基于视觉测量与神经网络的工业机器人位姿补偿

为提高六轴工业机器人的绝对定位精度,本文提出了一种利用视觉测量数据通过ELM(Extreme Learning Machine)神经网络实现机器人位姿补偿的新方法.利用固定在机器人末端的手眼相机获取机器人的末端位姿,并借助ELM实现机器人末端执行器从目标位姿到预测指令位姿之间映射,用修正转角代替原转角使机器人末端执行器运行至修正位姿,实现补偿.特别的是,对于使用的ELM,以网络预测均方误差为指标定量选取了网络的最佳参数.相比之前的方法,本文提出的算法具有能够同时高精度补偿姿态角及位置误差的显著优点.为验证该位姿误差补偿方法的有效性,本文进行了实验验证.结果表明,相比较于未补偿前的机器人末端位姿误差,经该方法补偿后的位姿误差被稳定控制在较低水平,平均位置误差降低89.1%;平均姿态角误差降低96.8%.除此以外,位置误差与姿态角误差的标准差也分别降低了85.66%和93.24%.

Potassium-induced bifurcations and chaos of firing patterns observed from biological experiment on a neural pacemaker

Changes of neural firing patterns and transitions between firing patterns induced by the introduction of external stimulation or adjustment of biological parameter have been demonstrated to play key roles in information coding.In this paper,bifurcation processes of bursting patterns were observed from an experimental neural pacemaker,through the adjustment of potassium parameter including ion concentration and calcium-dependent channel conductance.The adjustment of calcium-dependent potassium channel conductance was achieved by changing the extracellular tetraethylammonium concentration.The deterministic dynamics of chaotic bursting patterns induced by period-doubling bifurcation and intermittency,and lying between two periodic bursting patterns in a period-adding bifurcation process was investigated with a nonlinear prediction method.The bifurcations included period-doubling and period-adding bifurcations of bursting patterns.The experimental bifurcations and chaos closely matched those previously simulated in the theoretical neuronal model by adjusting potassium parameter,which demonstrated the simulation results of the theoretical model.The experimental results indicate that the potassium concentration and conductance of calcium-dependent potassium channel can induce bifurcations of the neural firing patterns.The potential role of these bifurcation structures in neural information coding mechanism is discussed.

Stochastic dynamics of conduction failure of action potential along nerve fiber with Hopf bifurcation

Action potentials can be induced by external electronic impulsive stimulations applied at one end of the unmyelinated fibers(C-fibers), while some action potentials fail to conduct to the other end of the fiber when the stimulation frequency becomes high. Such a phenomenon is called as conduction failure, which was observed in the biological experiments and related to the painful diabetic neuropathy, inflammation, and trauma in the previous studies. On-off firing pattern was recorded from the fiber when conduction failure happened. In the present study, the diffusion Hodgkin-Huxley(HH) model with resting state near a Hopf bifurcation is adopted to simulate the experimental observations. When the periodic electrical pulses with high frequency are applied to one end of the fiber described by the deterministic HH model, conduction failure and the corresponding firing patterns different from the on-off firing pattern are simulated. When noise is introduced to form the stochastic HH model, the firing pattern corresponding to conduction failure becomes the on-off firing pattern, which is characterised by transition behaviors between on-phase(continuous action potentials) and off-phase(a long quiescent state) and large variations in the durations of both phases. Furthermore, the increase of potassium conductance can enhance the conduction failure degree, which closely matches those observed in the experiment and is suggested to be related to the reduction of pain signals. The results show that noise is an important factor to evoke the on-off firing pattern, reveal the functional capability in the pain signals propagation along C-fiber,and present a possible measure for the treatment of chronic pain.

The nonlinear mechanism for the same responses of neuronal bursting to opposite self-feedback modulations of autapse

In the traditional viewpoint,inhibitory and excitatory effects always induce opposite responses.In the present study,the enhanced bursting activities induced by excitatory autapses,which are consistent with the recent experimental observations,and those induced by inhibitory autapses,which is a paradoxical phenomenon,were simulated using the Chay model.The same bifurcations and different ionic currents for the same responses were acquired with fast-slow variable dissection and current decomposition,respectively.As the inhibitory or excitatory autaptic conductance increased,the ending phase of the burst related to a homoclinic bifurcation of the fast subsystem changed to widen the burst duration to contain more spikes,which was induced by an elevated minimal potential(V_(min)) of spiking of the fast subsystem.Larger inhibitory and excitatory autaptic conductances induced smaller and larger maximal potentials(V_(max)) of spiking,respectively.During the downstroke,a weaker potassium current induced by the smaller V_(max) played a dominant role for the inhibitory autapse,and the stronger potassium current induced by the larger V_(max) became weaker due to the opposite autaptic current of the excitatory autapse,which induced the V_(min) elevated.The results present the nonlinear and biophysical mechanisms of the same responses to opposite effects,which extends nonlinear dynamics knowledge and provides potential modulation measures for the nervous system.