• Part 1
  • Part 2
  • Part 3

The History of Robokit - Part 1

Richard Pawson

Atari Robokit was designed by Personal Robots Ltd (PRL) in the late 1980s and licensed to Atari Corp.

PRL was formed in January 1985 by Julian Allason and me  -  as Chairman and Managing Director respectively. 

We had both been involved in the very start of the personal computer business, and both had decided that ‘personal robots’ would be ‘the next big thing’. This belief turned out to be misguided, but we were not alone in holding it:  Nolan Bushnell, founder of Atari, started two personal robotics companies in the mid-80s, neither of which ultimately succeeded.  PRL survived four and a half years (quite a good record for such companies!) before being voluntarily liquidated.  Atari UK bought some of the assets from the liquidator, including the head rights to Robokit; and PRL’s senior electronic engineer, Gary Lawman, went to work for Atari for a few years. Despite its ultimate failure, PRL was four and a half years of immense fun: at its peak it employing a dozen young engineers and designers, and developing an extraordinary range of prototypes and products.

In 1984 I had written The Robot Book -  a well-illustrated popular-science style of book about robotics, that included a large number of practical projects for building robotic devices from Lego Technic and from FischerTechnik  -  most of them linked to a home computer.  Some are shown here:


In forming PRL, our initial idea was to create a new construction set, purpose-designed for building intelligent robots.  Our first employee, Doug Cunningham, was an industrial designer, trained at the then Central School of Art and Design.  Apart from a knowledge of product design, plastics moulding, and other technical skills, Doug was a superb product illustrator.  (Today, almost all product design is done using CAD, and these extraordinary skills of product visualization using ‘Magic Markers’ and/or airbrushing  -  beloved of automotive designers  -  are virtually lost). Shown below are some of his
earliest visualisations of the product  - working to my brief -  including the presentation box  (this was before we had alighted on the name Robokit):

It was largely on the strength of Doug’s superb illustrations  (we had no working prototype at that point) that Julian and I successfully raised £250k in venture capital, to fund the expansion of the team and building prototypes of the physical components, electronics, and control software.  It’s worth explaining a few of the elements:

The main building block was called the ‘lock block’. The lock block connected with ‘tracking’  -  pieces of purpose-designed extruded aluminium, of varying cut lengths.  As shown in the photo below, the design of the lock block could connect the elongate tracks in a very flexible manner; it’s hollow design was to accommodate axles and other smaller elements.

Another key element was the ‘elbow joint’, which provided high-torque powered rotation through about 135 degrees.  The elbow joint adopted a servo motor design  -  meaning that it could be rotated precisely to any specified angle, and would hold that position against reasonable force.  The single working prototype, shown below along with the original design visualisation, was actually built from the components of a stripped-down Futaba radio-control model servomotor:

There were some other actuators, and a wide range of sensors, all designed to fit onto the track and/or lock-block.  Some of them are shown here:

All of these actuators and sensors were designed to plug into a microprocessor-based ‘slave board’, which was itself a physical construction component that could form the ‘base board’  for a static model, or be incorporated into a mobile robot.  The ‘slave board’ was designed to be hooked up to a home computer for the programming of the robot, but then be capable of functioning autonomously (and under battery power).  We built many working versions of the slave board, which. we ultimately used in prototypes of finished products that were quite different in conception to Robokit.

The next illustrations show some of the devices that we envisaged being able to build from Robokit:

The photographs below are physical mock-ups of some of those designs.  The ‘track’ pieces are real product, but the rest of the components in these photos were just painted balsa-wood. 

(As an aside, I recently became aware of Lego Mindstorms NXT.  The number of similarities between Mindstorms NXT (it is far better than the earlier incarnations of Mindstorms) and our conception of Robotkit in 1985 is extraordinary.  I do not mean to imply that they copied the idea  - we were by no means the only people who had the idea for an intelligent robotic kit, even then.  Nor do I seek to suggest that we were merely ahead of our time.  Although Mindstorms NXT seems to have already developed a cult following, it is far from clear whether it would really be commercially viable outside of Lego’s huge infrastructure.   I merely find it gratifying to see a good idea come to reality  -  whoever does it.  At that time, I took a rather narrow view of ‘intellectual property’  -   we did file a patent application covering some of the core ideas in Robokit, and had some very elaborate ideas about controlling the envisaged market for third-party plug ins.  Today my ideas of IPR are very different  -  and my current company, Naked Objects Group, open sources all its code.)

The History of Robokit - Part 2

Richard Pawson

Selling (or, rather, not selling) Robokit
It was always our intent to try license Robokit to one of the major toy manufacturers, who would take on the detailed mechanical design & tooling – with PRL doing the electronic and software design.  Surprisingly, as much as 40% of successful toys are invented by individuals or small companies and licensed to the majors (mostly American-owned).

Beyond this piece of information, Julian and I really knew nothing about the toy business. The only relevant skill we had was the ability - gained partly from spells in journalism - to talk our way into hard-to-access places. During the summer of '85 we would, quite literally, get on a plane to the USA - put ourselves up in the best hotel in town (Julian insisted that this was important) - and get on the phone. With no knowledge and no prior connections, we nonetheless talked ourselves into meetings with the CEOs and other senior management of most of the major US toy manufacturers. Being British probably helped a little: they do say that a British accent is worth 10 IQ points in America!

We cross-crossed the states from Calilfornia (Mattel), to Connecticut (Coleco), to Minnesota (Tonka). The most extraordinary place we visited was Dyersville, Iowa, to meet with Fred Ertl, Jnr, the president of Ertl toys, best known for model tractors, much beloved by mid-west farmers. To say that Dyersville is out of the way is putting it mildly: when we asked for directions from Dubuque Airport, the Hertz rep asked us if we were from 'outa town' - as she detected a strange accent. ‘Yes’, we said. 'In that case, I'll direct you via the black tops'. (We had no idea what this meant, but later that it meant metaled roads!) Some years later, Dyersville Iowa, became famous as the setting for 'Field of Dreams' – and you can apparently today visit the baseball diamond in the middle of the cornfields.

Our naiveté, in retrospect, was extraordinary. At Coleco I recall saying that 'one great thing about Robokit is its educational value.'  'Educational?' responded the Marketing Director, Al Kahn, the 'brains' behind the launch of Cabbage Patch dolls, 'That's a dirty word in the toy business.' At Mattel, I finished my pitch with the hubristic claim that 'We believe Robokit could be as big a video games!'. There was a long, pregnant silence, after which the CEO, Tom Kalinksi responded, a waery look on his face, 'We lost 400 million dollars on video games.'  Ah - I had forgotten about IntelliVision.

Yet despite our naiveté, the executives we met with were remarkably open, friendly and engaging. (I developed a very strong liking for America, which I have never lost). But it became very evident to us that Robokit just didn't hit the spot for any of these major toy makers. What we did gain was two very useful pieces of advice. The first was that we should consider applying our robotics know-how not to a kit, but to some kind of ready-made device.  The second was that we should consider partnering with a small UK-based company called Seven Towns, who were evidently quite well known in the toy business.

To be continued...