In this work, we consider the problem of 2 robots handling a rigid object, while model-parameter uncertainties are assumed. It is also assumed that the manipulators can push but not pull the object. Several control schemes proposed in the literature attempt to control the position of the object rather than its orientation. However, many industrial tasks require to move and to rotate the object. To this end, we propose an adaptive hybrid position/force control law based on time-variant holonomic constraints, which allow for object position and orientation control. Our approach guarantees that the force error asymptotically converges to 0; therefore, a stable grasp can be accomplished by means of a proper definition of the desired pushing force. In addition, a dynamic model of the cooperative system based on the load distribution and joint-space orthogonalization principles is developed. Experimental results are presented to validate the proposed dynamic model and control scheme.Consider the following auxiliary theorem: Theorem 2. 31 Let ⊂ ℜ n be a domain containing y = 0 and suppose f(t, y) is piecewise continuous in t and locally Lipschitz in y,J 1 (2, 1) = 11 c(q 11 ), J 1 (2, 2) = 21 c(q 21 ),For the matrix J 2 = J c2 J m2 , we haveJ 2 (2, 1) = 21 c(q 21 ) + 22 c(q 22 ) + ( 32 − r cx2 )c( 2 ) + r cy2 s( 2 ), J 2 (2, 2) = 22 c(q 22 ) + ( 32 − r cx2 )c( 2 ) + r cy2 s( 2 ), J 2 (2, 3) = ( 32 − r cx2 )c( 2 ) + r cy2 s( 2 ), J 2 (3, 1) = 1, J 2 (3, 2) = 1, J 2 (3, 3) = 1.
