DeepClaw

paper | poster | video

We present DeepClaw as a reconfigurable benchmark of robotic hardware and task hierarchy for robot learning. The DeepClaw benchmark aims at a mechatronics perspective of the robot learning problem, which features a minimum design of robot cell that can be easily reconfigured to host robot hardware from various vendors, including manipulators, grippers, cameras, desks, and objects, aiming at a streamlined collection of physical manipulation data and evaluation of the learned skills for hardware benchmarking. We provide a detailed design of the robot cell with readily available parts to build the experiment environment that can host a wide range of robotic hardware commonly adopted for robot learning. We also propose a hierarchical pipeline of software integration, including localization, recognition, grasp planning, and motion planning, to streamline learning-based robot control, data collection, and experiment validation towards shareability and reproducibility.

DeepClaw is a benchmarking system for robot manipulation designed to be modularized, extendable, and easy to use on real robots and the environment. As shown in Fig.1, DeepClaw consists of four components:

1. A standardized robot cell design.
2. A set of unified driver interfaces that serves as a connection layer between the benchmarking algorithms and the hardware, including robot arms, robot hands, visual sensors, tactile sensors.
3. A collection of baseline algorithms for segmentation, recognition, grasp planning, and motion planning.
4. A pipeline for task definition and functional integration.

For a detailed decription of DeepClaw and benchmarking tasks, please visit the website of DeepClaw

Quick Start

To be continued ... (the kind of code that anyone needs to get started with DeepClaw)