Manuel M. Crisóstomo

SVR Versus Neural-Fuzzy Network Controllers for the Sagittal Balance of a Biped Robot

The real-time balance control of an eight-link biped robot using a zero moment point (ZMP) dynamic model is difficult due to the processing time of the corresponding equations. To overcome this limitation, two alternative intelligent computing control techniques were compared: one based on support vector regression (SVR) and another based on a first-order Takagi-Sugeno-Kang (TSK)-type neural-fuzzy (NF) network. Both methods use the ZMP error and its variation as inputs and the output is the correction of the robots torso necessary for its sagittal balance. The SVR and the NF were trained based on simulation data and their performance was verified with a real biped robot. Two performance indexes are proposed to evaluate and compare the online performance of the two control methods. The ZMP is calculated by reading four force sensors placed under each robots foot. The gait implemented in this biped is similar to a human gait that was acquired and adapted to the robots size. Some experiments are presented and the results show that the implemented gait combined either with the SVR controller or with the TSK NF network controller can be used to control this biped robot. The SVR and the NF controllers exhibit similar stability, but the SVR controller runs about 50 times faster.

Human Gait Acquisition and Characterization

This paper analyzes human motion, more specifically the human gait in the sagittal plane. A video camera is used to acquire images of a walking person, fitted with a set of white light-emitting diodes (LEDs). The acquired trajectories of the light points are then used to specify joint trajectories in a biped robot. To analyze the stability of the human gait, a system was also developed to acquire the center of pressure (CoP). This system uses eight force sensors, four under each foot. The influence of the human torso angle on the CoP position during walking was confirmed. Some experiments were carried out on a biped robot, and the results show that the acquired human gait can be used in a biped robot, after scale conversion.

Helicopter motion control using adaptive neuro-fuzzy inference controller

This paper proposes an adaptive neuro-fuzzy inference controller using a feed forward neural network based on nonlinear regression. The general regression neural network is used to construct the base of an adaptive neuro-fuzzy system. This neural network uses a different learning capability when compared with the classical clustering algorithm. The parameters of the general regression neural network are obtained using the gradient descent and least squares algorithms. The simplification of the neuro-fuzzy architecture is done throw the elimination of the rules, maintaining the performance of the controller. In the simulation, the adaptive neuro-fuzzy controller is used to control the helicopter motion in the hover flight mode position. The longitudinal and lateral cyclic, the collective and pedals are used to enable the helicopter to maintain its position fixed in space. Results show the effectiveness of the proposed method.

Control of a Biped Robot With Support Vector Regression in Sagittal Plane

This paper describes the control of an autonomous biped robot that uses the support vector regression (SVR) method for its sagittal balance. This SVR uses the zero moment point (ZMP) position and its variation as input and the torso correction of the robots body as output. As the robot model used segments the robot into eight parts, it is difficult to use online. This is the main reason for using the artificial intelligence method. The SVR was trained with simulation data that was previously tested with the real robot. The SVR was found to be faster (with similar accuracy) than a recurrent network and a neuro-fuzzy control. This method is more precise than the model based on an inverted pendulum. The design of the feet is considered in terms of accommodating the force sensors used to estimate the center of pressure (CoP). The SVR was tested in the real robot using joint trajectories that are similar to those of human beings, and the results are presented.

Robot navigation using a sparse distributed memory

Despite all the progress that has been made in Robotics and Artificial Intelligence, traditional approaches seem unsuitable to build truly intelligent robots, exhibiting human-like behaviours. Many authors agree that the source of intelligence is, to a great extent, the use of a huge memory, where sequences of events that guide our later behaviour are stored. Inspired by that idea, our approach is to navigate a robot using sequences of images stored in a Sparse Distributed Memory-a kind of associative memory based on the properties of high dimensional binary spaces, which, in theory, exhibits some human-like behaviours. The robot showed good ability to correctly follow most of the sequences learnt, with small errors and good immunity to the kidnapped robot problem.

A neural-fuzzy walking control of an autonomous biped robot

In this paper, an adaptive neural-fuzzy walking control of an autonomous biped robot is proposed. This control system uses a feed forward neural network based on nonlinear regression. The general regression neural network is used to construct the base of an adaptive neuro-fuzzy system. The neural network uses an iterative grid partition method for the initial structure identification of the controller parameters. Comparison results are done between the proposed method and the ANFIS tool provided in the fuzzy MATLAB toolbox. The robots control system uses an inverted pendulum to balance of the gaits. The effectiveness of the proposed control system is demonstrated by simulation and experimental tests

Simulation control of a biped robot with Support Vector Regression

This paper describes the control of an autonomous biped robot that uses the Support Vector Regression (SVR) method for its longitudinal balance. This SVR uses the Zero Moment Point (ZMP) position and its variation as input and the longitudinal correction of the robots body is obtained as the output. The SVR was trained based on simulation data that was confirmed with the real robot. This method showed to be faster (with similar accuracy) than a recurrent network or a neuro-fuzzy control of the biped balance.

Automatic helicopter motion control using fuzzy logic

This paper describes the application of the fuzzy logic control (FLC) theory to control the helicopter flight on two basic flight modes: hovering and forward. Hovering is a formidable stability problem, where helicopter pilots typically train for weeks before managing to do it manually. Hence automating this operation is in itself an impressive achievement. In this work, four FLCs execute control rules that represent the linguistic knowledge used by helicopter pilots and found as verbal description in pilot operating manuals. For each flight mode it is used two FLCs where the outputs corresponds to the pitch angle of longitudinal and lateral cyclic. For the hover flight mode the controllers variable input are the pitch attitude and pitch rate of the fuselage for the first controller and the roll angle and roll rate of fuselage for the second controller to control the side and forward velocity through the pitch angle of lateral and longitudinal cyclic. In the forward flight mode, it is used three variable input for each controller where four of them are the same used in the hover flight mode and it is also used the forward velocity variable in both controllers. Simulation results are presented, showing the effectiveness of the proposed FLCs for both flight modes.

AI and memory: Studies towards equipping a robot with a sparse distributed memory

Traditional approaches to Artificial Intelligence (AI) and Robotics seem only to provide advances at a very slow pace. Many researchers agree that new, different approaches are needed to provide a breakthrough and allow the construction of robots with human-like capacities. Our approach consists in navigating a robot using vision a Sparse Distributed Memory (SDM), a kind of associative memory based on the properties of high dimensional binary spaces, which, in theory, exhibits some human-like behaviours. During learning the robot will store sequences of images in the SDM. During execution the robot will follow the sequence of images that is closest to its current view. Preliminary results show that the memory can store and predict sequences of images with a small error tolerance.

Image Processing to a Neuro-Fuzzy Classifier for Detection and Diagnosis of Induction Motor Stator Fault

In this paper a new algorithm for the detection of three-phase induction motor stator fault is presented. This diagnostic technique is based on the identification of a specified current pattern obtained from the transformation of the three- phase stator currents to an equivalent two-phase system. This new algorithm proposes a pattern recognition method to identify induction motor stator faults. The proposed neuro-fuzzy approach is based on the index of compactness, and also indicates the extension of the stator fault. This feature is obtained throw the image processing and used as an input in the neuro-fuzzy classifier. Using the neuro-fuzzy strategy, a better linguistic knowledge and an accurate learning capability underlying the motor faults detection and diagnosis process can be achieved. Simulation and experimental results are presented in order to verify the effectiveness of the proposed method.