Intelligent trajectory control using recurrent averaging learning

In this article, robotic trajectory control using artificial intelligence techniques is developed. The learning strategy is called recurrent averaging learning. It takes the average of initial states and final states after a cycle of training and sets this value as the new initial and final states for next training cycle. A three-layer neural network is used as a controller, it provides the control signals in each stage of a walking gait. A linearized inverse biped model is derived. This model calculates the error signals that will be used to back propagate to the controller in each stage. Through learning, the robot can develop skills to walk along a predefined path with specified step length, walking speed, and crossing clearance. This proposed scheme is tested with simulations of the BLR-G1 walking robot on horizontal and sloping surfaces.     

Fuzzy modelling control for aircraft automatic landing system

Aircraft landing control based on fuzzy modelling networks is presented. The proposed scheme uses a fuzzy controller combined with a linearized inverse aircraft model. A multi-layered fuzzy neural network is used as the controller, providing the control signals at each stage of the aircraft-landing phase. The algorithm used to train the network is the Backpropagation Through Time. The linearized inverse aircraft model provides the error signals that will be used to back-propagate through the controller at each stage. The objective of this study is to improve the performance of conventional automatic landing systems. The simulation results are described for the automatic landing system of a commercial aeroplane. Tracking performance and robustness are demonstrated through software simulations. Simulation results show that the fuzzy controller can successfully expand the safety envelope to include more hostile environments such as severe turbulence.     

Hybrid-state driven autonomous control for planar bipedal locomotion

The focus of this paper is on the development of a human inspired autonomous control scheme for a planar bipedal robot in a hybrid dynamical framework to realize human-like walking projected onto sagittal plane. In addition, a unified modelling scheme is presented for the biped dynamics incorporating the effects of various locomotion constraints due to varying feet-ground contact states, unilateral ground contact force, contact friction cone, passive dynamics associated with floating base etc. along with a practical impact velocity map on heel strike event. The autonomous control synthesis is formulated as a two-level hierarchical control algorithm with a hybrid-state based supervisory control in outer level and an integrated set of constrained motion control primitives, called task level control, in inner level. The supervisory level control is designed based on a human inspired heuristic approach whereas the task level control is formulated as a quadratic optimization problem with linear constraints. The explicit analytic solution obtained in terms of joint acceleration and ground contact force is used in turn to generate the joint torque command based on inverse dynamics model of the biped. The proposed controller framework is named as Hybrid-state Driven Autonomous Control (HyDAC). Unlike many other bipedal control schemes, HyDAC does not require a preplanned trajectory or orbit in terms of joint variables for locomotion control. Moreover, it is built upon a set of basic motion control primitives similar to those in human walk which provides a transparent and easily adaptable structure for the controller. These features make HyDAC framework suitable for bipedal walk on terrain with step and slope discontinuities without a priori gait optimization. The stability and agility of the proposed control scheme are demonstrated through dynamic model simulation of a 12-link planar biped having similar size and mass properties of an adult sized human being restricted to sagittal plane. Simulation results show that the planar biped is able to walk for a speed range of 0.1–2 m/s on level terrain and for a ground slope range of $+/-$20 deg for 1 m/s speed.     

Simulation of an autonomous biped walking robot includingenvironmental force interaction

This autonomous biped walking control system is based on reactive force interaction at the foothold. The precise 3D dynamic simulation presented includes: 1) a posture controller which accommodates the physical constraints of the reactive force/torque on the foot with quadratic programming; 2) a real-time COM (center of mass) tracking controller for foot placement, with a discrete inverted pendulum model; and 3) a 3D dynamic simulation scheme with precise contact with the environment. The proposed approach realizes robust biped locomotion because environmental interaction is directly controlled. The proposed method is applied to a 20 axes simulation model, and stable biped locomotion with velocity of 0.25 m/sec and a stepping time of 0.5 sec/step is realized     

Fuzzy SVM learning control system considering time properties of biped walking samples

To learn biped walking dynamics accurately, and then compensate time-varying external disturbances timely, a time-sequence-based fuzzy SVM (TSF-SVM) learning control system considering time properties of biped walking samples is proposed. For the first time, time-sequence-based triangular and Gaussian fuzzy membership functions have been proposed for the single support phase (SSP) and the double support phase (DSP), respectively, according to time properties of different biped phases, which provides an effective way to formulate time properties of biped walking samples in the context of time-varying external disturbances. In addition, a time-sequence-based moving learning window (TS-MLW) is proposed for online training of the proposed TSF-SVM. The performance of the proposed TSF-SVM is compared with other typical intelligent methods; simulation results demonstrate that the proposed method is more sensitive to occasional external disturbances, which increases the stability margin and prevents the robot from falling down.     

A direct adaptive neural command controller design for an unstable helicopter

This paper presents an off-line (finite time interval) and on-line learning direct adaptive neural controller for an unstable helicopter. The neural controller is designed to track pitch rate command signal generated using the reference model. A helicopter having a soft inplane four-bladed hingeless main rotor and a four-bladed tail rotor with conventional mechanical controls is used for the simulation studies. For the simulation study, a linearized helicopter model at different straight and level flight conditions is considered. A neural network with a linear filter architecture trained using backpropagation through time is used to approximate the control law. The controller network parameters are adapted using updated rules Lyapunov synthesis. The off-line trained (for finite time interval) network provides the necessary stability and tracking performance. The on-line learning is used to adapt the network under varying flight conditions. The on-line learning ability is demonstrated through parameter uncertainties. The performance of the proposed direct adaptive neural controller (DANC) is compared with feedback error learning neural controller (FENC).     

基于Fuzzy–CPG的双足机器人适应性行走控制

为提高双足机器人的环境适应性, 本文提出了一种基于模糊控制与中枢模式发生器(CPG)的混合控制策 略, 称之为Fuzzy–CPG算法. 高层控制中枢串联模糊控制系统, 将环境反馈信息映射为行走步态信息和CPG幅值参 数. 低层控制中枢CPG根据高层输出命令产生节律性信号, 作为机器人的关节控制信号. 通过机器人运动, 获取环境 信息并反馈给高层控制中枢, 产生下一步的运动命令. 在坡度和凹凸程度可变的仿真环境中进行混合控制策略的 实验验证, 结果表明, 本文提出的Fuzzy–CPG控制方法可以使机器人根据环境的变化产生适应的行走步态, 提高了 双足机器人的环境适应性行走能力.     

Development of a Complete Dynamic Model of a Planar Five-Link Biped and Sliding Mode Control of Its Locomotion During the Double Support Phase

This paper presents a complete dynamic model of a planar five-link biped walking on level ground. The single support phase (SSP), double support phase (DSP) and double impact occurring at the heel strike are included in the model. By modifying the conventional definition of certain physical parameters of the biped system, it is shown that the procedure of the derivation of the dynamic equations and their final forms are significantly simplified. For motion regulation during the DSP, our dynamic model is formulated as the motion of biped system under holonomic constraints, and the hip position and the trunk orientation are selected as the independent generalized coordinates to describe the constraint system and to eliminate the constraint forces from the equations of motion. Based on the presented dynamic formulation, we develop a sliding mode controller for motion regulation during the DSP where the biped is treated as a redundant manipulator. The stability and the robustness of the controller are investigated, and its effectiveness is demonstrated by computer simulations. To the best of our knowledge, it is the first time that a sliding mode controller is developed for biped walking during the DSP. This work makes it possible to provide robust sliding mode control to a full range of biped walking and to yield dexterity and versatility for performing specific gait patterns.     

学习人类控制策略的双足机器人步态控制研究

针对双足机器人面临的复杂环境下动态行走的适应性难题,提出了一种基于学习人类控制策略的双足机器人步态控制方法。利用三维线性倒立摆模型构造双足行走系统的状态方程,建立学习人类控制策略的参数化模型,设计了基于SVM的学习型控制器。该方法保证了躯干始终处于与地面近似垂直,增强了步态控制的鲁棒性,提高了双足机器人在复杂环境下行走的动态稳定性。实验验证了该方法的有效性。     

双足机器人稳定行走的仿人预测控制研究

针对双足机器人的稳定行走,提出了一种新的仿人预测控制在线步行模式生成方法。把期望零力矩点（ZMP）分解成离线规划好的参考ZMP和实时变化的可变ZMP之和,通过预测控制和其逆系统共同作用对质心运动进行控制,从而生成具有自适应性的步行模式。但单一的预测控制系统对诸如矩形齿状扰动的可变ZMP的跟踪存在较大的误差,结合仿人智能控制对误差的强抑制能力,设计了与预测控制相结合的仿人预测控制系统。仿真实验验证对矩形齿状扰动的可变ZMP,仿人预测系统也能实现较好的跟踪。     

Simultaneous design of optimal gait pattern and controller for a bipedal robot

Nowadays, biped robotics becomes an interesting topic for many control researchers. The biped robot is more adaptable than the other mobile robots in a varied environment and can have more diverse possibilities in planning the motion. However, it falls down easily and its control for stable walking is difficult. Therefore, generation of a desired walking pattern for the biped robot in the presence of some model uncertainties is an important problem. The proposed walking pattern should be also achievable by the designed controller. To achieve this aim and to reach the best control performance, the walking pattern and controller should be designed simultaneously rather than separately. In the present study, an optimal walking pattern is proposed to be tracked by a designed sliding mode controller. In this respect, a genetic algorithm (GA) is utilized to determine the walking pattern parameters and controller coefficients simultaneously. Here, high stability, minimum energy consumption, good mobility properties, and actuator limitations are considered as the important indexes in optimization. Simulation results indicate the efficiency of the proposed scheme in walking the understudy biped robot.     

A reflexive neural network for dynamic biped walking control

Biped walking remains a difficult problem, and robot models can greatly facilitate our understanding of the underlying biomechanical principles as well as their neuronal control. The goal of this study is to specifically demonstrate that stable biped walking can be achieved by combining the physical properties of the walking robot with a small, reflex-based neuronal network governed mainly by local sensor signals. Building on earlier work (Taga, 1995; Cruse, Kindermann, Schumm, Dean, & Schmitz, 1998), this study shows that human-like gaits emerge without specific position or trajectory control and that the walker is able to compensate small disturbances through its own dynamical properties. The reflexive controller used here has the following characteristics, which are different from earlier approaches: (1) Control is mainly local. Hence, it uses only two signals (anterior extreme angle and ground contact), which operate at the interjoint level. All other signals operate only at single joints. (2) Neither position control nor trajectory tracking control is used. Instead, the approximate nature of the local reflexes on each joint allows the robot mechanics itself (e.g., its passive dynamics) to contribute substantially to the overall gait trajectory computation. (3) The motor control scheme used in the local reflexes of our robot is more straightforward and has more biological plausibility than that of other robots, because the outputs of the motor neurons in our reflexive controller are directly driving the motors of the joints rather than working as references for position or velocity control. As a consequence, the neural controller and the robot mechanics are closely coupled as a neuromechanical system, and this study emphasizes that dynamically stable biped walking gaits emerge from the coupling between neural computation and physical computation. This is demonstrated by different walking experiments using a real robot as well as by a Poincaré map analysis applied on a model of the robot in order to assess its stability.     

基于深度强化学习的双足机器人斜坡步态控制方法

为提高准被动双足机器人斜坡步行稳定性, 本文提出了一种基于深度强化学习的准被动双足机器人步态控制方法. 通过分析准被动双足机器人的混合动力学模型与稳定行走过程, 建立了状态空间、动作空间、episode过程与奖励函数. 在利用基于DDPG改进的Ape-X DPG算法持续学习后, 准被动双足机器人能在较大斜坡范围内实现稳定行走. 仿真实验表明, Ape-X DPG无论是学习能力还是收敛速度均优于基于PER的DDPG. 同时, 相较于能量成型控制, 使用Ape-X DPG的准被动双足机器人步态收敛更迅速、步态收敛域更大, 证明Ape-X DPG可有效提高准被动双足机器人的步行稳定性.     

Control of a Biped Walking Robot during the Double Support Phase

This paper discusses the control problem of a biped walking robotduring the double-support phase. Motion of a biped robot during thedouble-support phase can be formulated as motion of robotmanipulators under holonomic constraints. Based on the formulation,the walking gait is generated by controlling the position of thetrunk of the robot to track a desired trajectory, referenced in theworld frame. Constrained forces at both feet were controlled suchthat firm contact is preserved between the feet and ground by using asimplified model of the double-support phase. The control scheme wasevaluated experimentally.     

基于模糊神经网络的5连杆双足机器人混杂控制

针对双足机器人单脚支撑期控制问题, 提出了一种新型的模糊神经网络混杂控制方法. 该种方法结合了模糊神经网络、H_∞ 控制及逆系统方法的优点. 应用了一种新的多层模糊CMAC神经网络对系统进行逼近, 一方面将模糊神经网络的构造误差看作系统的干扰, 利用H_∞ 控制对干扰进行抑制. 另一方面利用模糊神经网络对系统模型进行逼近, 为逆系统的构建和H_∞ 控制率的设计提供了有效的系统信息. 并证明了在采用本文提出的模糊神经网络和自适应算法后可以抑制 L₂ 增益.     

Control of variable speed gaits for a biped robot

We discuss a balance scheme for handling variable-speed gaits that was implemented on an experimental biped at the University of New Hampshire. The control scheme uses preplanned but adaptive motion sequences in combination with closed-loop reactive control. CMAC neural networks are responsible for the adaptive control of side-to-side and front-to-back balance. The biped is able to walk with variable-speed gaits and to change gait speeds on the fly. The slower gait speeds require statically balanced walking, while the faster speeds require dynamically balanced walking. It is not necessary to distinguish between the two balance modes within the controller. Following training, the biped is able to walk on flat, nonslippery surfaces at forward velocities in the range of 21 cm/min to 72 cm/min, with an average stride length of 6.5 cm     

A knowledge-base generating hierarchical fuzzy-neural controller.

We present an innovative fuzzy-neural architecture that is able to automatically generate a knowledge base, in an extractable form, for use in hierarchical knowledge-based controllers. The knowledge base is in the form of a linguistic rule base appropriate for a fuzzy inference system. First, we modify Berenji and Khedkar's (1992) GARIC architecture to enable it to automatically generate a knowledge base; a pseudosupervised learning scheme using reinforcement learning and error backpropagation is employed. Next, we further extend this architecture to a hierarchical controller that is able to generate its own knowledge base. Example applications are provided to underscore its viability.     

基于足迹采样的运动编辑

针对人体两足动画提出一种基于足迹采样的运动编辑算法．足迹很好地描述了两足动画中必须满足的时空约束，通过调整足迹的位置与朝向来编辑两足动画是一种较为自然、直观的交互方式．为有效、快速地生成编辑后的动画，采用一种实时的逆向运动学算法求解两足动画中的支撑脚约束，然后使用层次B样条技术构造偏移映射完成编辑．为了便于逆向运动学算法的求解，提出基于采样的方法来计算质心轨迹．     

Inverse kinematics and inverse dynamics for control of a biped walking machine

Analytical techniques are presented for the motion planning and control of a 12 degree-of-freedom biped walking machine. From the Newton-Euler equations, joint torques are obtained in terms of joint trajectories, and the inverse dynamics are developed for both the single-support and double-support cases. Physical admissibility of the biped trajectory is characterized in terms of the equivalent force-moment and zero-moment point. This methodology has been used to obtain reference inputs and implement the feedforward control of walking robots. A simulation example illustrates the application of the techniques to plan the forward-walking trajectory of the biped robot. The implementation of a prototype mechanism and controller is also described.