Program and papers

The special session took place on 23rd November 2017

14:00-15:00 Session TA1-R5: Autonomous driving and driver assistance systems (I)

Chairs: Arturo de La Escalera and Paulo Oliveira
Location: R5 - Aula 009

14:00 Andre Antunes , Carlos Cardeira and Paulo Oliveira - Application of Sideslip Estimation Architecture to a Formula Student Prototype

14:15 João Antunes , Carlos Cardeira and Paulo Oliveira - Torque Vectoring for a Formula Student Prototype

14:30 Jordi Perez Talamino and Alberto Sanfeliu - Path and velocity trajectory selection in an anticipative kinodynamic motion planner for autonomous driving

14:45 Seyed Amin Sajadi-Alamdari , Holger Voos and Mohamed Darouach - Deadzone-Quadratic Penalty Function for Predictive Extended Cruise Control with Experimental Validation


15:00-16:00 Session TA2-R5: Autonomous driving and driver assistance systems (II)

Chairs: Arturo de La Escalera and Paulo Oliveira
Location: R5 - Aula 009

15:00 Ana Rita Silva Gaspar , Alexandra Pereira Nunes , Andry Pinto and Aníbal Matos - Comparative study of visual odometry and SLAM techniques

15:15 Alireza Asvadi , Luis Garrote , Cristiano Premebida , Paulo Peixoto and Urbano Nunes - Real-Time Deep ConvNet-based Vehicle Detection Using 3D-LIDAR Reflection Intensity Data

15:30 Carlos Guindel Gómez , David Martín Gómez and José María Armingol Moreno - Modeling Traffic Scenes for Intelligent Vehicles using CNN-based Detection and Orientation Estimation.

15:45 Pablo Marín Plaza , Ahmed Hussein , David Martin Gomez and Arturo De La Escalera - Complete ROS-based Architecture for Intelligent Vehicles

Abstracts


Sarouthan Sriranjan, Ray Lattarulo, Joshué Pérez-Rastelli, Javier Ibanez-Guzman and Alberto Peña. Lateral Controllers using Neuro-Fuzzy Systems for Automated Vehicles: A Comparative Study
Abstract: Different implementations on automated vehicles are being introduced by researchers and manufacturers, particularly for longitudinal control. Some applications include traffic jam assistance, emergency assisted braking, Cruise Control, among others. However, lateral control applications are less common due to the complexities of the dynamic. In this paper, an Artificial Intelligence approach to control the steering wheel of an automated vehicle is presented. Two new lateral controllers are developed. One is based on human expertise (Fuzzy Logic), and the other is based on an Adaptive Network based Fuzzy Inference System (ANFIS) using expert driver data. Those controllers have been tested in a simulation environment, called Dynacar, and they were compared with a classical PID controller, giving promising results.
Andre Antunes, Carlos Cardeira and Paulo Oliveira. Application of Sideslip Estimation Architecture to a Formula Student Prototype
Abstract: This paper describes a estimator architecture for a Formula Student Prototype, based on data from an inertial measurement unit (IMU), a global positioning system (GPS), and from the underlying dy-
namic model of the car. Non-linear dynamic model of the car and realistic models for the sensors are presented. The estimates of attitude, rate-gyro bias, position, velocity and sideslip, based on Kalman filtering techniques. The resulting system is validated on a Formula Student prototype an assessed given ground truth data obtained by a set of differential GPS receivers installed onboard.
Alireza Asvadi, Luis Garrote, Cristiano Premebida, Paulo Peixoto and Urbano Nunes. Deep ConvNet-based Vehicle Detection Using 3D-LIDAR Reflection Intensity Data
Abstract: This paper addresses the problem of vehicle detection using a less explored LIDAR's modality: the reflection intensity. The reflectivity attribute is related to the type of surface the LIDAR reflection is obtained. A Dense Reflection Map (DRM) is generated from sparse 3D-LIDAR's reflectance intensity, and inputted to a Deep Convolutional Neural Network (ConvNet) object detection framework (YOLOv2 [1]) for the vehicle detection. The proposed approach is the first result using LIDAR's reflection value in the KITTI Benchmark Suite. Although only reflection intensity data is used in the approach presented in this paper, the performance is superior to some of the approaches that use LIDAR's range-value, and hence it demonstrates the usability of LIDAR's reflection for vehicle detection.
Jordi Perez Talamino and Alberto Sanfeliu. Path and velocity trajectory selection in an anticipative kinodynamic motion planner for autonomous driving
Abstract: This paper presents a novel approach for plan generation, selection and pruning of trajectories for autonomous driving, capable of dealing with dynamic complex environments, such as driving in urban scenarios. The planner first discretizes the plan space and searches for the best trajectory and velocity profile of the vehicle. The main contributions of this work are the use of G2-splines for path generation and a method that takes into account accelerations and passenger comfort for generating and pruning velocity profiles based on 3rd order splines, both fulfilling kinodynamic constraints. The proposed methods have been implemented in a motion planner in MATLAB and tested through simulation in different representative scenarios, involving obstacles and other moving vehicles. The simulations show that the planner performs cor- rectly in different dynamic scenarios, maintaining the passenger comfort.
Seyed Amin Sajadi-Alamdari, Holger Voos and Mohamed Darouach. Deadzone-Quadratic Penalty Function for Predictive Extended Cruise Control with Experimental Validation
Abstract: Battery Electric Vehicles have high potentials for the modern transportations, however, they are facing limited cruising range. To address this limitation, we present a semi-autonomous ecological driver assistance system to regulate the velocity with energy-efficient techniques. The main contribution of this paper is the design of a real-time nonlinear receding horizon optimal controller to plan the online cost-effective cruising velocity. Instead of conventional l2-norms, a deadzone-quadratic penalty function for the nonlinear model predictive controller is proposed. Obtained field experimental results demonstrate the effectiveness of the proposed method for a semi-autonomous electric vehicle in terms of real-time energy-efficient velocity regulation and constraints satisfaction.
João Antunes, Carlos Cardeira and Paulo Oliveira. Torque Vectoring for a Formula Student Prototype
Abstract: Torque Vectoring (TV) has the objective to substitute the need of a mechanical dierential, while improving the handling and response of the wheeled vehicle. This work addresses the design of a torque vectoring system in an rear wheel driven formula student prototype. The proposed solution resorts to a PID controller for yaw rate tracking with an evenly distributed torque to each wheel. Also an LQR scheme is discussed, for tracking the yaw rate and the lateral velocity. To assess and design, rst a 7 degree of freedom (DOF) non linear model is constructed, followed by a linear 2 DOF model, both validated with real data. The linear model, is used to design and simulate the proposed controllers.
When the controller is within the desired parameters it is tested in the non linear model. Tests with the vehicle are performed to verify the contribution of the controller to the overall performance of the vehicle.
Ana Rita Silva Gaspar, Alexandra Pereira Nunes, Andry Pinto and Aníbal Matos. Comparative study of visual odometry and SLAM techniques
Abstract: The use of the odometry and SLAM visual methods in autonomous vehicles has been growing. Optical sensors provide valuable information from the scenario that enhance the navigation of autonomous vehicles. Although several visual techniques are already available in the literature, their performance could be significantly affected by the scene captured by the optical sensor. In this context, this paper presents a comparative analysis of three monocular visual odometry methods and three stereo SLAM techniques. The advantages, particularities and performance of each technique are discussed, to provide information that is relevant for the development of new research and novel robotic applications.
Pablo Marín Plaza, Ahmed Hussein, David Martin Gomez and Arturo De La Escalera. Complete ROS-based Architecture for Intelligent Vehicles
Abstract: In the Intelligent Transportation Systems Society (ITSS), the research interest on intelligent vehicles is increasing during the last few years. Accordingly, this paper presents the advances in the development of the ROS-based (Robot Operating System) software architecture for intelligent vehicles. The main contribution of the architecture is its powerfulness, flexibility, and modularity, which allows the researchers to develop and test different algorithms. The architecture has been tested on different platforms, autonomous ground vehicles from the iCab (Intelligent Campus Automobile) project and in the intelligent vehicle based on Advanced Driver Assistance Systems (ADAS) incorporated from IvvI 2.0 (Intelligent Vehicle based on Visual Information) project.
Carlos Guindel Gómez, David Martín Gómez and José María Armingol Moreno. Modeling Traffic Scenes for Intelligent Vehicles using CNN-based Detection and Orientation Estimation
Abstract: Object identification in images taken from moving vehicles is still a complex task within the computer vision field due to the dynamism of the scenes and the poorly defined structures of the environment. This research proposes an efficient approach to perform recognition on images from a stereo camera, with the goal of gaining insight of traffic scenes in urban and road environments. We rely on a deep learning framework able to simultaneously identify a broad range of entities, such as vehicles, pedestrians or cyclists, with a frame rate compatible with the strong requirements of onboard automotive applications. The results demonstrate the capabilities of the full perception system for a wide variety of situations, thus providing valuable information to understand the scenario ahead of the vehicle.