Skip to main content

AI-Robot Platform for Scientific Exploration

Murilo Marques Marinho, Juan José Quiroz-Omaña, and Kanako Harada.

How to cite us
Check out our paper at the ArXiv version.
@misc{aiscienceplatform2022,
title = {Design and Validation of a Multi-Arm Robotic Platform for Scientific Exploration},
author = {Marques Marinho, Murilo and Quiroz-Oma\~na, Juan Jos\'e and Harada, Kanako},
doi = {10.48550/ARXIV.2210.11877},
url = {https://arxiv.org/abs/2210.11877},
year = {2022},
}

Exhibition on IROS 2022

Real Platform

The robotic system is composed of four branches able to independently rotate about a circular rail. Furthermore, each branch is composed of a serial manipulator mounted on a linear actuator that moves back-and-forth in the direction of the center of the rail.

Easy VFI Descriptions

Definitions of (self) collision-avoidance constraints by means of the VFI framework. The constraints are implemented using geometric primitives on CoppeliaSim and a configuration file.

Realistic Simulations

Digital twins implemented on Nvidia Isaac Sim. Before using the real platform, operators can be trained using simulation examples of peg transfer tasks.

Powered by ReactROSGNU/LinuxqpOASESSmartArmStackQtGNU/Linux SystemROSControllersKinematicWindows SystemMaster InterfaceCoppeliaSimIsaac SimPython APIDQ RoboticsHapticInterfaceReal Platform

Powered by Open Software Technologies

Available low level controllers on Python and C++ (including ROS packages) for the robots, sensors, actuators, and grippers. The robot modeling and control is implemented using the DQ Robotics library.

Long-Term Goal

Our long-term goal, part of the ambitious Moonshot Research & Development Program of the Cabinet Office of Japan is, by 2050 and beyond, have AI-robots that autonomously learn, adapt to their environment, evolve in intelligence and act alongside human beings.

About the Project

This project is under development at the Harada lab at the University of Tokyo and was funded by JST Moonshot R&D (JPMJMS2033-09).