai) and [large-language-model](https://www.wikiprompt.org/wiki/large-language-model) integration for natural language interaction and decision-making.

## Architecture and Components

Apollo's architecture is organized into several layers: the vehicle layer, the hardware layer, the software layer, and the cloud layer. The vehicle layer includes the physical car and its sensors, such as cameras, LiDAR, radar, and GPS. The hardware layer consists of computing units, often using [nvidia](https://www.wikiprompt.org/wiki/nvidia)-based GPUs, that run the software stack. The software layer is the core of Apollo, comprising modules for perception, prediction, planning, control, and localization.

The perception module uses [deep-learning](https://www.wikiprompt.org/wiki/deep-learning) models, including [residual-network](https://www.wikiprompt.org/wiki/residual-network) and [u-net](https://www.wikiprompt.org/wiki/u-net) architectures, to detect objects, lane lines, and traffic signs. The prediction module forecasts the future trajectories of other road users, while the planning module generates safe and efficient driving paths. The control module executes these plans by sending commands to the vehicle's actuators.

The cloud layer provides services such as high-definition map updates, simulation, and over-the-air software updates. Apollo's cloud platform also offers a data pipeline for collecting and analyzing real-world driving data, which is used to improve the system's performance.

## Open Source and Community

Apollo is released under the Apache 2.0 license, allowing free use and modification. The project is hosted on GitHub, where it has attracted thousands of contributors from companies, universities, and individual developers. Baidu also maintains a dedicated developer portal with documentation, tutorials, and forums.

The Apollo community includes partners such as [tomtom](https://www.wikiprompt.org/wiki/tomtom) for mapping, [intel](https://www.wikiprompt.org/wiki/intel) for hardware, and various automotive manufacturers. Baidu has established Apollo Innovation Centers in multiple cities to foster collaboration and testing. The platform has been used in numerous pilot projects, including autonomous taxis in Beijing and Wuhan.

## Applications and Partnerships

Apollo has been deployed in a variety of applications, including robotaxis, autonomous shuttles, and logistics vehicles. Baidu's own autonomous ride-hailing service, Apollo Go, operates in several Chinese cities, with over 5 million rides completed by 2023. The platform is also used by other companies, such as [waymo](https://www.wikiprompt.org/wiki/waymo) and [tesla-autopilot](https://www.wikiprompt.org/wiki/tesla-autopilot) competitors, to develop their own autonomous driving systems.

Baidu has partnered with automakers like Geely, Ford, and Hyundai to integrate Apollo into production vehicles. In 2021, Baidu announced a joint venture with Geely to produce electric vehicles under the Jidu brand, which uses Apollo's technology. Apollo also collaborates with [alibaba-cloud](https://www.wikiprompt.org/wiki/alibaba-cloud) and [amazon-web-services](https://www.wikiprompt.org/wiki/amazon-web-services) for cloud infrastructure and with [qualcomm](https://www.wikiprompt.org/wiki/qualcomm) and [arm-holdings](https://www.wikiprompt.org/wiki/arm-holdings) for edge computing solutions.

## Safety and Regulation

Safety is a critical aspect of Apollo's development. The platform includes redundant systems for critical components, such as braking and steering, and uses a fail-safe mechanism to bring the vehicle to a safe stop in case of system failure. Apollo also incorporates a simulation environment that allows extensive testing before real-world deployment.

Baidu has worked with regulatory bodies in China to establish guidelines for autonomous vehicle testing and deployment. Apollo vehicles have obtained licenses for road testing in multiple cities, and Apollo Go has received approval for commercial operation in certain areas. The platform's safety record has been positive, with no major accidents reported as of 2024.

## Impact and Future Directions

Apollo has significantly influenced the autonomous driving industry by democratizing access to advanced technology. Its open-source model has accelerated innovation and reduced development costs for many companies. Apollo's contributions to [deep-learning](https://www.wikiprompt.org/wiki/deep-learning) and [computer vision](https://www.wikiprompt.org/wiki/computer-vision) have also advanced the broader field of [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence).

Looking forward, Apollo aims to integrate more [large-language-model](https://www.wikiprompt.org/wiki/large-language-model) capabilities to enable natural language interaction with passengers and improve decision-making in complex scenarios. The platform is also exploring the use of [generative-ai](https://www.wikiprompt.org/wiki/generative-ai) for synthetic data generation to enhance training datasets. As autonomous driving technology matures, Apollo is positioned to play a key role in the transition to safer and more efficient transportation systems.

## See Also

- [waymo](https://www.wikiprompt.org/wiki/waymo)
- [tesla-autopilot](https://www.wikiprompt.org/wiki/tesla-autopilot)
- [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence)
- [deep-learning](https://www.wikiprompt.org/wiki/deep-learning)
- [machine-learning](https://www.wikiprompt.org/wiki/machine-learning)

---
Source: https://www.wikiprompt.org/wiki/baidu-apollo
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-09-09T01:54:47.73402+00:00