This paper presents the results and subsequent analysis of full-scale excavation experiments aimed at developing a practical understanding of how actuator forces evolve during excavation and how they relate to the interactions that occur between an excavator's end-effector and its environment. Our focus is on the excavation problem for fragmented rock, as is common in mining and construction applications. Based on an analysis of the experimental data, an example admittance-type autonomous excavation controller is postulated.

This paper presents a new approach to the problem of control synthesis and motion planning for compliant motions of manipulators. The scheme is based on decoupled linear impedance models of the robot and the environment, in which the errors due to control, task frame estimation, as well as environment uncertainty are explicitly taken into account. The technique consists of characterizing the position, velocity and force responses of the decoupled models, and expressing the task goals as constraints on these expressions. The impedance parameters of the controller are found by fitting uncertainty boxes inside the region of the parameter space that satisfies all task constraints. An experimental verification of this method is presented for the task of closing a circuit breaker.>

The motion of the bucket in an LHD (load-haul-dump) loader and the required forces in the hydraulic cylinder, and the pushing force during a loading operation are studied. Based on the required force/torque load at the front edge of the bucket and its configuration (height and orientation angle), these forces can be determined by considering the loading mechanism as a slow manipulator (static loads) through the Jacobian matrix. The Jacobian of the manipulator was analytically found in terms of the physical dimensions and the measurable angles, considering the joint variables for manipulation, which can also be used for velocity relations, when necessary. It is pointed out that the results of this work combined with studies on rock mechanics can provide some means of automatic loading action for a system equipped with pressure sensors to monitor the forces of the actuators.>

Traditional software design methodologies have been shown to have drawbacks in designing and implementing software systems for robotics. A novel dual-hierarchical object-oriented design methodology is presented, which is well suited to problems of this type. A practical example of the application of this methodology is presented, utilizing CLOS as the implementation vehicle. The methodology developed is shown to facilitate the programming and planning of complex robot tasks, and the provision of generic recovery procedures for exception handling.>

A qualitative geometric reasoning approach to the analysis of robot task kinematics is presented. Qualitative geometric reasoning refers to the process of deriving a solution based on a set of geometric rules in which the concepts are described in terms of symbolic qualitative values and partial-ordering relations. The qualitative kinematic analysis scheme permits computationally efficient approximate solutions of kinematic problems without using exact information on the geometry of mechanisms under consideration, and provides a basis for the functional simulation of mechanisms.>

The reduction of the dependence of the autonomous system on the human operator when recovering from error states is addressed. The focus is on the problem of formulating error recovery strategies, rather than that of the overall planning of complex tasks. A novel approach that utilizes explanation-Xased learning as a framework for the acquisition of error recovery knowledge is presented.>

Presents a qualitative reasoning approach to synthesizing robot motion plans from task-level specifications. This method relies on representing the knowledge about robot manipulation as a task grammar augmented with a domain-dependent task analyzer. The task analyzer tests constraints for deriving suboperations pertaining to a higher-level operation and performs commonsense reasoning about the manipulator-level motion strategies. Through several examples, the authors show how syntactical knowledge about robot operations can be formulated for real-world manipulation tasks and how commonsense knowledge can be augmented to constrain the syntactical reasoning.>

For pt.I see ibid., p.487-93 (1991). As augmentations to syntactical task planning, kinematic models facilitate the specification of robot-level motion strategies for mechanism-oriented manipulation tasks. The paper presents a qualitative geometric reasoning approach to the analysis of robot task kinematics. The qualitative approach permits computationally efficient, approximate, solution of kinematic problems without using exact information on the geometry of mechanisms and provides a basis for the functional simulation of robot manipulation. >

An investigation is reported of the design and implementation of a compliant control scheme based on damping control that can be used to perform position and force control of robot manipulators in teleoperation or autonomous control. The dynamic performance of such a system in constrained situations is very dependent on the environmental parameters. An adaptive controlled based on the model reference adaptive controller (MRAC) approach is developed for a single-axis manipulator model and is further extended to the multiaxis case using the concept of directional adaptation: the MRAC and the compliance are active only in the direction of the force constraints, thus allowing near-perfect velocity and force tracking in all directions. Simulation results show that the adaptive scheme can significantly improve the performance in force tracking and enhance stability at high stiffness by rendering the behavior independent of the environment and robot parameters. Some experimental implementation concerns are presented in light of a series of preliminary tests with a PUMA 560 manipulator.>