尚未检测到更新。GitHub 发布版本后会显示在这里。
按最近已完成的 8 小时观测回合间的 GitHub 星标增量排序
Dive deep into your data. A fast, native, multi-engine database client written in Rust. Free and open source, no telemetry.
翻译准备中
制作者的介绍仅有上面的一句话。 更多详情请阅读仓库 README。 仓库 README ↗
正在加载完整 README。
查看完整 README ↗Modular Intelligent Robot
MIR is an open-source modular robotics platform for education, research and experimentation.
Website, tutorials and course: mir-edu.fr (in French only for now)
Designed around interchangeable hardware modules, MIR makes it possible to build different robots while sharing a common hardware and software architecture. The platform is compatible with the LeRobot ecosystem for robot learning, while remaining suitable for more traditional robotics applications such as kinematics, control, teleoperation, perception and navigation.
The same hardware modules can be used to build different robotic systems, including :
Traditional educational robots are often designed for a single purpose. MIR follows a different approach : the same hardware modules can be reused to build different robotic systems. This allows students, teachers and researchers to explore a wide range of robotics topics without redesigning the platform.
This repository documents the development of MIR from its first functional prototype.
The MIR application lets you build a robot from its modules, simulate it (MuJoCo), drive it, record demonstrations, and train models (behavior cloning, LeRobot policies such as ACT, reinforcement learning with Stable-Baselines3), on the real robot or in simulation.
| Folder | Content |
|---|---|
mir_robot_editor/ | the MIR application (user interface) |
mir_robot/ | mirRobot, the single entry point to the robot (real or simulated), training and policies |
mir_sim/ | module catalog, MuJoCo model of the robot, simulated devices |
mir_devices/ | drivers for the real hardware (Feetech servo bus, cameras, ESP32, lidar) |
robots/ | example robots: configuration, scenes, scripts, training scripts |
doc/ | documentation (architecture, installation, hardware) |
cad/, assets/ | 3D models (STL) and pictures of the modules |
MIR runs on Windows (10 or 11, 64-bit), macOS (Apple Silicon) and Linux (64-bit). On each system, a single command installs it. No administrator rights are needed, and you don't have to install Python yourself: the uv tool downloads Python 3.12 and the exact versions of every package MIR needs (LeRobot, MuJoCo, PyTorch...).
You need an internet connection and about 5 GB of free disk space. The first installation takes around 10 minutes.
Open PowerShell (Start menu → type "PowerShell"), paste this line and press Enter:
powershell -ExecutionPolicy ByPass -c "irm https://raw.githubusercontent.com/MariusBrtn/mir/main/installer/windows/install_mir.ps1 | iex"
It installs uv, downloads MIR into your user folder (%USERPROFILE%\MIR), installs its packages and creates a MIR
shortcut in the Start menu and on the desktop. It starts with two simulated robots: bras_4ddl_cam_V2 (4-DOF arm with
a camera) and test_mobile (mobile robot with a lidar). The log is in %USERPROFILE%\MIR\logs\application.log. Run
the same line again to update MIR: the robots\ folder (your work) is never touched. Windows 7 and 8 are not
supported (Python 3.12, Qt 6 and PyTorch need Windows 10 or later).
Open the Terminal app (Spotlight → type "Terminal"), paste this line and press Enter:
curl -LsSf https://raw.githubusercontent.com/MariusBrtn/mir/main/installer/macos/install_mir.sh | bash
It installs uv, downloads MIR into your home folder (~/MIR), installs its packages and creates the MIR app in
the Applications folder (or in ~/Applications if your account may not write there). Find it with Launchpad or
Spotlight, and drag it to the Dock to keep it. It starts with the same two simulated robots. Run the same line again to update MIR: the robots/ folder (your work) is never touched. Intel Macs are not supported (PyTorch no longer exists for them).
Open a Terminal, paste this line and press Enter:
curl -LsSf https://raw.githubusercontent.com/MariusBrtn/mir/main/installer/linux/install_mir.sh | bash
If curl is missing (as on a fresh Ubuntu), use wget instead:
wget -qO- https://raw.githubusercontent.com/MariusBrtn/mir/main/installer/linux/install_mir.sh | bash
It installs uv, downloads MIR into your home folder (~/MIR), installs its packages and adds MIR to the
applications menu and to the desktop, with the same two simulated robots. On an Intel / AMD PC, plan for about
12 GB of free space: PyTorch for Linux comes with CUDA (NVIDIA graphics cards). Near the end, the installer offers
two optional steps, which need your password (sudo):
libxcb-cursor0, a library Qt needs to open MIR's window on Ubuntu / Debian;dialout group, allowed to use the serial port of a real robot (log out and back in
afterwards).Answer n if you have no administrator rights: MIR works in simulation, and the installer prints the commands for an
administrator. The log is in ~/MIR/logs/application.log. Run the same line again to update MIR (your robots/ folder
is never touched). To install elsewhere, put MIR_DIR=/path/to/MIR before bash, and use it again for each update.
The steps below do the same by hand, on Windows, macOS and Linux (useful to work on MIR's code).
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
curl -LsSf https://astral.sh/uv/install.sh | sh
Then close and reopen the PowerShell / Terminal window, so that the uv command is found.
以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗