Robotics demonstration

Originally published on my OU study blog while studying module TM129 'Technologies in Practice'

Just watched this. Broadcast via Stadium Live from the OU's School of Computing and Communications.

'Baxter' is a collaborative robot, in the sense of being designed for human-robot interaction. He was made by Rethink Robotics.

In terms of the chat window, I said hello but that was it. There were quite a few questions being asked in the chat window, some sensible, some flippant, but I tried to focus on the demo and not be reading the chat. In fairness, people did seem interested. I guess robots do capture the imagination.

We were asked how humanoid we thought Baxter is. Well, it (he?) has arms. From the waist down though, not so much. I was amused that he seemed to have an OU ID card around his neck. 

Baxter

Baxter's arms contain seven motors - apparently they are referred to as '7 degree of freedom' arms. There are two at each of the three joints (one for the hinge, one for rotation) plus an extra rotation one.

Baxter has ultrasonic sensors on his 'head'. I think he also had other types of sensor but I didn't write these down.

Baxter's movements were being dictated by a program running on the presenter's laptop, on a nearby table. The program was written in Python though I assume other languages could be used. He doesn't have his own power supply - he's plugged in.

Sines and cosines are involved in working out where his grippers are in space. An applied use for maths :-)

Someone in the chat spotted what looked like a red push button on the table behind the presenter. Turned out this was indeed an emergency off switch, if Baxter went berserk and started killing people. Only kidding. We were assured that Baxter has been designed in a way that makes him safe around humans. And he can only lift around 1-1.5kg apparently. The lightweight.

The red button

The presenter asked whether we thought Baxter obeyed Asimov's Three Laws. Although Asimov was a sci-fi writer rather than a scientist, these 'laws' are still apparently considered useful in thinking about the safety of robot systems. They're difficult to apply literally though as they assume a robot can distinguish what a human being is (Asimov's fictional robots were very clever) - when in reality no robot actually built thus far could do this.