A memory-based robotic control paradigm which learns relationships between a control effort and a change of state is introduced. It has been used to develop a learning control system which implements step responses in...
详细信息
A memory-based robotic control paradigm which learns relationships between a control effort and a change of state is introduced. It has been used to develop a learning control system which implements step responses in one dimension on a robotic gripper, with partial success. It was found that velocity as well as positional feedback were required to complete even simple movements. It is believed that aspects of this approach would readily extend to a tactile sensing system.< >
Trajectory planning methods and control algorithms for biped robot steady walking are discussed. The model following plus variable structure systems technique is developed for the control of dynamic walking in the sag...
详细信息
Trajectory planning methods and control algorithms for biped robot steady walking are discussed. The model following plus variable structure systems technique is developed for the control of dynamic walking in the sagittal plane. The nonlinear feedback plus modified alpha -computed torque technique is introduced for walking in the frontal plane. simulation results are provided for level walking and ascending on staircases for a five-link biped robot. These results indicate that the proposed algorithms could achieve robust trajectory tracking even in the presence of system modeling errors. Specifically, simulation results show that, even in the presence of disturbances, gait stability can be achieved within two steps after the start of walking. In addition, the biped robot can walk with prespecified patterns.< >
A system is presented which automatically constructs an assembly partial-order graph from an object-oriented model describing parts and their connections. The construction is based on the principle of assembly by disa...
详细信息
A system is presented which automatically constructs an assembly partial-order graph from an object-oriented model describing parts and their connections. The construction is based on the principle of assembly by disassembly and on assembly heuristics representing criteria for preferred subassemblies. The system first identifies disassemblable parts and subassemblies by reasoning geometric and physical constraints as well as resource requirements, and then assigns preference to the identified disassembly options based on assembly heuristics. The recursive application of this process to the selected disassemblies results in a hierarchical partial-order graph (HPOG). The HPOG not only specifies the required precedence in part assembly but also provides parallelism for implementing multiple-robot distributed and cooperative assembly. The software is organized under the cooperative-problem-solving paradigm.< >
The authors present a learning expert system which enables a robot to acquire fine motion skills automatically. The system follows the paradigm of Expert Assisted Robot Skill Acquisition (EARSA) proposed by the author...
详细信息
The authors present a learning expert system which enables a robot to acquire fine motion skills automatically. The system follows the paradigm of Expert Assisted Robot Skill Acquisition (EARSA) proposed by the authors (1987). EARSA is mainly concerned with the self-discovery of skills by a robot in conjunction with the transfer of human skills to a robot and emphasizes the distinctive difference in perceptual and physical capabilities between a human and a robot. The authors review the theory and mechanism of EARSA, describe the robot fine motion skill learning algorithm formulated on the basis of EARSA, and present the details of simulation on the robot learning of two-dimensional peg-hole insertion skills. The results of simulation indicate the dramatic improvement of performance as a result of skill learning.< >
Quasistatic mechanical systems, in which mass or acceleration is sufficiently small for the inertial term ma in F=ma to be negligible compared to dissipative forces, are discussed. It is pointed out that many instance...
详细信息
Quasistatic mechanical systems, in which mass or acceleration is sufficiently small for the inertial term ma in F=ma to be negligible compared to dissipative forces, are discussed. It is pointed out that many instances of robotic manipulation can be well approximated as quasistatic systems, with the dissipative force being dry friction. Energetic formulations of Newton's laws have often been found useful in the solution of mechanics problems involving multiple constraints. An intuitive minimum power principle is outlined which states that a system chooses at every instant the lowest-energy, or 'easiest', motion in conformity with the constraints. Surprisingly, the principle is in general false; but it is proved that the principle is correct in the useful special case that Coulomb friction is the only dissipative or velocity-dependent force acting in the system.< >
Real-time expert system techniques and applications to robot manipulator control systems are discussed. A novel type of intelligent controller structure, the expert learning controller prototype ELEC (expert learning ...
详细信息
Real-time expert system techniques and applications to robot manipulator control systems are discussed. A novel type of intelligent controller structure, the expert learning controller prototype ELEC (expert learning controller), is developed for the trajectory tracking control in repeat operations. ELEC, acting as an intelligent real-time controller in a closed-loop system, can modify the control series in a human-expert-like manner using the experience of previous operations in order to force the system output to converge to the prespecified desired trajectory. ELEC does not require the knowledge of system models, so it can be used in a fairly wide range of control problems. A numerical example for a two-link robot manipulator is given which shows the satisfactory performance of ELEC.< >
The fabrication of shallow p–n junction impurity profiles using Ga and In focused ion beam (FIB) implantation in conjunction with rapid thermal annealing is reported. A 75‐keV focused ion beam with 80 mA/cm2 current...
The fabrication of shallow p–n junction impurity profiles using Ga and In focused ion beam (FIB) implantation in conjunction with rapid thermal annealing is reported. A 75‐keV focused ion beam with 80 mA/cm2 current density and 0.5‐μm beam diameter was used to implant Ga+ and In+ into (100) Si substrates at doses ranging from 1×1013 to 5×1015/cm2. The annealing temperature was varied from 600 to 1000 °C for various times, 10–30 s. Secondary ion mass spectrometry and spreading resistance profiling were used to measure the implanted species atomic and carrier concentration depth profiles. As compared to the conventional broad beam implantation, the FIB implanted Ga concentration depth profiles exhibit higher activation percentage in the end‐of‐range and tail region. This results in thicker layers and deeper p+–n junctions, but with lower sheet resistance and leakage current. A similar phenomenon is also observed in In FIB‐implanted samples. The damage generated by the high dose rate implantation is possibly responsible for this phenomenon. p+–n diodes fabricated using Ga FIB doping exhibit very good diode electrical characteristics: ideality factor as low as 1.01, leakage current density below 1 nA/cm2 at −1 V reverse bias, and breakdown voltage of ∼35 V. Diodes fabrication with In generally exhibit shallower junctions, but higher leakage current. Bipolar transistors have been fabricated using FIB‐implanted Ga and In thin base layers. Common emitter current gain of 70 and 10 were obtained for Ga and In FIB‐implanted bipolar transistors, respectively.
A paradigm for optimum estimation of fused multiple sensor data is developed in order to best use the sensor information in the time evolving environment. Two basic approaches have been developed: dynamic moving quadr...
详细信息
A paradigm for optimum estimation of fused multiple sensor data is developed in order to best use the sensor information in the time evolving environment. Two basic approaches have been developed: dynamic moving quadratic curve fitting and weighted least mean square error. These two approaches are shown to be advantageous in terms of accuracy, speed, and versatility. The theoretical frameworks presented are supported by sets of simulation data.< >
暂无评论