The Start-up

Digithurst began trading in 1982, but the idea of using a camera as a computer peripheral was not new. Two years earlier, Stephen Cronk and I had been working for a small electronics company called Sands Whiteley selling microcomputers – most of them grey imports from the US. The company’s main business was building process control and test equipment for British Leyland. While discussing ideas for peripherals that we could sell alongside microcomputers, Rod Starksfield, the company’s chief engineer, suggested a small camera-like device made from an array of light-sensitive diodes. It would interface to a Commodore PET and allow an attached robot to track moving objects. Rod, Stephen and I put together a project plan envisaging the Colne Electronics educational robot as a tracking device.

In the back of a cupboard at home I had a description of an edge detection algorithm developed jointly by Robotron, an East German computer company, and Karl Marx Stadt Technical College. Stephen had already created a software package for the Acorn Atom called Atomic Pencil based on a draughting system I worked on at the CADCentre. Atomic Pencil had sold well and there seemed no reason why we couldn’t repeat its success with a ‘Robot Eye’ system.

The company’s management baulked at the idea of buying in a robot arm and suggested that, as I was a mechanical engineer, the company could build one in-house. We thought they were joking; swapping transformers in imported microcomputers was one thing, but building complex mechanical devices was another. Also, why reinvent the wheel? Unfortunately, they were serious and so Stephen and I spent a painful twelve months designing and building the SA2 robot.

There never was an SA1 and, as the recession of the early 1980s finally reached Sands Whiteley, what was essentially a production prototype was rushed onto the market. Over thirty were sold on day one. However, the SA2 and the Atomic Pencil failed to make up the revenue lost when US companies finally tightened up their overseas distributor channels and the grey market for microcomputers dried up. Stephen, who had been at Sands Whiteley on a work experience scheme, returned to Cambridge College of Arts and Technology. I became unemployed, the distress of being out of work in the middle of a recession tempered by the fact that I would never see another SA2. Sands Whiteley bravely soldiered on with the robot until, a year after Stephen and I left, the company closed.

Plan A was to set up a microcomputer business – something I had intended to do while at the CADCentre, before I was offered a job at Sands Whiteley. But there was no business as such, just me promoting myself as a specialist in 6809 and 68000 systems; it was brave, or foolhardy, as I only had access to 6502-based microcomputers. Still, there was a chance of becoming a UK agent for Force Computers, a German computer company. Stephen’s and I put together a quote to automate the back-office operations of an insurance broker. Unfortunately, we lost the sale after the dealer we tried to source terminals from discovered the name of the prospective customer. It seemed the market for microcomputer-based business systems had already matured and was now the domain of established companies.

In desperation, we resurrected the robot vision project, although building a camera would be out of the question. However, in the absence of a plan A, the Robotron algorithms came out of the cupboard again. Finance was going to be a problem as my only income was a fortnightly unemployment cheque. Walter Herriot, a business manager at the Cambridge branch of Barclay’s Bank and famous for handing over that cheque to Acorn Computers (BBC, The Micro Men), was the obvious person to talk to. As we sat in my living room, now doubling as my office and workshop, I explained the project. Walter said it sounded exciting and then asked if I would use my house as security for a loan, which I found just a little too exciting. You could see Walter’s point, Barclay’s realised Sands Whiteley and a number of other Cambridge hi-tech companies were already in trouble. The SA2 Robot seemed to hang around my neck like an albatross.

Adrian March Electronics marketed a GE CCD camera designed for robot vision. Adrian himself demonstrated it to me in his office in Farnborough, then offered to loan me one to interface to a microcomputer, even though he felt I was wasting my time. The last of my savings was used to buy a 32K Commodore PET and a single disk drive purchased from MMS in Bedford. The quote from Stirling Microsystems for a 6809 Fujitsu with a 68000 cross compiler was too high, as was a printer and the robot arm from Systems Control in Northallerton in Yorkshire. A rather distressed oak desk was rescued from a reclaimed building materials supplier and an ancient electric typewriter was purchased using revenue – at that point, still the fortnightly dole cheque.

Interfacing the GE camera to the Commodore was relatively straightforward – easier than writing driving software for the SA2 Robot. Even so, the high data rate necessitated the use of 6502 assembly language. Selling the concept was more difficult given the mismatch between the price of the camera – £2,000 – and the retail price of microcomputers, typically £500. How would we price the software: £500 or £2,000? Vision Dynamics, based in Watford, had interfaced the same camera to a PDP 11 and were selling imaging software for up to £12,000. Little wonder Adrian March had doubts about our venture.

It was time to test the market and so I arranged a photoshoot and wrote a press release. A wooden model was used in place of the robot arm. The location for the photoshoot was the local junior school; both the Commodore PET and the BBC Model B in the photograph belonged to the school. It was no great surprise the release drew little interest from the broadsheets. However, it was very disappointing that there was zero interest from the trade press. Even our local newspaper, The Cambridge Evening News, which had regularly given over column inches to Sands Whiteley, failed to give us a mention.

- 01010010 -

While searching for mentions of our camera system in PC magazines (these were the days before they were sealed in plastic envelopes), a copy of Wireless World caught my eye. It had a pixelated image on the cover and, inside, there was a circuit diagram for a low-cost video digitiser designed by Worcester-based electronics engineer Peter Howard. Coming on top of the deafening silence that greeted our press release, these comprehensive instructions telling microcomputer owners how to build their own computer vision systems was bad news. Showing the article to Stephen, I asked how long he thought it would be before versions of Peter Howard’s digitiser started flooding onto the market. He estimated three months, possibly two if we could get some help with the analogue part of the circuit. We? You mean we are going to manufacture this ourselves? …

(Extract from The Ghost in the Labyrinth by Peter Kruger)

Next chapter ...