Multiple arm systems have been introduced as offering enhanced capabilities over single arm designs. This is achieved only
if an effective coordination of their actions is accomplished. A symmetric kineto-static formulation that regards the system
composed by the manipulators and the held object as one integrated system proves very adequate for modelling purposes.
Manipulability ellipsoids have been defined to perform task space analysis and reconfiguration of multiarm systems.
Kineto-static ellipsoids have been introduced which are based on the mappings relating joint velocities to absolute and
relative velocities, and dually external and internal forces to joint torques. In the case when the dynamics of manipulators
and object is not negligible, a dynamic ellipsoid has been defined which is based on the mapping between joint driving
torques and object accelerations, taking into account the actuator torque limits and gravitational loads. A dexterous
reconfiguration scheme based on the use of the above ellipsoids has been proposed for two planar cooperating robots.
Regarding the control problem for multiarm systems, a nonlinear dynamic decoupling technique has been developed; the
goal is to achieve control of absolute object motion combined with control of internal forces between pairs of end effectors.
Furthermore, a computationally simpler control strategy has been proposed that assigns an impedance behavior to the
controlled variables and does not require force sensor measurements. A task-oriented formulation for two cooperative
spatial manipulators has been introduced which allows a straightforward description of general coordinated motion tasks in
terms of absolute and relative variables. An effective inverse kinematics scheme has been devised that exploits the above
formulation where the task Jacobians are expressed in terms of the single manipulator Jacobians. The scheme is extended to
handle the presence of redundant degrees of freedom in the system. The cases of rolling and sliding contact between the end
effectors and a commonly held object are handled with a minimum reformulation effort. The output of the inverse
kinematics algorithm have been used as reference inputs for a joint-space regulation scheme based on the concept of
kinetostatic filtering of internal forces at steady state. Alternatively, a task-space regulation scheme has been proposed,
where frame orientation is described in terms of unit quaternions so as to avoid representation singularities. Experimental
tests of kinematic control and operational space control have been conducted on the cooperative robot manipulator system
in the laboratory.