BuildBB-0022

An Ackermann-steered ROS 2 robot with an LLM agent and voice commands

Seif Salama built an Ackermann-steered robot that runs one ROS 2 Humble stack in Gazebo Harmonic and on a Raspberry Pi car. A custom ros2_control interface drives the motors over UART and the steering servo over I2C. A Gemini or Ollama agent takes spoken commands. We could not tell whether the README video shows the physical robot.

Open original source github.com · Seif Salama

Claimed
Why this evidence label?

README embeds a video; we could not confirm whether it shows the physical robot or the simulation.

How we check entries
Preview image for An Ackermann-steered ROS 2 robot with an LLM agent and voice commands
The source's own preview image, from github.com.

Briefing

Wheel odometry and the IMU feed an EKF in robot_localization. SLAM Toolbox maps new spaces and AMCL localizes on saved maps, while Nav2 plans with SmacPlannerHybrid on Reeds-Shepp paths and follows them with an MPPI controller set for Ackermann motion. The physical robot carries an RPLidar, and Grounding DINO object search from plain-language queries is listed as hardware-only.

The agent, the Whisper voice front end and Grounding DINO run on a separate dev machine; the Pi build excludes the agent and simulation packages, and Grounding DINO needs a GPU. The browser UI must be served over HTTPS so the microphone works across the LAN. The agent's default cloud model is gemma-3-27b-it, and the Docker images are marked as not yet implemented.

From the GitHub API

Repository facts

Repository facts for sseifsalama/ackermann_robot, from the GitHub API
Repositorysseifsalama/ackermann_robot
Stars8
LicenseApache-2.0 Permissive
LanguagePython
Last push6 May 2026 5 mo ago

Tags

Report a problem with this entry