AVLite

Autonomy, made lite

AVLite is a lightweight, modular autonomous-vehicle stack for perception, planning, and control — from a 2D simulator on your laptop to headless deployment on a real robot.

pip install avlite   # install
avlite               # launch the visualizer

Get Started Overview GitHub

BasicSim included  ·  CARLA / Gazebo / ROS2 ready  ·  GUI + headless  ·  Pure Python

•   Quick Start

  ---

  Install, launch the visualizer, and drive the built-in simulator in minutes.

  Get started

•   Architecture

  ---

  Layered strategy pattern with auto-registration and a capability system.

  How it fits together

•   Algorithms

  ---

  Global and local planning, including a greedy Frenet lattice planner.

  Planning internals

•   Plugin System

  ---

  Add perception, planning, control, or world-bridge strategies as plugins.

  Build a plugin

•   Configuration

  ---

  YAML profiles with schema validation, tooltips, and import/export.

  Settings naming

•   Full Overview

  ---

  Features, installation, components, and configuration in one place.

  Read the overview

Real-time Tk visualizer: live plots, per-layer tuning, and profile management.

Why AVLite

•   Lightweight

  ---

  The core stack needs only NumPy, Matplotlib, SciPy, Shapely, NetworkX, and Pydantic. No middleware lock-in.

•   Modular

  ---

  Swap perception, localization, planning, and control strategies at runtime — they auto-register and appear in the UI.

•   Sim to robot

  ---

  The same YAML profile drives the GUI and headless mode, so what you see in the visualizer is what the robot runs.

•   Multi-simulator

  ---

  BasicSim ships built in; CARLA, Gazebo, and ROS2 plug in through optional world-bridge plugins.

Ready to drive?

pip install avlite

Get Started Browse the docs