Browsing by Author "Deo, Arati Suresh"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Application of optimal damped least-squares method to inverse kinematics of robotic manipulators
    (1991) Deo, Arati Suresh; Walker, Ian D.
    Inverse kinematics for redundant robotic manipulators is typically computed using the pseudoinverse of the manipulator Jacobian. Though the pseudoinverse yields the most accurate minimum-norm joint velocity vector, it fails to prevent high joint velocities when the manipulator is in the neighborhood of singularities. The Singularity Robust Inverse (SRI), which arises from the Damped Least-Squares technique proves to be a better inverse kinematic solution than the pseudoinverse near singular configurations. The SRI computes damped joint velocities but causes some deviation of the end-effector from its planned trajectory. The SRI performs most satisfactorily when the damping factor, which represents the trade-off between the accuracy and feasibility of the computed solution is calculated optimally, i.e. it yields minimum deviation of the end-effector while ensuring the feasibility of the joint velocities at all points in the manipulator workspace. This work introduces a new method of computing the optimal damping factor. The SRI can also be used to optimize another sub-task criterion in addition to performing the main motion task and avoiding high joint velocities at singularities.
  • Thumbnail Image
    Item
    Inverse kinematics and dynamic control methods for robotic systems
    (1995) Deo, Arati Suresh; Walker, Ian D.
    This dissertation presents new algorithms for inverse kinematic computations of robotic manipulators and for the control of multiple cooperating manipulator systems. The results presented in this thesis can be classified into three parts. The first part is an extension of our earlier work in computing inverse kinematic solutions using the damped least squares method. An adaptive algorithm is presented, which switches from the damped least-squares model to a second-order model, in situations where the former is unable to converge to the desired configuration. This algorithm is insensitive to the reachability of the desired end-effector position. The second part introduces minimum-effort inverse kinematics for redundant robotic manipulators. The Euclidean norm has been universally used in optimizing various criteria for computing the joint velocities of a redundant arm. Here, the use of the infinity norm for defining these criteria is investigated. It is shown that in various applications, better physical representation of the performance criteria is obtained by using this norm instead of the Euclidean norm. The third section of the thesis deals with the formulation of dynamic equations and control law for a multiple cooperating manipulator system handling a common object, when the surfaces of the end-effectors and the object maintain rolling contact with each other. A new unified dynamic formulation for such a robotic system is derived, by modeling the rolling contacts as unactuated joints of the manipulators. This enables the formulation of trajectory planning methods that can be used to perform an additional subtask such as collision avoidance. In addition, a computed-torque type control law is designed, which explicitly controls the object trajectory, object internal forces and the contact trajectories.