Algorithmic processes for generating on-line drive transforms for robot manipulator.

摘要

Motion control schemes play major part in robotics. Joint and Cartesian motion schemes are two approaches in robot motion control. In the joint motion scheme, EOM positions are time coordinated with the joint variables. For implementation, EOM positions are transformed into joint variables instantaneously using appropriate inverse kinematic/kinetic models, whereas in the Cartesian motion scheme EOM positions are transformed into modified EOMs using straight line motions with transient (constant acceleration) and steady state (constant velocity) motions. Modified EOMs are then transformed into joint variables to feed joint actuators. One of the basic differences in joint motion and Cartesian motion scheme is that EOM is instantaneously converted into joint levels, whereas in the Cartesian motion scheme, EOM is modified before inversion. Objectives of this scheme are to avoid jerks and vibrations in the motion and to overcome obstacles and singularities. The major drawback is the evaluation time involved in generating modified EOM. Basic models for the drive transform are derived using two rotations and a translation, generally with PRR approach. Modified EOM in Cartesian motion scheme is primarily to address issues such as mode of travel and singularities.;The major work of this thesis is to: 1) Define a robot motion trajectory 2) Assign position on the trajectory for linear segmentation 3) Generate drive parameters for each segment 4) Generate modified motion points 5) Generate differential transforms, dT, and 6) Generate joint differential, dqi .