Minibot Proof of concept
This post shows a proof of concept build of our newly designed minibot.
Published on July 01, 2021 by Simon M. Haller-Seeber
diy robot
A new Educational Robot Platform
To give students a good introduction to robotics it is important to give them also hands-on experience. This requires a robot platform with certain capabilities. The platform itself should be affordable, easy to use, and robust, and one should still be able to showcase research and/or more complex algorithms. At the moment there is no suitable, off-the-shelf product out there: Either they are really cheap and do not provide any sensory feedback, or they are industrial or research products which are too expensive, the intended usage is too narrow, or they are unreliable.
Therefore a combination of consumer products building up a robot which is capable of performing interesting tasks such as grasping, self-localisation and mapping, planning is attractive.
We did a market search to identify feasible partial solutions to build such a robot:
As a mobile base platform Sphero RVR [1] is considered. Its software support is wide (ROS, Python, C++, etc.) and it is still not too expensive.
For grasping and manipulation a 4-DOF Lynxmotion Arm [2] is interesting: The LSS Motors provide additional sensor feedback (voltage, current, temperature, position, and RPM) and they can set properties such as angular stiffness, holding stiffness, acceleration and deceleration. Software support is not too wide but there is a complete, open reference and it should be possible to build a C library and ROS support within reasonable time.
As a compute module the Nvidia Jetson [3] in combination with a Intel depth camera [4] could be used for machine learning algorithms. This combination is currently also used and tested with the RoboCup @ Work setup by our UIBK Team.
References / Links
[1] Sphero RVR
[2] Lynxmotion LSS 4Dof
[3] Nvidia Jetson Nano
[4] Intel Realsense D435