by David Sands, RoboForth author.
Back in the 80s in a company called Sands Whiteley Research and Development Ltd, I had the idea to make an educational robot arm. Two engineers were given the project, Rod Starksfield and Peter Kruger. Rod Starksfield produced a prototype SA1 which was subsequently redesigned for low volume production. The SA2 robot arm was launched and the company started getting orders.
Peter Kruger was given overall project management while Rod Starksfield moved on to industrial automation. Kruger was tasked with producing robots with technician support and also with writing the software which he did in Basic for the BBC, Commodore Pet, TRS80 and other computers. The sister company Intelligent Artefacts was the marketing arm for the robot as well as many other successful products. Peter Kruger was general manager and William Miller was on marketing. Orders were coming in all the time but Kruger was spending increasing time on customer support and unauthorized projects of his own. After the inevitable parting of ways I got all the flack from customers calling about bugs such as "no such line number" and errors of that sort that can be common with Basic interpreters. I then spent an intense week writing the first RoboForth. Then when a customer called to complain I just mailed them a RoboForth disk and never heard from them again. We caught up the back orders and every one went out with RoboForth I. Peace and Light!
Unfortunately Barclays Bank closed Sands Whiteley's account and took all the assets from Intelligent Artefacts through a cross guarantee so
both companies closed. Together with a management expert I started a new venture called Cyber Robotics. A mechanical consultant and Catherine George redesigned the robot as the Cyber 3 using the same technology but much nicer styling. I wrote a new version of RoboForth to run on 6502 and Z80 computers and also a version for the Armdroid in collaboration with Colne Robotics. Cyber Robotics was subsequently sold to the Bibby corporation.
I left Cyber Robotics and started Imagecroft to make and market more professional robot arms for benchtop and lab use. An engineer from Labman Automation designed the first R12 for us using cast limbs and more powerful motors. RoboForth at that stage allowed the user to learn a list of joint coordinates which the robot would follow after typing RUN. Using the Forth feature CREATE DOES> I was able to give the user the ability to create multiple named lists called ROUTEs each with it's own list of coordinates. Then I added named positions called PLACEs which were self learning and self executing. Next was inverse kinematics (and forward kinematics) so that Cartesian coordinates could be used instead of joint values. Commodore Pet and computers like that became obsolete so I designed my own robot controller that could run RoboForth on its own, but tethered to an IBM PC.
The R17 was designed in the 90s with encoder feedback. Then we added a DSP to control the motors independently of the CPU, able to run multiple motors at different speeds in different directions, with programmable speed and acceleration and at the same time read back the encoders to the CPU.
Over the years many more functions have been added including programmable stop circuit, vectored execution, I/O, integrated development environment (IDE) and programmable JERK which is the rate of change of acceleration.