> I was a bit surprised to see that they could demo an early MIT CADR at Xerox PARC. These were large, fragile and rare machines at that time.
They weren't extraordinarily fragile; robot wirewrapping is pretty robust. The next year we shipped a couple of them to Paris and I used them just fine, along with a KL-20 that also made the trip OK.
This is not an “early” CADR, it is just a CADR. By ‘81 they were heavily used at the AI lab and as gumby mentions, they aren’t fragile little machines.
No running (real) CADRs exist, unless you consider the two FPGAs on my desk.
For a CADR simulator you can check https://tumbleweed.nu/lm-3 — I managed to restore the last system version for it last year or so, and we are continuing hacking adding and fixing things.
E.g. you can run the simulator against the Global Chaosnet and talk to other LispMs and ITS machines (simulated or not). And some of us do run it 24/7 as a file server for other LispMs.
The current HDL implementation though only works on a unobtanium FPGA board. We are slowly working on porting it over to something that can actually be bought these days. Help needed if you are keen on HDL hacking.
One of the initial reason for the Lisp Machine project back then was to run Macsyma, since the PDP-10 was too constrained, and a multi-user system that meant you had to share resources.
Which is yet another reason to show most embedded boards nowadays do just well with managed languages, it is only a matter of culture and urban myths preventing many people to do so.
What more can you say about or link to of the wirewrap robot? I remember seeing it in the lab, but never saw it in action, and I haven't been able to dig up anything more about it. I'd love to see a video of it doing its thing!
Maybe its corpse appears in this video from 1993, which might be years too late, but it does show off some of its beautiful work.
>This film from 1968 shows Marvin Minsky's tentacle arm, developed at the MIT AI Lab (one of CSAIL's forerunner labs). The arm had twelve joints and could be controlled by a PDP-6 computer or via a joystick. This video demonstrates that the arm was strong enough to lift a person, yet gentle enough to embrace a child.
The stuff at the end reminds me of Golan Levin's adorable googly-eyed worm robot (which was a menacing BB IRB-2400/16 underneath):
>"Double-Taker (Snout)" (interactive robotic installation, 2008) deals in a whimsical manner with the themes of trans-species eye contact, gestural choreography, subjecthood, and autonomous surveillance. The project consists of an eight-foot (2.5m) long industrial robot arm, costumed to resemble an enormous inchworm or elephant's trunk, which responds in unexpected ways to the presence and movements of people in its vicinity. Sited on a low roof above a museum entrance, and governed by a real-time machine vision algorithm, Double-Taker (Snout) orients itself towards passers-by, tracking their bodies and suggesting an intelligent awareness of their activities. The goal of this kinetic system is to perform convincing "double-takes" at its visitors, in which the sculpture appears to be continually surprised by the presence of its own viewers — communicating, without words, that there is something uniquely surprising about each of us. More information at http://www.flong.com/projects/snout/.
They weren't extraordinarily fragile; robot wirewrapping is pretty robust. The next year we shipped a couple of them to Paris and I used them just fine, along with a KL-20 that also made the trip OK.