previous day
all days

View: session overviewtalk overview

08:30-10:30 Session 10A: Autonomous Systems
Development of a Network Enabled System for Generic Autonomous Vehicles

ABSTRACT. This paper describes the development of a system for autonomous vehicle testing, utilising conventional network infrastructure for communication and control; allowing simultaneous control of multiple vehicles of differing vehicle types. A basic level of autonomy is achieved through the use of an Arduino based commercial autopilot (ArduPilot), which also allows for remote vehicle control via MAVLink protocol commands given through serial communication. Traditionally messages are sent using point-to-point wireless serial modems. As these are restricted in terms of bandwidth and flexibility, an improved set-up is suggested, where an embedded computer system is attached to each vehicle. A custom written Node.js program (MAVNode) is then used to encode and decode MAVLink messages onboard allowing communication over a Local Area Network via Wi-Fi. A selection of hardware configurations are discussed, including the use of conventional Wi-Fi and long range Ubiquiti airMAX wireless routers. Both software and hardware in the loop testing is discussed, in addition to the ability to perform control from Matlab/Simulink. With all the infrastructure in place, algorithms can be rapidly prototyped. As an example use of the system, a quad-rotor visually tracks a robot while using a remote Matlab installation for image processing and control.

Cooperative Source Seeking via Gradient Estimation and Formation Control (Part 1)

ABSTRACT. In this paper and its companion paper [16], the problem of cooperative source seeking by a formation of mobile agents is considered. Each agent is equipped with position and signal strength measurement sensors; their task is to find the maximum of the scalar field. Agents exchange information with neighboring agents through a communication network. In the first part of this couple of papers, a distributed gradient estimation for each agent and a decentralized navigation controller for single- and double-integrator models are presented. When the signal measurements are corrupted by noise, distributed consensus filters are used in order to estimate the gradient direction. The strategy is based on both a gradient estimating algorithm and a formation controller. Stability conditions are provided. Numerical simulations illustrate the effectiveness of the proposed control law. Part two extends this approach to general linear time-invariant models.

Cooperative Source Seeking via Gradient Estimation and Formation Control (Part 2)

ABSTRACT. In this paper and its companion paper [16], the problem of cooperative source seeking by a formation of mobile agents is considered. Each agent is equipped with position and signal strength measurement sensors; their task is to find the maximum of the scalar field. Agents exchange information with neighboring agents through a communication network. In the first part, a distributed gradient estimation for each agent and a decentralized navigation controller for single- and double-integrator models are presented. In this paper, the approach is extended to general linear time-invariant (LTI) models. Stability conditions are provided and our approach is verified using formation flight simulation for quad-rotor helicopters.

Cooperative Conflict Resolution by Velocity Obstacle Method
SPEAKER: Fatemeh Asadi

ABSTRACT. The automated and more efficient methods for resolution of conflicts between aircraft is necessary to support the sustained growth of air traffic. Distributed optimization is one of the proposed conflict resolution methods which can improve efficiency; but, sometimes it imposes the unequal burden on involved agents. This paper presents a method for conflict resolution by cooperation between agents which can lead to fair contribution of all agents in resolving the collision.

Estimation Of Time To Point Of Closest Approach For Collision Avoidance And Separation Systems

ABSTRACT. This paper proposes a method for estimating the time until two aircraft are at the their point of closest approach (TPCA). A range of simple methods, which use derivatives to estimate the time to collision, are analysed. These methods are only accurate when the angle subtended between the direction of the relative velocity vector and the bearing of the intruder aircraft, $\theta$, is small. An extended method is developed which calculates the exact TPCA from distance and bearing measurements. Representative levels of Gaussian white noise are introduced to the core equation variables for both the derivative and extended methods. It is found that as we increase the value of $\theta$, the extended method's accuracy increases beyond that of the derivative method. A fusion algorithm is developed to switch between methods and is shown to perform well for a range of conflicts. When the relative velocity between the two aircraft is small, the signal to noise ratio on the relative velocity variable reduces causing large errors to the TPCA estimation. It is therefore concluded that at a certain relative velocity threshold, $k_V$ (dependant on sensor and filter performance) both the derivative and extended TPCA estimation methods would become undesirable as risk estimators. It is suggested that in these situations distance could be better to use since it can be measured directly.

Ant Colony Optimization for Routing and Tasking Problems for Teams of UAVs

ABSTRACT. This paper presents an enhanced version of the ant colony optimization (ACO) for solving an improved model of vehicles routing problem (VRP), which is utilized for Unmanned Aerial Vehicle (UAV) task loitering and route planning. The improved VRP incorporates collision avoidance penalties, not only for the intersections between the vehicle's routes, but also for their departure and landing times. The ant colony algorithm uses a single objective function, consisting of penalties, ensuring in that way that the solutions with intersections will be evaluated. The ACO is a variation of the already known multi-colony algorithm where several ant colonies are assigned to different loitering steps for the same route between two tasks. Numerical experiments and comparison to previous work are illustrated to demonstrate the efficiency of the proposed algorithm.

08:30-10:30 Session 10B: Model Predictive Control
Efficient State Constraint Handling for MPC of the Heat Equation
SPEAKER: Tilman Utz

ABSTRACT. This contribution is concerned with model predictive control (MPC) of systems governed by partial differential equations (PDEs) and subject to state and input constraints. In particular, the numerically efficient handling of the underlying optimal control problem (OCP) is considered. It is shown that the state- and input-constrained OCP can be transformed by using saturation functions into an unconstrained OCP. This then can be solved numerically by means of well-established optimization methods. For a numerically efficient implementation of the MPC, a first discretize then optimize-approach is used in conjunction with a tailored gradient method. Both the transformation into an unconstrained OCP and its numerical solution are investigated for a simple heat conduction problem.

A Synthesis Strategy for Nonlinear Model Predictive Controller on FPGA
SPEAKER: unknown

ABSTRACT. This paper describes an implementation strategy of nonlinear model predictive controller for FPGA systems. A high-level synthesis of a real-time MPC algorithm by means of the MATLAB HDL Coder as well as the Vivado HLS tool is discussed. In order to exploit the parallel processing of FPGAs, the included integration schemes are parallelized using a fixed- point iteration approach. The synthesis results are demonstrated for two different example systems.

Networked predictive control for linear systems with unknown communication delay

ABSTRACT. This paper is concerned with the problem of controller design for linear systems with unknown communication delay. A control scheme is proposed to deal with the unknown communication delay on the basis of networked predictive control method. The closed-loop system is modeled as a switched system under constraint switching taking an important property of communication delay into account. A sufficient stability condition is derived using switched Lyapunov function approach. How to choose an important parameter in the control law is also considered. Finally, two numerical examples are given to illustrate the effectiveness of the proposed method.

Review of MPC Applications in Wind Turbines
SPEAKER: Wai Hou Lio

ABSTRACT. This paper aims to give an overview of the recent development and benefits of model predictive control in wind turbines and its future potential. For a modern large wind turbine, the main objective of control is to maximise the power production while maintaining the fatigue loads to be minimal. With such multiple objectives, a multivariable system and actuators constraints the popular PI controller may become ineffective or hard to tune whereas MPC provides a systematic approach for designing a multivariable controller incorporating the knowledge of actuator constraints. This paper reviews the wind turbine control problem and in particular gives a survey of the use of model predictive control on wind turbines.


Gradient Filter Methods for Predictive Control with Simple Constraints
SPEAKER: Eric Kerrigan

ABSTRACT. One of the main limitations with model predictive control (MPC) is the need to solve an online optimisation problem at each sampling instant. In this paper, a framework for developing computationally efficient, first-order optimisation solvers using gradient filtering techniques is presented. These solvers are similar in form to Nesterov's fast gradient method, however they are differentiated by the increased number of historical iterations used in their gradient filter. This approach was found to accelerate the convergence rate of the fast gradient method and could compute the solution of a test case quadratic programming problem, based on MPC of an atomic force microscope, in a shorter time than both the fast gradient and interior point methods.

The Effect of Model and Objective Function Mismatch in Model Predictive Control (MPC) for a Solar Heating System with a Heat Pump
SPEAKER: unknown

ABSTRACT. This paper investigates the effect of model mismatch on the performance of model predictive control (MPC) when applied to the heating system. The controller uses a linear model and a quadratic cost function, while the actual process is non-linear in nature with a linear cost function. A genetic algorithm (NSGA II) is used to find the optimal solution to the actual problem and a number of variations, which are then compared the performance of the MPC controller. The results show that the model mismatch has a small but significant effect on the control performance, and it does prevent effective load shifting in certain situations.

08:30-10:30 Session 10C: Invited Session - Advanced Robotics and Brain Computer Interface
Design and Implementation of Serious Games for Training and Education

ABSTRACT. Serious games are considered to be effective in the field of training and education to supplement the conventional methods, since their addictivity is effective to keep the motivation of trainees or learners. Although many of serious games have been developed and used, many of them lacks the quantitative analysis of their effectiveness. Moreover, in order that serious games become more popular, an effective software development process which considers the evaluation as an important phase in the process. In this paper, we propose a serious game design process “SGDP” and show the serious games that that applied SGDP. We also discuss the method to evaluate serious games by monitoring brain activities of the players.

Calculation Algorithm for Motion Control Target for the Platooning of Autonomous Heavy Vehicles
SPEAKER: unknown

ABSTRACT. In this study, we develop an algorithm for the real-time generation of a trajectory for a control target to realize the autonomous steering control needed to realize a platooning system for heavy trucks. The proposed algorithm consists of a driver model, which considers the risk potential, and a vehicle dynamics model. This algorithm was validated through simulations using a vehicle dynamics model with multiple degrees of freedom. In addition, the feasibility of real-time processing for the control target was determined through field tests in which the proposed algorithm was implemented on the control hardware of an experimental vehicle.

Study on Control System of Rider Robot for Motorcycle
Development of Portable Brain-Computer Interface Using NIRS

ABSTRACT. This paper is directed toward the development of a portable NIRS-BCI system, through investigation of different methods of brain activity classification in BCI. We first compare the performance of three classifiers (perceptron, BP network, and SVM) in classification of an ideal signal. We then investigate their performance in classification of actual data acquired in an experiment with volunteer participants and apply the findings to construction of a portable BCI system showing a capability for real-time implement manipulation.

Classification of functional near-infrared spectroscopy signals applying reduction of scalp hemodynamic artifact
SPEAKER: Takanori Sato

ABSTRACT. Functional near-infrared spectroscopy (fNIRS) has been considered the application to brain-computer interfaces (BCIs) in many studies because of its simplicity of use and portability. However, scalp-hemodynamics often contaminates fNIRS signals as an artifact, and causes the markedly-degradation of the signal-to-noise ratio for functional signals. Although some studies have reported reduction method for the artifacts, no study has investigated its effects on BCIs. In our previous study, we also proposed the artifact reduction method that estimates the global scalp-hemodynamic component from minimal number of short source-detector distance channels (Short-channels), and removes its influence from standard source-detector distance channels using a general linear model (GLM) that incorporates the scalp-hemodynamics in the design matrix. In this study, we investigated the effects of applying scalp-hemodynamic reduction to classifications for four tasks: a ball grasping with right, left, or both hands, or resting. We used a support vector machine (SVM) and binary-tree multi-classification method, and compared five types of the ΔOxy-Hb feature: time samples of raw data, of data subtracted scalp-hemodynamics, and of estimated scalp-hemodynamics, and β values for the cerebral-hemodynamic component into GLM obtained by analyzing by a standard GLM without scalp hemodynamic model and the proposed GLM. As results, the proposed method was successfully improved the signal-to-noise ratio of ΔOxy-Hb signals, and the feature types of β values estimated by the proposed method showed the highest accuracy for classification. These results suggest that the reduction of the scalp-hemodynamic artifacts may provide the more accurate fNIRS-BCI.

08:30-10:30 Session 10D: Invited Session - Real-time Adaptive Networked Control of Rescue Robots 2
Hopping robot motion dynamic control using semi-active control systems (TBC)
The development of an smart chair to assist sit-to-stand transferring process

ABSTRACT. Standing up from a seated position, known as sit-to-stand (STS) movement, is one of the activities of daily living (ADLs) on a daily basis. As people age, physiological changes occur including reduced muscle strength and mass as well as sensory capacity. This may lead to difficulties in STS transferring process, with which the elderly may encounter sedentary lifestyle and contracted social space. There exist market available assistive lift devices with performance far from satisfaction, for the reason being that they fail to provide appropriate assistance. Thus, an intent-based automatic lift chair is proposed and partially developed aiming to analyse user’s physiological condition through pattern recognition. The idea of assistance-as-needed is also introduced which may help encourage the elderly to improve their own motor function by offering personalised assistance with adaptation to the change of conditions.

Design of a new solution for the wheeled hopping robot
SPEAKER: unknown

ABSTRACT. In recent years many researchers are trying to combine wheeled movement with legged movement or hopping movement to strengthen the robots adaptability to environment. Hybrid robots are playing a more and more important role in our modern life. However, their controls are mostly complicated. So in this paper we present a new solution for the wheeled hopping robot. Based on the mechanical design, this robot has two movement modes: wheeled and hopping movement mode. While the center of gravity of the mechanism is under the wheels center, the wheeled hopping robot which is like a tumbler can keep its stability even when the wheels’ velocity is zero. In this paper, we firstly introduce the description and design of the wheeled hopping robot. Then we explain how to choose the robot parameters and lastly we describe, how the robot through hopping, successfully traverses obstacles and ditches.

Enhanced Extenics Controller for Real Time Control of Rescue Robot Actuators
SPEAKER: unknown

ABSTRACT. The paper presents a new control method for the real time control of rescue robots’ actuators, using the concepts and ideas of Extenics Theory. The controller uses the Dependent Function to measure the degree of compatibility of the process variable and combines this information with the derivative of the system error. It then takes the appropriate action to force the system into convergence around a desired set point. This builds upon the previous single-input Extenics controller and leads to a slightly more complex implementation with markedly better results, while retaining the ease of defining and adjusting the associated parameters and rule base. The output of the Dependent Function classifies the process variable value into one of four categories, concurrent with the nested intervals used in Extenics Theory. The rationality and validity of the proposed model are demonstrated through simulation in the Matlab/Simulink environment. The controller is then compared to a selection of other possible controllers. Throughout the paper, opportunities for further improvement and research are highlighted and discussed.

Human Skill Performance to Control an Underactuated Pendulum-Driven Capsule System

ABSTRACT. This paper investigates human learning and skill performance to control an underactuated pendulum-driven capsule system within an interactive virtual simulation environment. The experiment conducted with 9 participants who learned to control the robot via the provided joystick interface within the virtual simulation platform. The results show the difference in learning and skill performance among the participants. Right-handed and left-handed participants achieved their highest trial on the opposite side of their handedness. High learning participant tends to achieve high performance whereas participant who has steady learning tends to produce stable performance either low or high. The variance of the displacements achieved appears to be a learning indicator while the high frequency of joystick oscillation at the right portion and interval gives high performance results.

Mechanism Topology Design for Novel Parallel-Parallel Hexapod Robot

ABSTRACT. The Octopus is a hexapod robot designed for research on emergency rescue missions in nuclear plants. The robot focused on load-bearing capability and adaptability in different unknown radiation scenarios. Compared with other legged robots, the main difference is that the robot applied 3 degree of freedom (DOF) parallel mechanism which improves the payload capability significantly. In this paper, first the background and the robot system is introduced. Then mechanism topology design methodology is mainly discussed. In our design process, the GF set theory is used which design the limb type from the end-effector requirements. Finally the relative experiments are briefly shown.

11:00-12:00 Session 11: Plenary - Chris Edwards
Sliding Mode Approaches for Fault Diagnosis and Fault Tolerant Control
Disclaimer | Powered by EasyChair Smart Program