The Autonomous Robot Architecture (AuRA) provides multi-level representation and planning capabilities. This paper addresses the task of navigational path-planning, which provides the robot with a path guaranteed to be free of collisions with any modeled obstacles. Knowledge supporting visual perception can also be embedded, facilitating the actual path traversal by the vehicle.

A multi-level representation and architecture to support multi-sensor navigation (predominantly visual) are described. A hybrid vertex-graph free-space representation based upon the decomposition of free space into convex regions capable for use in both indoor and limited outdoor navigation is discussed. This “meadow map” is produced via the recursive decomposition of the initial bounding area of traversability and its associated modeled obstacles. Of particular interest is the ability to handle diverse terrain types (sidewalks, grass, gravel, etc.) “Transition zones” ease the passage of the robot from one terrain type to another.

The navigational planner that utilizes the data available in the above representational scheme is described. An A* search algorithm incorporates appropriate cost functions for multi-terrain navigation. Consideration is given to just what constitutes an “optimal” path in this context.

This paper considers theoretically and experimentally the control problem of the biped locomotion system on the double support phase. The motion on the double support phase is described by the constrained dynamic system with two static constraints. The reduced order equations are derived by eliminating Lagrange's multiplier from the constrained equations, and a control algorithm for stabilizing the motion is obtained in reduced space. Finally, this algorithm is applied to an actual robot constructed in our laboratory and its usefulness is shown.

The pulsed time-of-flight laser rangefinding technique has been used in many industrial measurement applications, including 3D-coordinate measuring devices, hot surface profilers and mobile robot sensors. Optical fibres, typically 1–10 m in length and 100–400 μm in diameter can be used to guide optical pulses to the separate sensing head of the measurement device. The use of a large multimode fibre may cause problems, however, when aiming at millimetre accuracy, as the construction and adjustment of the optics of the sensor head may affect the transit time linearity and measurement accuracy via multimode dispersion. Environmental effects, such as bending, vibration due to the moving sensing head and temperature, also cause measurement errors. The error sources are studied and characterized in this paper.

This paper presents a method for automatic forming and solving dynamic equations of motion of manipulator with elastic joints and rigid segments. For the minimal configuration of a cylindrical manipulator the numerical integration of these equations is performed, and vibrations in the joints are obtained as dynamic errors with respect to the prescribed trajectory.

In this paper the authors present a mathematical model of geometric and kinematic behaviour of an original passive compliant device provided with two rotation centres, called DCR–LAI system. This device is designed for a robotic assembly of parts with very small tolerances including a chamfer at the hole. The given modélisation may be used as a decision aid for the choice of a compliant device with regard to characteristics of parts being assembled.

After the enthusiasm for creating “intelligent robots” in the early 1980's, progress of robotics research in the past decade has not fulfilled our expectations but revealed various difficulties in understanding motor control by man and implementing intelligent functions in robotic machines. To regain the initiative in the development of intelligent machines, this paper first presents a critical review of the state of the art of robot control and points out the necessity for improving robot servo-loops in order to facilitate skilled and dexterious motions in robotic manipulators and mechanical hands. It is then shown that the introduction of a quasi-natural potential in Lagrange's formulation of robot dynamics gives rise to the design of hyperstable PID servo-loops, which establish global asymptotic stability of set-point control. The hyperstability theoretical framework is then applied to the design of control commands in various control problems, such as hybrid (position/force) control, impedance control, model-based adaptive control, and learning control. In all cases, the passivity concept of residual robot dynamics plays a vital role in conjunction with the concept of feedback connections of two hyperstable nonlinear or linear blocks.

In this paper a pneumatic vibratory wrist operated with a PWM controller is developed for robotic assembly. In the vibratory assembly system, the vibratory wrist can perform random search motion of a hole to compensate the position error at the early stage of an insertion process. Since the vibration characteristics of the wrist, such as the amplitude and trajectory of the vibration, are critical to assembly performance, they are experimentally investigated for various system controller parameters. In addition, a series of insertion experiments are performed to evaluate the assembly performance of the proposed wrist. The results show that within a wide range of operating conditions the wrist vibration can effectively compensate for large positioning errors when this wrist is used for a chamferless peg-in-hole task.

A method for solving the recognition of partially occluded parts is presented. It is based on the automatic generation of features from a set of primitive features which are configurations of pairs of fixed length segments of boundary edges of the parts. The procedure that creates the recognition features assigns a number in the range (0,1) that indicates the importance of the feature in the recognition strategy. This number is referred to as the feature's saliency. The method assumes that the parts that can occur in a scene come from a known set of parts. An example illustrates how automatically generated features can be used to count the number of identical parts in a heap.

Position estimation is a key issue for an ALV Autonomous Land Vehicle) in navigating a mountainous area. The unevenness of the terrain makes mechanical velocity sensors inaccurate (due to wheel slippage), and the lack of appropriate landmarks complicates the problem. In this paper, we present a solution method using features of the skyline. The skyline from the vision system is assumed given, and compared with a computer map, called the CAD-MAP. The algorithm is composed of: a) Identification of the peak points in the camera skyline, b) Computing the ALV position for the identified peak points, and c) Searching for the corresponding peak point in the CAD-MAP. Heuristics for computational efficiency and solution accuracy are also included in the algorithm. To test the validity and effectiveness of the algorithm, numerous simulations were performed and analyzed.

This paper argues the case for extracting as complete a set of sensory data as practicable from scenes consisting of complex assemblages of objects with the goal of completing the task of scene analysis, including placement, pose, identity and relationship amongst the components in a robust manner which supports goal directed robotic action, including collision-free trajectory planning, grip site location and manipulation of selected object classes.

The emphasis of the paper is that of sensor fusion of range and surface colour data including preliminary results in proximity, surface normal directionality and colour based scene segmentation through semantic-free clustering processes. The larger context is that of imbedding the results of such analysis in a graphics world containing an articulated robotic manipulator and of carrying out experiments in that world prior to replication of safe manipulation sequences in the real world.

This paper presents a general method to identify the geometric parameters of robots. An algorithm is given to calculate the identifiable geometric parameters. The robot location and the tool location parameters are taken into account. The algorithm is generalized to tree structure robots. The problem of selecting the optimum robot configurations to be used during the identification is discussed and a solution is proposed.