The Jet Propulsion Laboratory (JPL) is embarking on a series of exploration missions to Mars. Long-range rovers with several science instruments on board will traverse many kilometers on the surface. This paper describes some of the research underway at JPL as they pertain to navigation of these long-range rovers. A brief description is provided of the Sojourner rover which recently completed a short-range, near-lander Mars mission. A research rover called Rocky-7 under development at JPL implements a number of new technologies for long-range Mars traverses. The kinematic configuration of this vehicle is described as is tile sensor set. The paper then focuses on state estimation techniques used on-board Rocky-7. A brief review of rover-related state-estimation work is presented and an Extended Kalman Filter framework is introduced. This includes process models for translation, small-angle/quaternion based rotations, gyro bias and wheel contact as well as observation models for a gyro, sun-sensor, and accelerometer. A novel slip constraint technique is developed to allow incorporation into the filter of the highly non-linear kinematics resulting from the rover's rocker-bogey suspension mechanism and wheel-to-ground contact point variations. Simulation results are discussed to allow detailed comparisons with ground-truth data and experimental results from Rocky-7 are presented.

We present a general framework for modeling a class of mechanical systems for robotic manipulation, consisting of articulated limbs with redundant tendinous actuation and unilateral constraints. Such systems, that include biomorphically designed devices, are regarded as a collection of rigid bodies, inter-acting through connections that model both joints and contacts with virtual springs. Methods previously developed for the analysis of force distribution in multiple whole-limb manipulation are generalized to this broader class of mechanisms, and are shown to provide a basis for the control of co-contraction and internal forces that guarantee proper operation of the system. In particular, in the presence of constraints such as those due to limited friction between surfaces or object fragility, the choice of tendon tensions is crucial to the success of manipulation. An algorithm is described that allows to evaluate efficiently set-points for the control of tendon actuators that “optimally” (in a sense to be described) comply with the given constraints.

To limit cumulative errors due to the odometer, we propose an absolute location method based on the extraction and the use of geometrical landmarks. This method also avoids the need of prior knowledge of the robot environment (no training steps) and its specified fitting (no beacons). If different measurements of the environment around the robot taken by an ultrasonic telemeter are merged, geometrical primitives appear. They are then discriminated and rebuilt. Those primitives are the landmarks used by the locating system. Hence a set of definitions and laws have been established to conjointly use few landmarks, in order to obtain the mobile robot absolute location by triangulation.

In this paper, we deal with a localization system allowing one to determine the position and orientation of a mobile robot. This system uses active beacons distributed at the ceiling of the navigation area. These beacons can transmit a coded infrared signal which allows the robots to identify the sender. A CCD camera associated to an infrared receiver allows one to compute the position with a triangulation method which needs reduced processing time. Calibration and correcting distortion stages are performed to improve accuracy in the determination of the position. Dynamic localisation is established for most actual mobile robots used in indoor areas.

This paper introduces an event-based decentralized control scheme for the cooperation between multiple manipulators. This is in contrast to the common practice of using only centralized controls for such cooperation which, consequently, greatly limit the flexibility of robotic systems. The manipulators used in the present system are very simple with only two degrees of freedom, while even one of them is passive. Moreover these manipulators use very few and commonly available sensors only. Computer simulations indicated the applicability of the event-based decentralized control scheme for multi-manipulator cooperation, while real-life experimental implementation has proved that the proposed decentralized control scheme is fairly applicable for very simple and even under-actuated systems too. Hence, this work has opened new doors towards further research in this area. The proposed control scheme is expected to be equally applicable for any mobile or immobile multi-robotic system.

Draper Small Autonomous Aerial Vehicle (DSAAV), MITy and SMART micro-rovers, Companion mini-rover, and Vorticity Control Unmanned Undersea Vehicle (VCUUV) are highlighted. DSAAV demonstrated autonomy with GPS/INS integration and vision processing. The micro- and mini-rovers investigated ground based autonomy with extensive mapping and planning integration. VCUUV is a flexible-hull UUV which propels and maneuvers like a tuna.

This paper presents a novel method for ultrasound speed compensation in ultrasonic robot tracking. It employs optimization algorithms to estimate the best value of the ultrasound speed for every computation of the robot end-effector position. Compared with the other existing method for ultrasound speed compensation, this method has the advantage that no extra equipment is required to monitor the environment conditions (temperature etc.). The simulation, which is described at the end of this paper, shows that the real trajectory is extracted with very good precision from a set of distorted data caused by serious ultrasound speed drifting.

Most studies in the past on the control of biped locomotion considered only level surfaces. However, in the real world the ground is rarely completely flat. More research on locomotion on less structured surfaces is needed. In this study, we investigated a control design method that searches for suitable control and trajectory parameters using a Genetic Algorithm (GA). Many sets of parameters are generated through the search and the best set is selected based on a robustness measure developed from the linearized Poincaré map. This technique reduces tedious analysis and is favorably applicable to the design for locomotion on unstructured surfaces, for which analytical approaches are less appropriate. Simulations have been performed. Control parameters for different slopes were obtained and stored in a database. During the control, the control parameters suitable for the current surface slope were retrieved and the trajectory for a level surface was modified according to the surface slope. The control parameters changed values according to the terrain. Simulation results were promising.

Fuzzy control has been widely applied in industrial controls and domestic electrical equipment. The automatic learning of fuzzy rules is a key technique in fuzzy control. In this paper, a software development system for fuzzy control is presented. Since the learning of fuzzy rules can be seen as finding the best classifications of fuzzy memberships of input-output variables in a fuzzy controller, it can also be seen as the combination optimization of input-output fuzzy memberships. Multi-layer feedforward network and genetic algorithms (GA) can be used for the automatic learning of fuzzy rules. The algorithms and their characteristics are described. The software development system has been successfully used for the design of some fuzzy controllers.

This paper proposes a grasp planning for a multifingered hand-arm robot in the presence of obstacles. When the hand grasps an object, neither the hand nor the arm should collide with obstacles. However, it is very difficult to generate a collision-free grasp configuration of the hand-arm robot due to its abundant degrees of freedom of motion. To overcome this difficulty, we first decompose the hand-arm robot into the multifingered hand part and the arm part. We develop two methods of the grasp planning. In one method, a collision-free configuration of the hand part is planned first. Then, the arm configuration is planned so as to enable the hand configuration. In the other method, the collision-free configuration of the arm part is planned first and the multi-fingered hand configuration second. The best method is selected according to the space structure of the environment. Various heuristics are developed to improve the efficiency of the planning and implementation of the system. Simulation results are presented.

This paper describes the design of an Internal Model Control (IMC) system for a planar two-degree-of-freedom robot. IMC was investigated as an alternative to the basic inverse control scheme which is difficult to implement. The proposed IMC system consisted of a forward internal neural model of the robot, a neural controller and a conventional feedback controller, all of which were realised easily. Both the neural model and the neural controller were based on recurrent networks which were trained using the backpropagation (BP) algorithm. The paper presents the results obtained with two types of recurrent networks as well as a conventional PID system.

This paper addresses the link position setpoint control problem of n–link robotic manipulators with amplitude-limited control inputs. We design a global-asymptotic exact model knowledge controller and a semi-global asymptotic controller which adapts for parametric uncertainty. Explicit bounds for these controllers can be determined; hence, the required input torque can be calculated a priori so that actuator saturation can be avoided. We also illustrate how the proposed control algorithm in this paper can be slightly modified to produce a proportional-integral-derivative (PID) controller which contains a saturated integral term. Experimental results are provided to illustrate the improved performance of the proposed control strategy over a standard adaptive controller that has been artificially limited to account for torque saturation.

Manipulability ellipsoids are effective tools to perform task space analysis of robotic manipulators in terms of velocities, accelerations and forces at the end effector. In this paper a new definition of a dynamic manipulability ellipsoid for redundant manipulators is proposed which leads to more correct results in evaluating manipulator capabilities in terms of task-space accelerations. The case of manipulators in singular configurations is also analyzed. Two case studies are presented to illustrate the correctness of the proposed approach.

This paper describes an algorithm for generating a tripod turning gait which, when given an arbitrarily located turning centre, firstly maximises the rotation angle (angular stride) and then secondly optimises the stability. It does not require a specific walking machine layout or leg workspace shape, and it can deal with changes in the position of the CG caused by walking on a gradient or by uneven loading.

This paper mainly concerns a vibrotactile sensor capable of generating impulsive high-frequency signals for exclusively distinguishing a slip phase from various other contact phases. We describe the structure of the vibrotactile sensor with surface ridges and, after analyzing the slip signals, conclude that the output signal originates in the free vibration of the ridge. By using a two-fingered robot hand with the sensor attached to its compliant endtips, we verify that the vibrotactile sensor has a slip phase isolating function and demonstrate object regrip control in real time.

In future real-time systems such as those required for intelligent autonomous vehicle control, we need flexibility in choosing the set of services to support under varying environmental conditions and system states. It is not feasible to make an optimal choice of services at run-time, so we propose a method of ranking the services pre-run-time, based on the ‘utility' of each service. This paper focuses on the problem of calculating a ‘value' for the utility of each service alternative. We show how to derive values systematically and rationally, using Measurement Theory and Decision Analysis. The approach relies on engineering judgement and data input by a domain expert. In the context of autonomous vehicles, we believe that such knowledge would be available, making ‘value-based scheduling' a feasible approach.

In this paper, a new application of the ZMP (Zero Moment Point) control law is presented. The objective of this control method is to obtain a smooth and soft motion based on a real-time control. In the controller, the ZMP is treated as an actuating signal. The coordinates of the robot body are fed back to obtain its position. The proposed control method was applied on two different biped robots, and its validity is verified experimentally.

This paper describes a new vision-based playback method of a wheeled mobile robot. The path of the robot is divided into straight line parts and curved line parts. The curved line parts were approximated by a Bézier curve to decrease the volume of data. In order to track the curved line parts with good accuracy, tracking errors were detected by referring to two pre-recorded images on one part of the curve. Then, the errors were corrected by adjusting the control points of the Bézier curve. Outdoor experiments demonstrated good repeatability by using the proposed method.

This paper presents a new formulation of computing three-dimensional (3-D) frictional form-closure grasps of n robotic fingers. As 3-D form-closure grasps involve 6-D wrench space, we first propose a recursive reduction technique to transform the complicated problem in the 6-D space into a simpler 3-D one. Next, we rewrite the sufficient and necessary condition for form-closure grasps into its equivalent form of two sets of linear inequalities. Then, according to the linear inequality theory, the problem is transformed to searching for a set of points which ensure the inconsistency of each of the two linear inequality systems. To search for such points, we proposed two methods: The first one is based on testing whether the convex region formed by each linear inequality system is empty, while the second one relies on the potential field method. We have implemented the algorithm and confirmed their efficiency for the synthesis of 3-D form-closure grasps.

The problem of the control of the object cooperative manipulation during the work of multiple non-redundant six degrees-of-freedom manipulators is considered in this paper. The problem of the cooperative manipulation control is, like all its problems, solvable only if the system is considered as the elastic one, taking into account all existing constraints. The controlled system is with the output number greater than the available number of inputs, therefore, in the first stage the desired motions are selected from the set of the possible nominal ones, containing the trajectories of the manipulated object mass centre and slave manipulators contacts. Afterward, the classification of control tasks is performed. The procedure for the calculation of the driving torques introduced into the joints of the manipulators, necessary to obtain the nominal trajectory tracking, is proposed. The theoretical analysis of cooperative system closed loop behaviour is exposed, particular attention being paid to the uncontrolled variables. The procedure is illustrated on the example of the simple closed loop cooperative system, consisting of the manipulated object and two one degree-of-freedom manipulators. For this system, the behaviour is determined and the driving torques are calculated.