Leopoldo Armesto

Fast Ego-motion Estimation with Multi-rate Fusion of Inertial and Vision

This paper presents a tracking system for ego-motion estimation which fuses vision and inertial measurements using EKF and UKF (Extended and Unscented Kalman Filters), where a comparison of their performance has been done. It also considers the multi-rate nature of the sensors: inertial sensing is sampled at a fast sampling frequency while the sampling frequency of vision is lower. the proposed approach uses a constant linear acceleration model and constant angular velocity model based on quaternions, which yields a non-linear model for states and a linear model in measurement equations. Results show that a significant improvement is obtained on the estimation when fusing both measurements with respect to just vision or just inertial measurements. It is also shown that the proposed system can estimate fast-motions even when vision system fails. Moreover, a study of the influence of the noise covariance is also performed, which aims to select their appropriate values at the tuning process. The setup is an end-effector mounted camera, which allow us to pre-define basic rotational and translational motions for validating results.

Multi-rate fusion with vision and inertial sensors

This work presents a multi-rate fusion model, which exploits the complimentary properties of visual and inertial sensors for egomotion estimation in applications such as robot navigation and augmented reality. The sampling of these two sensors is described with size-varying input and output equations without assumed synchronicity and periodicity of measurements. Data fusion is performed with two different multi-rate (MR) filter models, an extended (EKF) and an unscented Kalman filter (UKF). A complete dynamic model for the 6D-tracking task is given together with a method to calculate the dependencies of the covariance matrices. It is further shown that a centripetal acceleration model and the precise description of quaternion prediction for a constant velocity model highly improve the estimation error for rotary motions. The comparison demonstrates that the MR-UKF provides better estimation results at higher computational costs.

SLAM based on Kalman filter for multi-rate fusion of laser and encoder measurements

In this paper, extended Kalman and unscented Kalman filters are developed for multi-rate systems in the context of the SLAM problem. These multi-rate filters have been extensively compared and tested with experimental data taken from a parking lot. Both multi-rate filters have improved the estimation with respect to the conventional single-rate Kalman filter. A performance index is introduced, showing that EKF gives better performance than UKF. In order to complete the SLAM solution, well-known techniques for feature extraction, data association and map building have been also implemented.

Dual-rate high order holds based on primitive functions

Absfracf- This paper establishes a new and general methodology for computing dual-rate high order holds based on primitive functions. This methodology includes the formulation of all the well known single-rate (N=l) and dual-rate hold circuits. Different polynomial primitive functions can be considered. They allow the formulation of the Impulse-Hold, ZOH, FOH and SOH for single and dual-rate sampling. The general methodology also allows considering dual-rate holds based on non-polynomial functions reducing complexity formulation. This is especially interesting when trying to match the behavior of dynamic systems. In this sense, the exponential and sinusoidal functions allow to approach the response of first and second order systems at high frequency.

A generalization of the metric-based Iterative Closest Point technique for 3D scan matching

Scan matching techniques have been widely used to compute the displacement of robots. This estimate is part of many algorithms addressing navigation and mapping. This paper addresses the scan matching problem in three dimensional workspaces. We propose an generalization of the Metric based Iterative Closest Point (MbICP) to the 3D case. The main contribution is the development of all the mathematical tools required to formulate the ICP with this new metric, including the derivation of point to plane distances based on the new metric. We also provide experimental results to evaluate the algorithms and different combinations of ICP and MbICP to illustrate the advantages of the metric based approach.

On multi-rate fusion for non-linear sampled-data systems: Application to a 6D tracking system

Egomotion estimation, e.g. for robot navigation or augmented reality applications, requires the fusion of non-linear sampled-data system with different sensors. An example is to fuse the complimentary characteristics of visual and inertial sensors. Existing approaches either use Kalman filters in conventionally sampled systems or use Particle filters to accommodate the uncertainty of motion models. This paper introduces an approach that models multi-rate non-linear systems to exploit the characteristics of both sensors, assuming synchronicity and periodicity of measurements. The final contribution of this paper is an in-depth analysis and performance comparison of the Extended Kalman filter, the Unscented Kalman filter and three particle filters (Bootstrap, Extended and Unscented). While there is large debate over the pros and cons of these two approaches, this work shows the following results for fusing visual and inertial data in 6 DOF (position and orientation) in a tracking application: the Bootstrap Particle filter gives higher estimation error than Extended and Unscented Particle filters, which give very similar results than Extended and Unscented Kalman filters, but with considerable higher computational burden.

Automation of industrial vehicles: A vision-based line tracking application

This paper presents some intermediate results of a Research Project regarding with Auto-Guided Vehicles (AGVs) and vision-based guidance. The paper describes three different degrees of autonomy, form the basic one to the full autonomous one. In particular, the paper focuses on the intermediate degree of autonomy, named as “Guided” driving, which includes several types of application under the same approach such as manual-assisted driving, teleoperated driving and a vision-based line tracking application. The paper describes on the line-tracking problem with AGVs where a simple and robust line detection algorithm has been described and implemented on an embedded vision system. It also describes a line tracking control algorithm, which has been validated through experimental and simulation data. Finally, the paper discusses future work in the context of the project.

Factory management and transport automation

This paper presents a global solution developed for integrating the warehouses transportation system based on a fleet of automated industrial forklifts in the factory management. Each industrial forklift has been completely automated using external and internal sensors. An ERP system, as information system, has been installed in order to handle the whole factory management process. A SCADA application has also been implemented for monitoring the industrial vehicles in the plant. It also acts as an interface between the ERP and the process itself. Finally, three vehicle transportation modes have been developed and validated in the system: teleoperation, path-tracking and line tracking.

Robust and Efficient Mobile Robot Self-localization using Laser Scanner and Geometrical Maps

In this paper, we present a set of robust and efficient algorithms with O(N) cost for object detection and pose estimation with a laser ranger based on regular geometrical maps. Firstly, a multiple line fitting method is described, related with walls at the environment, which minimizes the sum of squared distances for contiguous lines and constitutes a global pattern with regular constrained angles. Secondly, beacons, related with columns at the environment, are estimated with the circle algebraic method. Two pose estimation methods are presented, based on detected beacons and lines. These methods have proved to be robust and faster than other methods. In addition to this, a redundant kinematic model is used for dead-reckoning sensors, which may improve estimations. Finally, the robot localization is performed with an EKF during normal navigation and a global localization method, based on Monte Carlo simulation methods, at initialization

Inspection system based on artificial vision for paint defects detection on cars bodies

This paper describes the design and implementation of a novel inspection system for detecting defects on car bodies based on artificial vision, implemented in Ford Factory at Almussafes (Spain). The system is based on the principle of performing a lightning sweeping with static imagining system, which causes shadows surrounding defects when merging consecutive images, coined as defect augmentation phenomena. As a result, we can detect millimetric defects of 0.3mm diameter or greater with different shapes which were very hard to detect with existing technology without that phenomena. The project has generated two PTC patents, the first one protects the defect augmentation phenomena, while the second protects the industrial system itself. The main innovation of this industrial project is the development of a system that improves in almost 100% the human inspection. As a consequences, it reduces the number of invalid vehicles, energy consumption, saving painting which also implies a significant cost reduction. It also improves working conditions for workers by reducing ocular fatigues.