Method for controlling the movement of an industrial robot at and near singularities

A method for controlling an industrial robot such that a tool supported by the robot can follow a path determined by a number of consecutive points, and wherein the robot while following the path ends up near or at a singularity, said robot having a plurality of movement axes (A1-A6) and the configuration of the robot being determined by the angles of rotation (.theta..sub.1 -.theta..sub.6) of the movement axes, the ratio of the angular velocities of the axes to the velocity of the tool being given by a Jacobian matrix, the angles of rotation which the robot is to assume in order to obtain the correct position and orientation for the tool at the next point on the path being calculated as follows:

• a) setting start values of the angles of rotation,
• b) calculating the errors in the position and orientation of the tool in relation to the desired position and orientation for the start values,
• c) calculating the total error,
• d) calculating new start values, based on the errors in the position and orientation of the tool,
• e) allowing steps b-d to constitute an iteration and repeating it until the total error is lower than a tolerance value for the total error,
• characterized in that the iteration comprises the following steps:
• f) determining with the aid of the Jacobian matrix whether the robot is at or near a singularity,
• g) if the robot is at or near a singularity, carrying out the steps of:
• 1) identifying those axes which are singular and setting their joint angle values to their start values,
• 2) weighting the errors in the position of the tool in relation to the errors in the orientation of the tool by a weighting factor (V),
• 3) increasing the tolerance value for the total error by a tolerance factor (k) such that it exceeds the tolerance for the total error for points outside the singularities,
• 4) based on the weighted errors, calculating new start values for the next iteration.