Modeling and control of robot manipulators

Mastery of the design and operation of complex motorized mechanisms, or "machines", has always been an important factor in technological progress, and sometimes also in social and economic progress, in various fields: transport, industrial production, public works, exploration and work in hostile environments, medical imaging, etc. The design, manufacture and control of these machines have been made possible by scientific and technical knowledge in mechanics, thermodynamics, electrical engineering and hydraulics. These machines are designed to enhance human capabilities in terms of speed of movement and action, strength and range of action, particularly in the performance of arduous, dangerous and/or repetitive tasks. The functions generally concerned are :

• moving over varying distances on land, at sea, in the air and underwater;

• manipulation" in the broadest sense of the term:

  • move a tool to pick up objects or materials, transport and set them down,

  • environmental efforts ;

• the combination of the two previous functions.

A robot manipulator can therefore be considered in general terms, as seen by its environment, as a generator of movements and forces in various directions in space.

In terms of the most common applications, we can distinguish :

• industrial robots, generally working at a fixed station, in a totally autonomous way, and whose "tasks" are programmed on site by learning, or off-line by using a specialized language or computer-aided design means;

• robots for intervention and exploration in hostile and unfamiliar environments (nuclear, planetary, submarine, etc.), which are usually remotely operated, but can be equipped with a degree of local autonomy, given the difficulties associated with transmission delays and low bandwidth. The "virtual reality" tools emerging in many laboratories and industries are likely to help operators control the manipulators fitted to robotized vehicles.

In all these cases, where robot manipulators are not directly remote-operated and must have a certain degree of autonomy of action, their automatic control systems must be aware of and compensate for any inaccuracies, since the human operator is not directly in the servo loop, in order to adapt to the characteristics of the machines and their environment. This requires precise mathematical modeling of the geometry and dynamics of the manipulator arms.

The aim of this article is to familiarize the reader with the main concepts involved...