Simple rules to modify pre‐planned paths and improve gross robot motions associated with pick & place assembly tasks

Abstract

PurposeThis paper aims to describe real time improvements to the performance and trajectories of robots for which paths had already been planned by some means, automatic or otherwise. The techniques are applied to industrial robots during the gross motions associated with pick and place tasks. Simple rules for path improvement are described.Design/methodology/approachThe dynamics of the manipulator in closed form Lagrange equations are used to represent the dynamics by a set of second‐order coupled non‐linear differential equations. The form of these equations is exploited in an attempt to establish some simple rules. Sub‐optimal paths are improved by considering simple rules developed from the model of the machinery dynamics. By considering the physical limitations of the manipulator, performance was improved by refining pre‐calculated paths. Experiments were performed with a prototype robot and an old Puma 560 robot in a laboratory environment. Once the method had been tested successfully then experiments were conducted with a Kuka KR125 Robot at Ford Motor Company. The measured quantities for all the robots were drive currents to the motors (which represented the torques) and the joint angular positions.FindingsThe method of path refinement presented in this paper uses a simplified model of the robot dynamics to successfully improve the gross motions associated with a pick and place task. The advantage of using the input‐output form described was that intermediate non‐linearities (such as gear friction) and the motor characteristics were directly incorporated into the model.Research limitations/implicationsEven though many of the theoretical problems in manipulator dynamics have been solved, the question of how to best apply the theories to industrial manipulators is still being debated. In the work presented in this paper, information on system dynamics is used to produce simple rules for “path improvement”.Practical implicationsMost fast algorithms are for mobile robots and algorithms are scarcer for manipulators with revolute joints (the most popular type of industrial robot). This work presents real time methods that allow the robot to continue working while new global paths are automatically planned and improved as necessary.Originality/valueMotion planning for manipulators with many degrees of freedom is a complex task and research in this area has been mostly restricted to static environments, offline simulation or virtual environments. This research is applied in real time to industrial robots with revolute joints.