control-toolbox - The Control Toolbox - An Open-Source C++ Library for Robotics, Optimal and Model Predictive Control

•    C++

---

This is the Control Toolbox, an efficient C++ library for control, estimation, optimization and motion planning in robotics. Link to the wiki, quickstart! Find the detailed documentation here.

cpp robotics automatic-differentiation control-systems trajectory-optimization optimal-control model-predictive-control rigid-body-dynamics lqr-controller extended-kalman-filter ilqg ilqr disturbance-observer multiple-shooting riccati-solver

free_gait - An Architecture for the Versatile Control of Legged Robots

•    C++

---

NOTICE: This software is not supported anymore! The authors of this software have changed their affiliation and do not work on this project anymore. Please excuse any inconvenience this might cause. If you are interested in working with the ANYmal robot, please reach out to ANYbotics. Free Gait is a software framework for the versatile, robust, and task-oriented control of legged robots. The Free Gait interface defines a whole-body abstraction layer to accommodate a variety of task-space control commands such as end effector, joint, and base motions. The defined motion tasks are tracked with a feedback whole-body controller to ensure accurate and robust motion execution even under slip and external disturbances. The application of this framework includes intuitive tele-operation of the robot, efficient scripting of behaviors, and fully autonomous operation with motion and footstep planners.

api control robot robotics motion motion-planning ros apis motion-design whole-body-control legged-robots free-gait whole-body-api

rex-gym - OpenAI Gym environments for an open-source quadruped robot (SpotMicro)

•    Python

---

The goal of this project is to train an open-source 3D printed quadruped robot exploring Reinforcement Learning and OpenAI Gym. The aim is to let the robot learns domestic and generic tasks in the simulations and then successfully transfer the knowledge (Control Policies) on the real robot without any other manual tuning. This project is mostly inspired by the incredible works done by Boston Dynamics.

machine-learning reinforcement-learning robot robotics tensorflow openai-gym python3 artificial-intelligence inverse-kinematics openai reinforcement-learning-algorithms legged-robots quadruped robotic-arm openai-gym-environments pybullet gym-environment quadruped-robot-gaits quadruped-robot spotmicro

planet - Learning Latent Dynamics for Planning from Pixels

•    Python

---

This project provides the open source implementation of the PlaNet agent introduced in Learning Latent Dynamics for Planning from Pixels. PlaNet is a purely model-based reinforcement learning algorithm that solves control tasks from images by efficient planning in a learned latent space. PlaNet competes with top model-free methods in terms of final performance and training time while using substantially less interaction with the environment. PlaNet models the world as a compact sequence of hidden states. For planning, we first encode the history of past images into the current state. From there, we efficiently predict future rewards for multiple action sequences in latent space. We execute the first action of the best sequence found and replan after observing the next image.

yarp - YARP - Yet Another Robot Platform

•    C++

---

YARP is a library and toolkit for communication and device interfaces, used on everything from humanoids to embedded devices. Regular YARP builds use the ACE library. On Linux and macOS, YARP can be compiled without ACE by adding -DSKIP_ACE=TRUE when running cmake.

middleware yarp robotics-libraries robotics devices humanoids communication ipc interprocess-communication network robot research human-robot-interaction artificial-intelligence

evQueue - Job scheduler and queueing engine

•    C++

---

evQueue is an open source job scheduler and queueing engine. It features an event-driven C++ engine and a PHP / MySQL web control interface which provides tasks monitoring and creation. It provides both simple task execution and complex task chaining (workflow) using an easy to use drag & drop web interface. Workflow description includes output linking to input, conditions, loops... Queues management provides an easy way for task parallelization and resource control.

scheduler job-scheduler workflow cron high-availability cluster automation event job-queue task-queue

pytorch-dense-correspondence - Code for "Dense Object Nets: Learning Dense Visual Object Descriptors By and For Robotic Manipulation"

•    Python

---

Abstract: What is the right object representation for manipulation? We would like robots to visually perceive scenes and learn an understanding of the objects in them that (i) is task-agnostic and can be used as a building block for a variety of manipulation tasks, (ii) is generally applicable to both rigid and non-rigid objects, (iii) takes advantage of the strong priors provided by 3D vision, and (iv) is entirely learned from self-supervision. This is hard to achieve with previous methods: much recent work in grasping does not extend to grasping specific objects or other tasks, whereas task-specific learning may require many trials to generalize well across object configurations or other tasks. In this paper we present Dense Object Nets, which build on recent developments in self-supervised dense descriptor learning, as a consistent object representation for visual understanding and manipulation. We demonstrate they can be trained quickly (approximately 20 minutes) for a wide variety of previously unseen and potentially non-rigid objects. We additionally present novel contributions to enable multi-object descriptor learning, and show that by modifying our training procedure, we can either acquire descriptors which generalize across classes of objects, or descriptors that are distinct for each object instance. Finally, we demonstrate the novel application of learned dense descriptors to robotic manipulation. We demonstrate grasping of specific points on an object across potentially deformed object configurations, and demonstrate using class general descriptors to transfer specific grasps across objects in a class. To prevent the repo from growing in size, recommend always "restart and clear outputs" before committing any Jupyter notebooks. If you'd like to save what your notebook looks like, you can always "download as .html", which is a great way to snapshot the state of that notebook and share.

computer-vision deep-learning robotics pytorch artificial-intelligence vision manipulation 3d self-supervised-learning

Rusty - Light-weight, user-space, event-driven and highly-scalable TCP/IP stack

•    C++

---

Rusty is a light-weight, user-space, event-driven and highly-scalable TCP/IP stack. It has been developed to run on a EZChip TILE-Gx36 processor. Rusty is a light-weight, user-space, event-driven and highly-scalable TCP/IP stack. It has been developed to run on a EZChip TILE-Gx36 processor. Rusty takes full control of cores it runs on.

tcp-ip-stack tcp protocol-stack socket event-driven

f-stack - F-Stack is an user space network development kit with high performance based on DPDK, FreeBSD TCP/IP stack and coroutine API

•    C

---

With the rapid development of Network Interface Cards the poor performance of data packet processing with the Linux kernel has become the bottleneck in modern network systems. Yet, the increasing demands of the Internet's growth demand a higher performant network processing solution. Kernel bypass has emerged to catch more and more attention. There are various similar technologies such as: DPDK, NETMAP and PF_RING. The main idea of kernel bypass is that Linux is only used to deal with control flow; all data streams are processed in user space. Therefore, kernel bypass can avoid performance bottlenecks caused by kernel packet copying, thread scheduling, system calls, and interrupts. Furthermore, kernel bypass can achieve higher performance with multi-optimizing methods. Within various techniques, DPDK has been widely used because of it's more thorough isolation from kernel scheduling and active community support. To deal with the increasingly severe DDoS attacks the authorized DNS server of Tencent Cloud DNSPod switched from Gigabit Ethernet to 10-Gigabit at the end of 2012. We faced several options: one is to continue to use the original network stack in the Linux kernel, another is to use kernel bypass techniques. After several rounds of investigation; we finally chose to develop our next generation of DNS server based on DPDK. The reason is DPDK provides ultra-high performance and can be seamlessly extended to 40G, or even 100G NIC, in the future.

pinocchio - A fast and flexible implementation of Rigid Body Dynamics algorithms and their analytical derivatives

•    C++

---

Pinocchio instantiates the state-of-the-art Rigid Body Algorithms for poly-articulated systems based on revisited Roy Featherstone's algorithms. Besides, Pinocchio provides the analytical derivatives of the main Rigid-Body Algorithms like the Recursive Newton-Euler Algorithm or the Articulated-Body Algorithm. Pinocchio is first tailored for robotics applications, but it can be used in extra contexts (biomechanics, computer graphics, vision, etc.). It is built upon Eigen for linear algebra and FCL for collision detection. Pinocchio comes with a Python interface for fast code prototyping, directly accessible through Conda.

c-plus-plus robotics kinematics dynamics automatic-differentiation conda motion-planning ros code-generation urdf rigid-body-dynamics cppad fcl casadi analytical-derivatives pinocchio

habitat-sim - A flexible, high-performance 3D simulator for Embodied AI research.

•    C++

---

The design philosophy of Habitat is to prioritize simulation speed over the breadth of simulation capabilities. When rendering a scene from the Matterport3D dataset, Habitat-Sim achieves several thousand frames per second (FPS) running single-threaded and reaches over 10,000 FPS multi-process on a single GPU. Habitat-Sim simulates a Fetch robot interacting in ReplicaCAD scenes at over 8,000 steps per second (SPS), where each ‘step’ involves rendering 1 RGBD observation (128×128 pixels) and rigid-body dynamics for 1/30sec. Habitat-Sim is typically used with Habitat-Lab, a modular high-level library for end-to-end experiments in embodied AI -- defining embodied AI tasks (e.g. navigation, instruction following, question answering), training agents (via imitation or reinforcement learning, or no learning at all as in classical SensePlanAct pipelines), and benchmarking their performance on the defined tasks using standard metrics.

simulator ai computer-vision cplusplus robotics sim2real embodied sim agents

SwiftQ - Distributed Task Queue

•    Swift

---

SwiftQ is a distributed task queue for server side swift applications. Task queues are used as a mechanism to distribute work across machines. SwiftQ uses messages to communicate between clients and workers. In this case the message broker is Redis. SwiftQ uses the reliable queue pattern. This ensures that all tasks get processed even in the event of networking problems or consumer crashes. SwiftQ can be used for real time operations as well as delayed execution of tasks. SwiftQ can consist of multiple producers and consumers allowing for high availability and horizontal scaling. It is recommended that a separate Redis database is used to avoid conflicting name space.

server-side-swift

hexapod - Blazing fast hexapod robot simulator for the web.

•    Javascript

---

You can use this web app to solve inverse kinematics, simulate various gaits, and more. In real time, you can also view all the angles the robot's eighteen joints make at any particular pose. All the computations are solely done in your browser, nothing's fetching data from somewhere else, so it should be fast. Another (somewhat) cool thing is that this app does NOT depend on any external mathematics library; it only uses Javascript's built-in Math object. If you'd like to build you're own user interface with Node, you can download the algorithm alone as a package: Hexapod Kinematics Library. There is also a "fork" modified where you can use the app to control a physical hexapod robot as you can see in the gif below.

react simulator robot robotics simulation reactjs kinematics plotly inverse-kinematics hexapod kinematics-model hacktoberfest forward-kinematics visualizers kinematics-modelling

libuavcan - Portable reference implementation of the UAVCAN protocol stack in C++ for embedded systems and Linux

•    C

---

WARNING libuavcan v1 is not yet complete. This is a work-in-progress. Portable reference implementation of the UAVCAN protocol stack in C++ for embedded systems, Linux, and POSIX-compliant RTOSs.

communication-protocol uav cpp robotics space aerospace cpp14 cpp11 embedded-systems can drones uavcan can-bus misra socketcan vehicle-bus

Time Sid Manager

•    C++

---

Who is your best Sid Author? And Tune? TSID wants to answer to it. TSID library collects your listening time statistic about tunes in HVSC from a patched sidplayer. Now TSID2 (Next generation of TSID) is being develop.

concurrencpp - Modern concurrency for C++

•    C++

---

concurrencpp is a tasking library for C++ allowing developers to write highly concurrent applications easily and safely by using tasks, executors and coroutines. By using concurrencpp applications can break down big procedures that need to be processed asynchronously into smaller tasks that run concurrently and work in a co-operative manner to achieve the wanted result. concurrencpp also allows applications to write parallel algorithms easily by using parallel coroutines. concurrencpp is a task-centric library. A task is an asynchronous operation. Tasks offer a higher level of abstraction for concurrent code than traditional thread-centric approaches. Tasks can be chained together, meaning that tasks pass their asynchronous result from one to another, where the result of one task is used as if it were a parameter or an intermediate value of another ongoing task. Tasks allow applications to utilize available hardware resources better and scale much more than using raw threads, since tasks can be suspended, waiting for another task to produce a result, without blocking underlying OS-threads. Tasks bring much more productivity to developers by allowing them to focus more on business-logic and less on low-level concepts like thread management and inter-thread synchronization.

cpp scheduler coroutines concurrency tasks multithreading concurrent-programming asynchronous-programming threading await async-await task-scheduler thread-safety timers executor-service coroutine threadpool cpp20 executors

pytorch-a2c-ppo-acktr-gail - PyTorch implementation of Advantage Actor Critic (A2C), Proximal Policy Optimization (PPO), Scalable trust-region method for deep reinforcement learning using Kronecker-factored approximation (ACKTR) and Generative Adversarial Imitation Learning (GAIL)

•    Python

---

PPO is great, but Soft Actor Critic can be better for many continuous control tasks. Please check out my new RL repository in jax. Also see the OpenAI posts: A2C/ACKTR and PPO for more information.

reinforcement-learning deep-learning deep-reinforcement-learning pytorch atari hessian second-order continuous-control actor-critic ale mujoco proximal-policy-optimization ppo advantage-actor-critic a2c acktr natural-gradients roboschool kfac kronecker-factored-approximation

mujoco - Multi-Joint dynamics with Contact. A general purpose physics simulator.

•    C

---

MuJoCo stands for Multi-Joint dynamics with Contact. It is a general purpose physics engine that aims to facilitate research and development in robotics, biomechanics, graphics and animation, machine learning, and other areas which demand fast and accurate simulation of articulated structures interacting with their environment. DeepMind has acquired MuJoCo, and we are currently making preparations to open source the codebase. In the meantime, MuJoCo is available for download as a free and unrestricted precompiled binary under the Apache 2.0 license from mujoco.org.

robotics physics mujoco

gpt-3 - GPT-3: Language Models are Few-Shot Learners

•    

---

Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions – something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches. Specifically, we train GPT-3, an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model, and test its performance in the few-shot setting. For all tasks, GPT-3 is applied without any gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model. GPT-3 achieves strong performance on many NLP datasets, including translation, question-answering, and cloze tasks, as well as several tasks that require on-the-fly reasoning or domain adaptation, such as unscrambling words, using a novel word in a sentence, or performing 3-digit arithmetic. At the same time, we also identify some datasets where GPT-3's few-shot learning still struggles, as well as some datasets where GPT-3 faces methodological issues related to training on large web corpora. Finally, we find that GPT-3 can generate samples of news articles which human evaluators have difficulty distinguishing from articles written by humans. We discuss broader societal impacts of this finding and of GPT-3 in general.

acado - ACADO Toolkit is a software environment and algorithm collection for automatic control and dynamic optimization

•    C++

---

ACADO Toolkit is a software environment and algorithm collection for automatic control and dynamic optimization. It provides a general framework for using a great variety of algorithms for direct optimal control, including model predictive control, state and parameter estimation and robust optimization. ACADO Toolkit is implemented as self-contained C++ code and comes along with user-friendly MATLAB interface. The object-oriented design allows for convenient coupling of existing optimization packages and for extending it with user-written optimization routines. More information about the toolkit can be found at the homepage.