Fernando Moita

Sensor fusion for precise autonomous vehicle navigation in outdoor semi-structured environments

This paper presents a guidance system for autonomous vehicles navigation in semi-structured outdoor environments. It integrates redundant encoders data and absolute positioning data provided by landmarks and artificial beacons. Natural features are localized using a laser range sensor, and magnetic sensing rulers were developed to detect magnetic markers buried in the ground. In the first fusion stage, data from four wheel encoders and one steering encoder are fused by means of an EKF, providing robust odometric information, namely in face of undesirable effects of wheels slippage. Next, a second fusion stage is processed for integrating odometric and absolute positioning data. Simulation and real experiments using a four-wheels actuated electrical vehicle are presented.

Semantic fusion of laser and vision in pedestrian detection

Fusion of laser and vision in object detection has been accomplished by two main approaches: (1) independent integration of sensor-driven features or sensor-driven classifiers, or (2) a region of interest (ROI) is found by laser segmentation and an image classifier is used to name the projected ROI. Here, we propose a novel fusion approach based on semantic information, and embodied on many levels. Sensor fusion is based on spatial relationship of parts-based classifiers, being performed via a Markov logic network. The proposed system deals with partial segments, it is able to recover depth information even if the laser fails, and the integration is modeled through contextual information-characteristics not found on previous approaches. Experiments in pedestrian detection demonstrate the effectiveness of our method over data sets gathered in urban scenarios.

3D laser-based sensor for robotics

The paper presents the analysis and design of a distance and orientation sensor based on the triangulation principle, using three laser diodes and a position sensitive detector (PSD). A 16-bit microcontroller controls the acquisition and the lasers, processes the data and integrates the sensor in a distributed sensor network (DSN). A DSP based processing system gathers the multisensor data, for real-time control of a PUMA 560 robot manipulator through the DSN network. A sensor model based on theoretical and experimental results is emphasized.<<ETX>>

Multi-echo technique for feature detection and identification using simple sonar configurations

This work presents the study and development of sonar sensor configurations and processing methods with capability to detect and identify planes, edges and corners. The systems presented can be easily applied to settled robots or new platforms to improve the quality and quantity of ultrasonic data. No special configurations as arrays or cells of sensors are used in this work, only standard configurations that can be found and applied in a great number of mobile robots. New multi-echo functionalities are introduced and combined with EERUF, providing a new ME-EERUF (Multi-Echo Error Eliminating Rapid Ultrasonic Firing) method that simultaneously increases the sonar acquisition rate and provides inter-sensor measurements, necessary to implement feature detection and identification algorithms. Sonar feature-based and grid-based maps are built and integrated online using Bayesian probabilistic approaches. Experimental results using data collected in a specular indoor environment are presented in the paper.

Autonomous electric vehicle: Steering and path-following control systems

This paper describes the ISRobotCar, an experimental autonomous electric vehicle that integrates Robot Operating System (ROS) and several sensors such as IBEO laserscanner, Inertial Measurement Unit (IMU), RTK-GPS, vision cameras, and magnet detectors for magnetic guidance. The development of a power steering controller and of a fuzzy path-following controller are particularly addressed. Finally simulations and experimental results of autonomous path-following control are presented.

An outdoor guidepath navigation system for AMRs based on robust detection of magnetic markers

This paper presents an outdoor guidepath navigation system for autonomous mobile robots (AMR) that use permanent magnetic markers embedded in the ground. The odometric data provided by the wheel encoders is fused with the data from magnetic markers. The extended Kalman filter (EKF) was chosen for the fusion process. The AMR is equipped with a magnetic sensing ruler (MSR) developed at ISR-UC that is able to perform a robust detection of magnetic markers. The detection is based on a 3-D algorithm that includes longitudinal-fitting detection (LFD), and cross-fitting detection (CFD). Both, the LFD and the CFD are based on the least squares fitting (LSF) of the measurement data with the 3-D model of the vertical magnetic field. The experimental results with Robchair (intelligent wheelchair being developed at ISR-UC) primarily show that the detection system is robust, since it is able to detect true magnetic markers, and to eliminate noisy magnetic distortions and false markers. The design, and implementation of the navigation algorithm in the Robchair were carried out, and results are presented.

ISRobotCar: The autonomous electric vehicle project

The video shows an overview of the ISRobotCar, an experimental autonomous electric vehicle that integrates the robot operating system (ROS.org) and several sensors such as an IBEO laser scanner, inertial measurement units (IMU), RTK-GPS, vision cameras, and magnet detectors for magnetic guidance. The main purpose of this vehicle is to serve as a platform for experimental testing of algorithms required for autonomous navigation and cooperative navigation in urban environments. In the video, we present the ISRobotCar moving autonomously in a parking-like area as well as its main hardware and control modules.

A Fast Firing Binaural System for Ultrasonic Pattern Recognition

A binaural sonar configuration with capability to detect and identify walls, edges and corners on real-time is presented in this work. A new multi-echo ultra-fast firing method increases the sonar acquisition rate, and provides crossed measurements without interference. A feature map is built on-line using Bayesian updating and classification rules. Three classifiers are implemented and analyzed: minimum risk (MR), maximum a posteriori (MAP), and minimum distance (MD). Experimental results of ultrasonic reflector recognition, using data collected in a specular indoor environment are presented in the paper.

Ultrasonic reflectors recognition with a fast firing system

Abstract A binaural sonar configuration and processing algorithms with capability to detect and identify planes, edges and corners are presented in this work. Furthermore, a new multi-echo ultra-fast firing method, witch simultaneously increases the sonar acquisition rate by four to six times than conventional ones and provides crossed measures without interference, are presented and modelled. Feature maps are built on-line using Bayesian updating and classification rules. Experimental results of ultrasonic reflectors recognition, using data collected in a specular indoor environment are presented in the paper.

Mapping with a binaural system

A binaural sonar configuration with the capability to detect and identify planes, edges and corners is presented in this work. A new ultra-fast firing method with acquisition rates four to six times greater than the conventional ones, shows high interference rejection capabilities, preserving a very low level of errors. Experimental results have shown successful rejection of undesired crosstalk and good acceptance of true measurements. With just two firing cycles, this system provides fast and more confident information, leading to a higher obstacle detection probability. Together with a new feature detection and identification algorithms, on-line feature maps are built making use of the Bayesian updating and classification rules. Experimental results of ultrasonic reflectors recognition, using data collected in a specular indoor environment are presented in the paper.