Historical Perspective of Industrial Robotics in the United States
Beginning with sales of the first UNIMATE Industrial Robots built and sold by Joseph Engelberger in 1961, robotics sales continued a steadily expanding growth through the 1980’s, mostly through heavy investment from automobile manufacturers.  Purchase of expensive robotic systems were justified based on high projected working hours and low mean-time-between-failures.  Robotics companies sold systems on the justification that robots would reduce factory labor costs by performing tasks quickly and accurately around the clock, referred to as flexible automation, since robots could be programmed for different factory production needs.  Robots were typically placed as stations along a production line.  In service, reliability problems and resultant "down-time" lead to work stoppage at the station and possibly of substantial sections of the production line.  Eventually companies could not afford the cost of production stoppage and unfortunately, we saw a decline in the use of robots for manufacturing.  Thus, buyers discovered robotic systems to be less versatile and less reliable than originally advertised and as a result, many robotics companies collapsed.  By the 1990’s, only a handful of American companies remain.
Having programmed several robots in the 1980’s myself, I believe there were three causes for the decline of robotics use and sales: inaccurate marketing of the robot’s capabilities, the need for special end of arm tools for different tasks and controls efficiency.  Controls efficiency is the ease in which a robot could successfully complete a given task through programming language and feedback techniques.  In summary, robots in the factory couldn’t adapt to the many varied factory environments which they were marketed and met their demise -ala Darwin’s Theory of Evolution.

Lessons Learned
Robotic systems must be flexible, reliable and easy to program to survive in today’s marketplace.  Unfortunately, the more complex a robot becomes, i.e. more joints, the greater the control complexity.  This tends to decrease reliability and ease of use.  However, design complexity does not necessarily result in control complexity, based upon a control technique that couples a robot design having anatomical similarity to humans and a body movement simulator, worn by a person, which accurately monitors joint angles of the person’s fingers, arms and upper body.  The fundamental premise of this control technique is rooted in the principle of anatomical consistency that all humans generally share.  Human anatomical features such as finger, hand and arm lengths possess length ratios, such that similarity from one person to the next would allow the average person the ability to easily control an "anatomically averaged" Robot.  It is the coupling of anatomically similar robotic link dimensions with a human operator that distills complex computer control of some 58 robotic upper body joints into relatively simple human motor-skills.  This control technique leverages millennia of developed, complex human hand-eye coordination tasks, which today are generally taken for granted.  Motor skills such as tying shoe laces, writing and typing are easily acquired skills, yet programming a robot for such a task would be extremely difficult.  In developing my Personal Assistant, computer resources are used for robot /teleoperator joint position and finger tip sensing, not multi-joint kinematic solutions, therefore eliminating motion delays associated with real-time kinematic motion control of many joints.  Furthermore, special tooling to interface robot hands to the task are not emphasized, utilizing instead, common tools manufactured for use by the human hand.