Precise motion control of a robot by controlling its various robotic manipulators may be organized by the required level of task specification. The levels include object-level control, which describes the ability to control the behavior of an object held in a single or a cooperative grasp of the robot; end-effector control, which is control of the various manipulators such as robotic fingers and thumbs; and joint-level control. Collectively, the various control levels achieve the required mobility, dexterity, and work task-related functionality.
This invention relates to the automatic control of a force- or impedance-controlled robot in an operating environment in which a robotic manipulator of the robot may come into unexpected direct physical contact with an object and/or an operator within its workspace.
A method of controlling a robotic manipulator of a force-or impedance-controlled robot within an unstructured workspace includes imposing a saturation limit on a static force applied by the manipulator to its surrounding environment. Should the robotic manipulator unexpectedly contact an object in its workspace, the static force applied by the manipulator is limited. Upon contact, the robotic manipulator proceeds with its assigned task without “fighting” through the object after the object is encountered.
The control strategy also may include automatically executing a predetermined dynamic reflex upon a threshold contact force with the object. The dynamic reflex alleviates any inertial impulse of the contact that is not already addressed by the saturation-limited static force. The dynamic reflex can rely on either dynamic modeling or exteroceptive sensing in order to detect the contact.
The controller provides a control strategy for the robot using an algorithm. The control strategy provides a static compliance phase at all times, and may also provide an additional dynamic reflex phase to help ensure the safe operation of the robot. The controller may be a server or host machine with one or more microprocessors or CPUs. Individual control algorithms resident in the controller may be stored in ROM (read only memory) or other suitable memory and automatically executed to provide the control functionality.
This work was done by Philip Strawser of Johnson Space Center; John Yamokoski and Brian Hargrave of Oceaneering Space Systems; and Muhammad Abdallah of General Motors Corp. MSC-25121-1