explaingit

apolloauto/apollo

★ 26,618C++Audience · developerComplexity · 5/5MaintainedLicenseSetup · hard

TLDR

Baidu's open-source platform for building self-driving car software, handling perception, planning, and vehicle control from sensors to steering.

Mindmap

mindmap
  root((Apollo))
    What it does
      Perceives environment
      Plans routes
      Controls vehicle
      Real-time decisions
    Key components
      Camera processing
      Lidar integration
      Radar fusion
      Drive-by-wire control
    Use cases
      Urban driving
      Highway automation
      Research testing
      Company development
    Tech stack
      C++ core
      Linux servers
      GPU acceleration
      Sensor drivers
    Audience
      Auto companies
      Research teams
      Autonomous tech devs

Things people build with this

USE CASE 1

Build a self-driving car platform by integrating Apollo's perception and planning modules with your vehicle's sensors and hardware.

USE CASE 2

Research autonomous driving behavior in urban and highway scenarios using Apollo's simulation and testing framework.

USE CASE 3

Develop custom perception pipelines for lidar, camera, and radar fusion on top of Apollo's sensor abstraction layer.

USE CASE 4

Prototype autonomous vehicle features like unprotected left turns and narrow street navigation using Apollo's decision-making engine.

Tech stack

C++LinuxCUDAROSPythonProtobuf

Getting it running

Difficulty · hard Time to first run · 1day+

Requires Linux environment, CUDA GPU, ROS installation, multiple C++ dependencies, and complex sensor/vehicle simulation setup.

Apache 2.0, use freely for any purpose including commercial, with attribution and liability disclaimer.

In plain English

Apollo is Baidu's open-source platform for building self-driving car software. It's the full technology stack needed to make a vehicle drive itself, from perceiving the environment with cameras, lidar (laser-based 3D sensors), and radar, to planning routes, making real-time driving decisions, and controlling the vehicle's steering, brakes, and throttle. The platform is designed for automotive companies, research institutions, and developers working on autonomous vehicle technology. It's not a finished product you deploy; it's a development and testing framework that teams build on top of. Think of it as the operating system for a self-driving car, you bring your own vehicle hardware and sensors, and Apollo provides the software architecture that makes them work together. Apollo has evolved through multiple versions from basic GPS waypoint following on closed tracks (version 1.0) to handling complex urban intersections, unprotected left turns, narrow residential streets, and highway driving at speed (versions 5.x and beyond). The platform currently requires serious hardware: a vehicle with drive-by-wire capability (electronic control of steering/brakes), a high-end GPU, and Linux server hardware, plus sensors that cost tens of thousands of dollars. For most non-technical founders and vibe coders, this is far outside the scope of something to deploy directly. Its relevance is more about understanding the landscape: this is one of the most significant open-source contributions to the autonomous driving space, used by dozens of car companies and research teams globally as a starting point for their own self-driving programs.

Copy-paste prompts

Prompt 1
How do I set up Apollo's perception pipeline to process camera and lidar data from my vehicle's sensors?
Prompt 2
Walk me through Apollo's planning and decision-making modules, how does it decide when to turn or brake?
Prompt 3
What hardware and sensors do I need to run Apollo on a real vehicle, and what are the minimum specs?
Prompt 4
Show me how to use Apollo's simulation environment to test autonomous driving behavior before deploying to a real car.
Prompt 5
How does Apollo handle sensor fusion across cameras, lidar, and radar to build a 3D model of the road?
Open on GitHub → Explain another repo

Generated 2026-05-18 · Model: sonnet-4-6 · Verify against the repo before relying on details.