The International Journal of Robotics Research

Symmetry can simplify the control of dynamic legged sys tems. In this paper, the symmetries studied describe motion of the body and legs in terms of even and odd functions of time. A single set of equations describes symmetric running for systems with any number of legs and for a wide range of gaits. Techniques based on symmetry have been used in laboratory experiments to control machines that run on one, two, and four legs. In addition to simplifying the control of legged machines, symmetry may help us to understand legged locomotion in animals. Data from a cat trotting and galloping on a treadmill and from a human running on a track conform reasonably well to the predicted symmetries.

For practical reasons, compliant materials are often used on the gripping surfaces of robotic hands. Such materials are not well described by the Coulomb friction law or by simple point-contact or line-contact kinematics. In this paper, a shearing model is used to describe the contact friction. Models ofpointed, curved, flat, soft, and soft-curved fingertips are then developed and compared in terms of their contribu tion to the stiffness and stability of a simple grasp. There is a spectrum of contact conditions defined by the fingertip radius and contact area relative to the object size. This spectrum provides insights for designing and controlling robotic fingers.

In this paper, we describe a technique for geometrically hash ing two-dimensional model objects. Used in conjunction with other methods for recognizing partially obscured and over lapping objects, this technique enables us to recognize over lapping, two-dimensional objects selected from large data bases of model objects without significant performance degradation when the database is enlarged. This technique is based on use of a synthetic attribute of an object, which we will call its footprint. Experimental results from databases of up to 100 objects are presented.

This paper describes a general method for estimating the nominal relationship and expected error (covariance) between coordinate frames representing the relative locations of ob jects. The frames may be known only indirectly through a series of spatial relationships, each with its associated error, arising from diverse causes, including positioning errors, measurement errors, or tolerances in part dimensions. This estimation method can be used to answer such questions as whether a camera attached to a robot is likely to have a particular reference object in its field of view. The calculated estimates agree well with those from an independent Monte Carlo simulation. The method makes it possible to decide in advance whether an uncertain relationship is known accu rately enough for some task and, if not, how much of an improvement in locational knowledge a proposed sensor will provide. The method presented can be generalized to six degrees offreedom and provides a practical means of esti mating the relationships ( position and orientation) among objects, as well as estimating the uncertainty associated with the relationships.

This paper studies two different approaches to the inverse- kinematics problem for a six-degree-of-freedom robot manip ulator having three revolute joint axes intersecting at the wrist. One method uses three rotational generalized coordi nates to describe the orientation of the body. The other method uses equivalent Euler parameters with one constraint equation. These two approaches have been incorporated into two different computer algorithms, and the results from each are compared on the basis of computational complexity, time simulation, singularity, etc. It was found that Euler parame ters were less efficient than three rotational angles for solving the inverse-kinematics problem of the robot considered, and that the physical singularities caused by the robot mecha nism could not be eliminated by using either of the two ap proaches.

A mobile robot work center has been constructed for the Navy by Southwest Research Institute for the purpose of re moving rivets from airplane wings during overhaul and re pair. This system is referred to as a deriveter. Two features of the robotic deriveter system required the derivation of new mathematical relationships. The first feature is the computa tion of the transformation relating the robot and wing refer ence frames to each other based on measurements made by the robot, which is mounted on a movable vehicle. Four methods for finding this transformation are presented and the applicability of each to the deriveter is discussed. The second feature is that of positioning the end-effector, tool cen terline so that it is perpendicular to the wing surface, with the tool center point at a given offset from the wing based on measurements made by the robot end effector.

In this paper, we present a general purpose manipulator control system. The system is run under the Unix operating system. Manipulator programs are written in the "C" lan guage and make use of primitive functions included in a library. Manipulator control is thus integrated within the lan guage in the same manner as is input-output. The system includes a world modeler and a trajectory generator that are accessed through two sets of primitive functions. The sys tem's structured world modeler is designed for an easy inte gration of sensors. The first part of the paper reviews the functional organization of the system, going through world modeling, trajectory generation, force control, and synchroni zation. The second part describes actual robot programming examples.

The redundancy of robot manipulators plays an important role in increasing their flexibility and versatility, especially in an environment with obstacles. In this paper, the redundancy is utilized for enabling collision-free motion for a given tra jectory of a manipulator hand. The calculation of the corre sponding trajectory in the space of joint coordinates is per formed using several performance criteria that take into account the vicinity of the obstacles. The advantages and shortcomings of the proposed algorithms are discussed. The implementation of the algorithms is real-time due to modest computational complexity.