Sensor-based Control of Industrial Manipulators

Abstract

Automation is an important field in industrial research. Robots offer the possibility to automate a wide range of complex tasks. Especially with the use of sensor-based control, it is possible to automate tasks which today are performed by humans.

The industry in high cost countries like Norway must increase their effectiveness in order to stay competitive. This has often resulted in outsourcing production to countries with lower labor cost. An attractive alternative for increasing competitiveness, is the use of automation. The offshore oil and gas industry in particular, which is the largest industry sector in Norway, sees automation as a way to increase health and safety issues, as well as increasing the reliability of operations.

To meet the demands of future applications, which go beyond existing soultions such as repeating pre-computed motion, new sensors and control architectures must be used. This thesis investigates and develops control strategies for several applications for which there is industrial demand. Emphasis is placed on implementability on currently available hardware. This approach will by its nature include analysis which are application specific, but the generality of the methods developed is also considered. The main research topics and contributions found in this thesis include: A theoretical stability analysis is carried out for output-feedback control with redundant tasks for industrial manipulators. It is shown that many results which hold for open-loop trajectory generation, also generalize to output-feedback control.Development and analysis of a leader-follower control system for coordinated remote inspection with industrial manipulators. A gradient weighting scheme is proposed to fulfill joint limit constraints in combination with sub-task control such as collision avoidance. An experimental setup consisting of one robot carrying a stereoscopic camera and another robot which carries out remote maintenance tasks, is used to evaluate the effectiveness of the proposed control method.Development of a cooperative object-handling control algorithm for industrial robots with compliant human interaction. A variable admittance control scheme allows for intuitive human interaction while the two robots move their commonly held object. Experiments which detail the high performance of the control scheme is carried out using two fully redundant industrial manipulators.Development of a coordinated sewing control scheme designed for coordinated robot manipulators. Experiments show that leader-follower synchronization can be used to effectively eliminate position mismatches along the finished seam.A derivation of a new and computationally efficient Coriolis matrix factorization. The proposed Coriolis matrix factorization is due to its structure, especially useful in the theoretical analysis of a class of globally stable output-tracking controllers.