Robot-sensor integration

The vast majority of installed industrial robots operate in a repetition mode, executing a sequence of learned movements. Reactivity to external events is very limited, often limited to basic reactions triggered by "all-or-nothing" events.

The environment must therefore be constant and known with great precision. For a robotic welding station, for example, the workpieces to be welded must be precisely positioned in front of the robot(s) before it performs its learned task. Any error in this positioning will have a direct impact on the precision of the task.

On the other hand, when the robot is working in contact (as in the case of insertion, polishing or screwing tasks, for example), in the absence of an additional sensor and if the dimensions or positioning of the part deviate from the nominal values, the forces generated can increase dramatically, even to the point of destroying the part or tool.

To give the robot greater flexibility and adaptability, it is necessary to include external sensory feedback in its control system.

Vision can be used to take into account variations in the robot's working environment. This is known as "vision control" or visual "servo-control". When the robot is working in contact, a force transducer can be used to measure and limit interaction forces. This is known as "force control".

This article is dedicated to robot-sensor interaction for the two most commonly used exteroceptive sensors in robotics: the camera and the force sensor.

The theoretical aspects of the various control architectures are dealt with in a very concise way, giving the reader the opportunity to delve deeper into the subject by citing references.

Sensor technology, examples and real-life case studies form the practical side of the article.