Open-source alternatives to SO ARM100

Legged Gym is a high-performance simulation platform and toolkit engineered for training autonomous robotic agents in complex, physics-based environments. It provides a comprehensive framework for developing legged locomotion control policies, enabling robots to learn movement strategies for navigating uneven terrain and managing physical disturbances through reinforcement learning. The platform distinguishes itself by utilizing hardware-accelerated physics and headless execution to maximize computational throughput during training. It incorporates a domain randomization pipeline that injects

AgiBot-World is a suite of software pipelines and tools designed for robotic policy training, dataset standardization, embodiment transfer, and performance benchmarking. It provides infrastructure for developing bimanual manipulation policies using foundation models and human-reference trajectory data. The project features a robot embodiment transfer suite that adapts pre-trained models to different robot bodies without requiring new multi-embodiment training data. It also includes a specialized evaluation framework for validating vision-language-action models through open-loop testing and ph

OpenCat-Quadruped-Robot is a development framework and motion control API for building four-legged robots. It provides a comprehensive environment for quadruped robot development, featuring tools for locomotion gait design, inverse kinematics, and a layered control architecture that separates high-level intelligence from low-level motion. The project distinguishes itself as an embedded AI deployment tool, allowing users to train, quantize, and deploy machine learning models to vision modules for tasks such as object detection and visual target tracking. It further supports complex human-robot

ElectronBot is a robotic system comprising custom PCB designs, an embedded circular display driver, an I2C servo controller, and a communication bridge. It integrates hardware schematics and layouts for control boards and sensor arrays with firmware designed to render images and animations on integrated circular screens. The system features a control bridge that links the physical robot hardware to the Unity game engine for real-time visualization and high-level control. Data exchange between the host computer and the embedded hardware is handled via a custom USB CDC serial communication prot

This project is a Java library for establishing raw serial port connections between Android devices and external hardware using USB host mode. It provides a communication bridge for interacting with microcontrollers and adapters without requiring root access. The library includes a device identification framework to detect and map serial hardware using vendor and product IDs. It features specialized driver implementations for chipsets such as CDC and FTDI to enable data exchange across different hardware standards. It covers hardware integration capabilities including device discovery, seria

Racer is a self-driving remote-controlled car platform that integrates hardware schematics, an autonomous vehicle training framework, and a machine learning control system. It provides the necessary tools to build, train, and operate autonomous vehicles. The platform includes 3D-printable design files and electronic schematics for constructing custom vehicle chassis and circuitry. This hardware is supported by a software toolkit that translates model predictions into steering and speed commands for the physical vehicle. The system covers supervised learning processes that map sensor data to

Video-Pre-Training is a machine learning framework designed for training autonomous agents to perform complex tasks by observing and mimicking human behavior from video recordings. It provides a comprehensive toolkit for imitation learning and reinforcement learning research, enabling the development of agents that can replicate human actions within simulated digital environments. The framework distinguishes itself through its ability to process large-scale, unlabeled video datasets to bootstrap agent capabilities. It utilizes inverse dynamics modeling to infer control inputs from frame trans

Practical_RL is a comprehensive educational curriculum and course for learning to design and implement agents that solve complex decision processes. It provides a structured study program covering the fundamentals of reinforcement learning, from basic trial-and-error behavior to advanced deep reinforcement learning. The project includes specialized guides and frameworks for imitation learning based on expert demonstrations, model-based reinforcement learning using planners, and the training of recurrent neural networks to solve partially observed environments. The materials cover a broad ran

ManiSkill is a GPU-accelerated robot simulation framework designed for training robotic manipulation skills, benchmarking learning algorithms, and generating synthetic datasets. It serves as a reinforcement learning environment where robot control policies can be developed and evaluated using parallelized physics and rendering on the GPU. The platform is distinguished by its ability to perform sim-to-real transfer, allowing policies trained in virtual environments to be deployed onto physical robotic hardware. It features ray-traced parallel rendering for producing high-frame-rate RGBD and se

TrackWeight is a digital weighing instrument and pressure-to-weight converter designed to measure the mass of objects. It functions as a pressure sensor weight scale that translates data from a touch surface into weight measurements in grams. The system integrates pressure sensor data to enable digital weight measurement and embedded weight tracking. It converts raw pressure readings into standardized weight units to quantify the mass of objects or fingers in contact with a sensory surface. The implementation includes analog-to-digital signal conversion, linear regression calibration for acc

Baselines is a comprehensive suite of frameworks for reinforcement learning algorithm implementation, imitation learning, and training orchestration. It provides a library of standardized learning algorithms used to benchmark and replicate research results, alongside a deep learning policy framework for constructing neural network architectures such as multi-layer perceptrons, convolutional networks, and long short-term memory networks. The project includes a specialized imitation learning toolkit that enables agents to mimic expert behavior through behavior cloning and generative adversarial

Diffusion Policy is a robot learning framework that uses diffusion models to map visual observations to precise action trajectories. It functions as an imitation learning toolkit and visuomotor policy learner, providing a system to train neural networks that replicate human behavior by generating robotic movements based on image and sensor data. The framework employs a conditional denoising process to sample sequences of robotic movements, allowing it to handle multimodal action distributions where multiple valid trajectories may exist for a single state. It utilizes score-based action modeli

DeepMimic is a deep reinforcement learning framework and physics-based motion imitation tool designed to teach simulated characters and robots to reproduce human movements. It provides a pipeline for integrating motion capture data into physics simulations to train agents that can mimic complex physical skills. The system utilizes the PyBullet simulation environment to execute motion policies and visualize character interactions in real time. It includes a motion capture integration pipeline that imports and processes animation sequences to serve as reference targets for imitation learning ag

Rerun is a multimodal data visualizer and robotics data logger designed for rendering synchronized streams of 3D spatial data, images, and time-series metrics. It functions as a tool for capturing high-frequency sensor data and AI outputs into a queryable columnar format, providing a dedicated interface for viewing MCAP recording files and analyzing physical environments. The project distinguishes itself as a machine learning dataset streamer, capable of feeding logged recordings directly into GPU buffers and PyTorch training pipelines without intermediate exports. It supports a high-performa

Doccano is a collaborative data labeling platform and machine learning dataset management system. It provides a web-based interface for teams to import raw text, mark datasets, and export structured annotations for model training. The project specifically supports text annotation for classification and named entity recognition tasks. It enables teams to coordinate multiple users on a single project to maintain consistent labeling guidelines and increase the speed of dataset creation. The system includes tools for data management and team coordination, providing the ability to import raw data

labelImg is a computer vision labeling tool and image bounding box annotator used to create training datasets for machine learning models. It functions as a desktop utility for drawing rectangular labels on images and saving object coordinates and class names in common machine learning formats. The tool is specifically designed to generate and edit PascalVOC formatted XML files and create image labels in the text-based format required by YOLO object detection pipelines. The software covers object detection annotation and training data preparation, including the ability to manage label catego

Label Studio is a multi-modal data annotation platform designed to create and manage high-quality training datasets for machine learning. It functions as a self-hosted, containerized environment that supports secure, private deployments, including air-gapped configurations. The platform provides a centralized workspace for labeling diverse media types, such as images, text, audio, and time-series data, to support supervised and reinforcement learning workflows. The platform distinguishes itself through deep integration with machine learning backends, enabling active learning loops, automated

This project is a reinforcement learning toolkit and simulation-based AI trainer for creating intelligent agents within Unity simulations. It provides a multi-agent simulation framework for configuring cooperative or competitive scenarios and includes an environment wrapper that bridges simulations with standard machine learning libraries using gym-style interfaces. The system features a native cross-platform inference engine that executes trained neural network models for real-time decision making without external dependencies. It enables the acceleration of the learning process by running m

This project provides open-source hardware designs for a high-end, repairable laptop. It defines a modular architecture centered on a precision-milled CNC aluminum chassis and a hardware layout that prioritizes accessibility and user-replaceable parts for long-term maintenance. The design is distinguished by the integration of a 4K AMOLED display and a custom wireless mechanical keyboard utilizing low-profile switches. The system employs a modular approach that allows the input system to be detached and used as a standalone peripheral. The technical scope covers high-performance motherboard

VoTT is a computer vision annotation software and machine learning dataset preparation tool. It is a desktop application designed for drawing bounding boxes and assigning tags to objects in images and videos to create training datasets for object detection models. The application utilizes a cross-platform desktop interface to manage image and video assets. It features a local-first storage integration to handle large media assets directly from the host machine's file system and includes frame-rate controlled video sampling to extract specific images from video streams for labeling. The softw

Viser is a Python 3D visualization framework and remote scene server that renders 3D primitives, point clouds, and meshes in a web browser. It functions as a server-client system that synchronizes scene state and camera poses to a web client via WebSockets. The framework provides specialized capabilities for robotics and computer vision, including a URDF robot visualizer for loading robot models and joint states, as well as a GPU-accelerated Gaussian splatting viewer for high-fidelity volumetric rendering. It also supports the visualization of human body models and skinned meshes for pose ana

DRL is a curated educational resource that teaches deep reinforcement learning through a structured series of lectures and videos. It covers the three main families of reinforcement learning methods: actor-critic architectures, value-based algorithms like Q-learning and DQN, and policy-based techniques that directly optimize an agent's action-selection strategy. The curriculum extends beyond these core topics to include imitation learning, multi-agent training, and methods for handling continuous action spaces. Content is organized as markdown-driven documentation that generates static, navig

Dokploy is a self-hosted platform-as-a-service designed to simplify the deployment and management of containerized applications and databases. It provides a centralized control plane that decouples administrative management from application workloads, allowing users to oversee infrastructure across multiple server nodes through a unified web interface or a command-line tool. The platform distinguishes itself through an extensive library of pre-configured application templates, enabling the rapid deployment of databases, identity providers, and various productivity or development tools. It sup

This project is an educational platform and research toolkit designed to teach deep learning through a combination of mathematical theory, visual diagrams, and executable code. It provides a comprehensive environment for building, training, and evaluating neural networks, grounding complex concepts in interactive computational notebooks that allow for hands-on experimentation. The framework distinguishes itself by interleaving theoretical foundations—including linear algebra, calculus, and probability—with practical implementations across multiple industry-standard libraries. It supports flex

Mobile Alohas is a bimanual mobile manipulation framework designed to learn whole-body robot control. It provides an integrated imitation learning pipeline that manages the process of collecting human demonstration data and training behavior models to automate complex physical tasks. The system features a robotic teleoperation interface that maps human movements to a mobile robot with dual arms. It includes a whole-body motion dataset tool used for recording, visualizing, and replaying joint and sensor data from manipulation sessions. The framework covers several capability areas, including

Hub is a multimodal AI data lake and vector database designed for storing and querying embeddings, text, audio, and images. It functions as a dataset version control system and a machine learning data streaming engine to support large-scale model training. The system utilizes a serverless PostgreSQL vector store to index high-dimensional embeddings for semantic search. It provides a visual interface for inspecting multimodal datasets and viewing annotations such as bounding boxes and masks. The platform handles cloud-agnostic storage synchronization and implements lazy, compressed data strea