ROBOT2017: ROBOT'2017 - THIRD IBERIAN ROBOTICS CONFERENCE
PROGRAM FOR WEDNESDAY, NOVEMBER 22ND
Days:
next day
all days

View: session overviewtalk overview

15:00-16:00 Session Plenary I: Ocean One: A Robotic Avatar for Oceanic Discovery.

Author: Oussama Khatib (Stanford University)

Location: Salón de actos
16:00-17:30 Session WA1-R1: General track: Mobile robotics
Location: R1 - Sala Juan Larrañeta
16:00
Landmark detection for docking tasks

ABSTRACT. With the objective to use robots to perform tasks that require docking manoeuvres, these have to be carried out with high precision and accuracy. It is possible to divide these manoeuvres into two parts. One is responsible for the landmark detection and the second one is responsible for guiding the robot towards a position to dock. Keeping this in mind, the objective of this article is to test possible ways to detect a landmark using a laser rangefinder. We will test a beacon-based localisation algorithm and an algorithm based on natural landmarks already implemented, however we will apply modifications to such methods. To verify the possibility of docking using these methods, we will conduct experiments with a real robot.

16:15
Rao-Blackwellized particle filter SLAM with prior map: An experimental evaluation

ABSTRACT. Robotic applications demanding high-precision positioning, such as material handling by autonomous forklifts, require accurate and consistent maps of the environment. Moreover, exact correspondence between features of the architectural floor plan and the vehicle navigation map is desirable, as pick-up and drop-off locations are typically laid out on the floor plan. In this paper, we present a modification of the well-established grid-based Rao-Blackwellized particle filter (RBPF), which enables efficient initialization of its internal data structures using information from the architectural floor plan. This prior information simultaneously improves mapping accuracy and yields a map that has a high degree of correspondence with the floor plan. The proposed method is evaluated experimentally, on a publicly available localization dataset from the MIT Stata Centre, and on a dataset acquired with a mobile platform at the University of Zagreb.

16:30
Object Tracking in a Moving Reference Frame

ABSTRACT. Object tracking in a moving frame is becoming a common requirement in a lot of mobile robotic applications, such as search and rescue, monitoring and surveillance, and even in some scientific applications, such as robotic soccer. In all these applications, the robots must be capable of estimating the target position and, sometimes, velocity on their own. Depending on the application and on the current scene situation, the estimates must be more or less accurate, depending on the robot intention to interact with the target, whether to catch it, follow it, etc. The problem is that a robot moves along the working area, having some uncertainty in its pose estimation. This paper proposes an approach based on a Kalman Filter to estimate the object position and velocity, regardless the robot pose. As a testbed, a Middle-Size League soccer robot tracking a moving ball example will be used. This approach allows the agent to follow and interact with a moving object, even if its localization is not available.

16:45
Vision-based Control for an AUV in a Multi-Robot Undersea Intervention Task

ABSTRACT. This paper presents a novel vision-based framework for controlling an Autonomous Underwater Vehicle (AUV). In our application, this AUV is in charge of providing an alternative point of view of a predefined target during a multi-robot intervention mission, where two vehicles cooperate in order to perform the required task. Given this scenario, our framework is based on two main modules: on the one hand, a target detection and tracking module is used to determine the position of the target in the scene; on the other hand, a visual servoing module generates the required velocities for controlling the platform according to the estimated position of the target in the image plane. Results for a set of experiments in different environments are reported and discussed.

17:00
The K-Framed Quadtrees approach for path planning through a known environment

ABSTRACT. One of the most important tasks for a mobile robot is to navigate in an environment. The path planning is required to design the trajectory that generates useful motions from the original to the desired position. There are several methodologies to perform the path planning. In this paper, a new method of approximate cells decomposition, called K-Framed Quadtrees is present, to which the algorithm A-star is applied to determine trajectories between two points. To validate the new approach, we made a comparative analysis between the present method, the grid decomposition, quadtree decomposition and framed quadtree decomposition. Results and implementation specications of the four methods are presented.

17:15
Collision avoidance for Multi-robot Systems with Coincident Paths based on Fictitious Collision Points using Nonlinear Formulation

ABSTRACT. This paper addresses the problem of collision avoidance along specified paths in multiple mobile robot systems. These collisions can be represented by points of intersection or coincident segments between paths. The proposal of the work is to model these segments where the collision is possible through fictitious points. In addition, the advantages of the nonlinear versus mixed integer linear formulation, widely used in the literature, are verified. Comparisons were made and it's proved the superiority of the proposed method with respect to complexity, computational time and inclusion of nonlinear constraints. Moreover, the simulations performed using this technique indicate that the method is promissory for applications in real systems.

16:00-17:30 Session WA1-R2: New particular developments in marine robotics
Location: R2 - Salón de grados
16:00
Coordinated Multi-UAV Exploration Strategy for Large Areas with Communication Constrains

ABSTRACT. The use of multiple low-cost small Unmanned Aerial Vehicles (UAVs) is becoming very popular because of their large number of possible applications, including exploration, search and photography. To accomplish those missions with a high level of autonomy, the vehicles need to be coordinated and each vehicle task assigned. In this work we present a coordination control strategy applied in a searching scenario of oceanographic research buoys taking into account the problem of communication constraints. The proposed control strategy include the assignment of two different teams of UAVs that search and acquire data from buoys with unknown locations in complex coastal or remote oceanic regions. We present a 3-layer architecture that search for the buoys, enables communication with them and creates a UAV communication network with a quality-of-service scheme to recover the data. Extensive simulations illustrate the performance of the proposed coordinated control strategy.

16:15
A Nonlinear Model Predictive Control for an AUV to Track and Estimate a Moving Target using Range Measurements

ABSTRACT. In this paper, we propose a Nonlinear Model Predictive Control system that is employed by an Autonomous Underwater Vehicle (AUV) to track and estimate a moving target using range measurements. The problem of estimating the position of a target using range measurements is a highly nonlinear estimation problem with existence of state and input trajectories of the AUV which makes the system unobservable. A standard stabilizing NMPC based approach augmented with an economic cost function is utilized to steer the system through highly observable trajectories in order to guarantee a good estimate of the position of the target. The efficacy of the proposed solution is demonstrated through simulations.

16:30
Architecture of a Flight Endurance Enhancement System for Maritime Operations with Fixed Wing UAS

ABSTRACT. This paper presents the functional architecture of a flight endurance enhancement system that is suitable for the operation of battery-powered Unmanned Aerial Systems (UAS) in maritime and coastal regions. The flight duration problem is subject of various research efforts; different techniques have arisen to address different aspects aiming to prolong the airborne time with minimum resources. This is a great challenge for low-cost platforms that are typically operated with Lithium-Polymer (LiPo) batteries, which have inherent cost and weight restrictions, and also they have not reached sufficient safety and reliability levels to operate in hazardous environments, such as the ocean. This paper depicts the architecture of novel system for flight endurance enhancement. This system is based on Atmospheric Energy Harvesting (AEH) techniques for exploitation of spatial and temporal wind gradients, which is a bio-mimetic principle observed in the flight of albatrosses in the southern ocean. This paper summarizes the high level and low level functional architecture of the proposed modules, including those that have been fully designed, implemented and tested (wind estimation gen1, wind feature identification, communication framework, trajectory generation) and those that are still subject for research, e.g. trajectory tracking and the next generation wind identification system.

16:45
Immersive Touring For Marine Archaeologywith an AUV. An application of a new compact omnidirectional camera to mapping the Gnalić shipwreck.

ABSTRACT. This paper presents the use of omnidirectional underwater cameras on underwater robots for the rapid high-resolution mapping of shipwrecks for marine archaeology applications. In collaboration with the University of Zadar the methodology was recently demonstrated on the Gnalic shipwreck during the “Breaking the Surface 2016” workshop held in Biograd de Moru (Croatia). The robot was programmed to survey the shipwreck and the data collected was used to build 360º panoramic video, topological panoramic maps and 3D optical reconstructions. The paper presents a recently built multicamera system comprising 5 compact, high resolution video cameras. It outlines the methodology used and reports the results obtained. The paper is completed with the challenges that marine archaeology pose to underwater roboticists, as well as the contribution that this technology may bring to archaeology community.

17:00
Towards Inspection of Marine Energy Devices using ROVs: Floating Wind Turbine Motion Replication

ABSTRACT. Inspection, repair, and maintenance operations are of crucial importance for the safety and survival of marine renewable energy devices. Current ROV manipulator technology used for subsea operations in offshore oil and gas industry is not adequate for offshore renewables. Target devices for inspection and monitoring operations in oil and gas are close to static, unlike their marine energy counterparts which are in motion due to a highly dynamic environment in which they operate. In this paper we describe motion analysis methods for marine energy devices implemented on the OC3-Hywind floating wind turbine in order to understand the conditions in which inspection and intervention operations are to be performed. Additionally, we present experiments carried out on a laboratory rig which is able to replicate the motion of the OC3-Hywind floating platform, and which will be used in future research on control systems for automated visual inspection and intervention using ROV manipulators.

17:15
SHORT INDUSTRIAL PAPER: High reliability unmmanned vessels

ABSTRACT. Developing USV (Unmanned Surface Vehicles) for real applications needs a professional approach to get high reliability systems to increase the operational availability.

UTEK is a new company specialized in the development of high reliability and advanced performances USV.

16:00-17:30 Session WA1-R3a: Cooperative and active perception for robotics
Location: R3a - Aula 002
16:00
Multi-Robot Planning for Perception of Multiple Regions of Interest

ABSTRACT. In this paper we address the problem of allocating perception tasks among a set of multiple robots. We consider a set of target regions of interest in a mapped environment that need to be sensed by any of the robots, and the problem is to find paths for the robots that cover all the target regions with minimal cost. Our approach can be used in many multi-robot problems, from inspection and surveillance to search in structured environments. We use PA* to determine possible perception points for every robot, which we cluster and then use as possible waypoints that can be used to construct paths for all the robots. Given the combinatorial characteristics of path determination in this setting, we use a construction heuristic to find good paths that guarantee full coverage of all the feasible perception target regions. We use a 2D gridmap to represent the environment where ground robots move, and we assume robots are heterogeneous regarding their geometric properties, such as size and maximum perception range. Thus, we contribute an algorithm for heterogeneous robots path planning in perception scenarios with target regions that need to be covered by at least one robot. Our technique reduces the intrinsic combinatorial complexity of this allocation problem, allowing for replanning during execution in case of dynamic obstacles and changes in the map. We consider multiple simulated scenarios where we show the benefits of our approach that enables fast multi-robot planning for perception of multiple regions of interest.

16:15
Low resolution lidar-based multi object tracking for driving applications

ABSTRACT. Vehicle detection and tracking in real scenarios is a key component to develop assisted and autonomous driving systems. Lidar sensors are specially suitable for this task, as they bring robustness to harsh weather conditions while providing accurate spatial information. However, the resolution provided is very scarce in comparison to camera images. In this work we explore the possibilities of Deep Learning (DL) methodologies applied to low resolution 3D lidar sensors such as the Velodyne VLP-16 (PUCK), in the context of vehicle detection and tracking. For this purpose we have developed a lidar-based system that uses a Convolutional Neural Network (CNN) to perform point-wise vehicle detection using PUCK data, followed by a Multi-Hypothesis Extended Kalman Filters (MH-EKF) that estimate the actual position and velocities of those detected vehicles. Comparative studies between the proposed lower resolution (VLP-16) tracking system and a high-end system using Velodyne HDL-64 have been carried out on the Kitti Tracking Benchmark dataset. Moreover, to analyze the influence of the CNN-based vehicle detection approach, comparisons are also performed against a geometric-only detector. The results demonstrate that the proposed low resolution Deep Learning architecture is able to successfully perform the vehicle detection task, outperforming the geometric baseline approach. Moreover, it shows that our system achieves similar tracking performance at close range as the high-end HDL-64 sensor, and at long range, detection is limited at half the distance than the higher-end sensor.

16:30
From Fast to Accurate Wireless Map Reconstruction for Human Positioning Systems

ABSTRACT. Indoor localization systems for humans are becoming commonplace for context-aware applications. In many public areas such as shopping malls or airports, existing wireless infrastructures can be used for localization, often through approaches based on fingerprinting. Although those systems do not require additional installation, a previous calibration phase is needed. This calibration task becomes tedious and time consuming for large scenarios, since the wireless signal must be measured in many different locations. This paper proposes an algorithm to perform this wireless map calibration autonomously by means of a robot. Instead of sampling thoroughly the full scenario from the beginning, our algorithm fosters a more sensible behavior when the calibration time may be limited: first, the robot tries to explore all areas to gain an overall view of the map; and then, it improves the accuracy by sampling more deeply each sector if there is remaining time. For this purpose, full coverage of individual rooms is ranked lower if others are still unexplored. Moreover, we propose some metrics to evaluate this kind of behavior and evaluate our exploration algorithm against a traditional coverage system in two different simulated scenarios.

16:45
Cooperative Cloud SLAM on Matrix Lie Groups

ABSTRACT. In this paper we present a Cooperative Cloud SLAM on Matrix Lie Groups (C2LEARS), which enables efficient and accurate execution of simultaneous localization and environment mapping, while relying on integration of data from multiple agents. Such fused information is then used to increase mapping accuracy of every agent itself. In particular, the agents perform only computationally simpler tasks including local map building and single trajectory optimization. At the same time, the efficient execution is ensured by performing complex tasks of global map building and multiple trajectory optimization on a standalone cloud server. The front-end part of C2LEARS is based on a planar SLAM solution, while the back-end is implemented using the exactly sparse delayed state filter on matrix Lie groups (LG-ESDSF). The main advantages of the front-end employing planar surfaces to represent the environment are significantly lower memory requirements and possibility of the efficient map exchange between agents. The back-end relying on the LG-ESDSF allows for efficient trajectory optimization utilizing sparsity of the information form and exploiting higher accuracy supported by representing the state on Lie groups. We demonstrate C2LEARS on a real-world experiment recorded on the ground floor of our faculty building.

17:00
Circle Formation in Multi–Robot Systems with Limited Visibility

ABSTRACT. Pattern Formation in multi-robot systems was proposed in the 1990s. Since then it has been extensively studied and applied in various ways. To date, the majority of the proposed algorithms that aimed to achieve geometric patterns in the literature have overlooked the visibility limitation in physical robots. In addition, a methodology to reach a complete coordinate agreement has not been adopted by many researchers as a prerequisite towards a successful formation. It should be stressed that such limitation and methodology have a strong effect on the desired pattern approach. In this paper, a decentralized algorithm for circle formation is highlighted. The main advantage of forming a circle is the flexibility to be generated with different initial distributions. Moreover, circle arrangement can be utilized as a preliminary sub-task for more complex activities in multi-robot systems. To handle the aforementioned realities, this approach is proposed under a realistic robot model i.e. one that has a short visibility range and performs the task autonomously relying on the information picked by itself, or by the vicinity. In addition, robots do not initially have a predefined leader nor unique IDs. Simulation results have validated the robustness and flexibility of the proposed algorithm, where a circular pattern has been successfully constructed in a self-organized manner.

17:15
A Multidrone Approach for Autonomous Cinematography Planning

ABSTRACT. This paper proposes a multidrone approach for autonomous cinematography planning. The use of drones for aerial cinematography is becoming a trend. Therefore, the aerial cinematography opens a new field of application for autonomous platforms that need to develop intelligent capabilities. This becomes even more challenging if a team of multiple drones are considered for cooperation. This paper introduces the novel application of planning for cinematography, including the challenges involved and the current state of the art. Then, it proposes a first version of an architecture for cooperative planning in cinematography applications, like filming sport events outdoors. The main features for this architecture are the following. The system should be able to reproduce typical shots from cinematography rules autonomously, shooting static and mobile targets. It should also ensure smooth transitions along the shots, implementing collision avoidance and being aware of no-fly zones, security and emergency situations. Finally, it should take into account the limited resources of the drones (e.g. battery life).

16:00-17:30 Session WA1-R4: Ontologies and knowledge representation for robotics
Location: R4 - Sala de Juntas
16:00
Heterogeneous ontologies and hybrid reasoning for service robotics: the EASE framework

ABSTRACT. As robots are expected to accomplish human-level manipulation tasks, the demand for formal knowledge representation techniques and reasoning for robots increases dramatically. In this paper we describe how to make use of heterogeneous ontologies in service robotics. To illustrate the vision, we take the action of pouring as an example.

16:15
Towards an Ontology for Task and Planning in Autonomous Systems: an Emergency Scenario

ABSTRACT. Emergency scenarios where chemical explosions or fire outbreaks take place become dangerous environments to send a human team to. UAVs and UGVs working together with the human rescue team can prevent endangering the human team. To perform the emergency tasks, it is necessary to define what the UAV/UGV robotic system will do in the emergency area, detailing what each of the robots will specifically carry out. Next, the different actions should be planned to perform such tasks. An ontology--based approach is described in this paper, where the cohesive element to specify the planning process is a task ontology. We describe the initial contents of our ontology for task and planning in autonomous robots, with an example of its use within an emergency scenario where a combined UAV/UGV robotic system is supposed to act.

16:30
Knowledge and Tasks Representation for an Industrial Robotic Application

ABSTRACT. The paper presents an implementation of knowledge representation and task representation, based on ontologies for an Industrial Robotic Application. The industrial application is to insert up to 56 small pins, sealants, in a harness box terminal for the automotive industry. The number of the sealants and its insertion pattern vary significantly with the production requests. Based on the knowledge representation of the robot and also from the tasks to be performed, plans are built and then sent to the robot controller based on the seal pattern production order. Moreover, the robotic system is capable to do a re-planning when an insertion error is reported by a machine vision system. The ontology based approach was used to define the robot, the machine vision system, and the tasks that are needed to be performed by the robotic system. The robotic system was validated experimentally, i.e., showing its capability to correct seals insertion errors, while re-planning itself

16:45
Physics-based Ontology for Manipulation Planning

ABSTRACT. In manipulation planning, dynamic interaction between the objects and the robots is playing a significant role. In this scope, physics-based engine such as ODE allows to consider the dynamic interaction between rigid bodies for motion planning. In this context, the knowledge representation of manipulation planning can reduce the computational cost by allowing a semantic-based reasoning on manipulation regions and behaviors. In this work, physics-based ontologies are used to identify the manipulation regions and manipulation behaviors. A semantic map is constructed to categorize and assign the manipulation constraints based on, the robot, and object constraints, and the type of actions. The proposed follows the RoSCa ontological framework and is illustrated with didactic examples scenarios.

17:00
What Can Ontologies Do for Robot Design?

ABSTRACT. In this paper we address the problem of automatic design of the abstract structure of a robot. The design is driven by the desired behaviors which the robot should be able to perform. To this aim, an extension for the IEEE Standard Ontology for Robotics and Automation has been developed. We present an intelligent system which infers abstract, dimensionless robot designs from this ontology by relating robot actions to necessary structural parts. Then, these abstract structures can be materialized into physical robots that should be able to perform requested behaviors. We show this implementation using a modular robotics platform as a demonstrator.

17:15
Deep Semantic Abstractions of Everyday Human Activities: On Commonsense Representations of Human-Object Interactions

ABSTRACT. We propose a deep semantic characterisation of space and motion categorically from the viewpoint of grounding embodied human-object interactions. Our key focus is on an ontological model that would be adept to formalisation from the viewpoint of commonsense knowledge representation, relational learning, and qualitative reasoning about space and motion in cognitive robotics settings. We demonstrate key aspects of the space \& motion ontology and its formalisation as a representational framework in the backdrop of select examples from a dataset of everyday activities. Furthermore, focussing on human-object interaction data obtained from RGBD sensors, we also illustrate how declarative (spatio-temporal) reasoning in the (constraint) logic programming family may be performed with the developed deep semantic abstractions.

16:00-17:30 Session WA1-R5: Vision and Learning for robotics (I)
Location: R5 - Aula 009
16:00
Robot Semantic Localization through CNN Descriptors

ABSTRACT. Semantic localization for mobile robots involves an accurate determination of the kind of place where a robot is located. Therefore, the representation of the knowledge of this place is crucial for the robot. In this paper we present a study for finding a robust model for scene classification procedure for a mobile robot. The proposed system uses CNN descriptors for representing the input perceptions of the robot. First, we develop comparative experiments in order for finding a model. Experiments include the evaluation of several pre-trained CNN models and training a classifier with different classifications algorithms. These experiments were carried out using the ViDRILO dataset and compared with the benchmark provided by their authors. The results demonstrate the goodness of using CNN descriptors for semantic classification.

16:15
3D object mapping using a labelling system

ABSTRACT. 3D data has arisen as the most used information for environment representation thanks to the advent of low cost RGB-D cameras. We propose a 3D map representation that uses not only depth information but the information provided by an expert system. This expert consists on a Convolutional Neural Network trained with deep learning techniques for scene labelling purposes. For every partial 3D map captured, we receive a set of labels with their associated probability of presence in that scene. The final map is obtained by registering and merging all these partial maps. The semantic labels from the expert system are used to recognise and locate objects in the environment.

16:30
Robust Hand Pose Regression Using Convolutional Neural Networks

ABSTRACT. Hand pose estimation is useful for several human-computer interaction applications, like sign language recognition, the identification of more complex behaviors such as hand gestures and interaction in virtual reality applications. In this work, we propose a system which is able to predict the 2D hand joints using a monocular color camera. To do that, we propose to use a 3D hand tracking sensor for collecting ground truth information that is projected to the camera image plane. We present a novel pipeline that leverages deep learning techniques for hand hand pose estimation. The proposed Convolutional Neural Networks (CNN) is able to infer the joints of the hand from an image without the need of any additional sensor.

16:45
3D semantic maps for scene segmentation

ABSTRACT. The semantic segmentation problem has been widely studied in the computer vision community. However, state-of-the-art solutions based on deep learning are only available for 2D images. The lack of large annotated datasets makes more difficult the training of models with 3D images. In this work we propose to use the already available 2D deep learning based solutions to semantically segment the 3D environment for robotic applications. Concretely, deep learning applications provide the semantic labeling, and the geometrical information from RGB-D cameras along with the robot pose provides the 3D position.

17:00
Movement Direction Estimation using Omnidirectional Images in a SLAM algorithm

ABSTRACT. This work presents a method to estimate the movement direction of a mobile robot using only visual information, without any other additional sensor. This visual information is provided by a catadioptric system mounted on the robot and formed by a camera pointing towards a convex mirror. It provides the robot with omnidirectional images that contain information with a field of view of 360 degrees around the camera-mirror axis. A SLAM algorithm is presented to test the method that estimates the movement direction of the robot. This SLAM method uses two different global appearance descriptors to calculate the orientation of the robot and the distance between two different positions. The method to calculate the movement direction is based on landmarks extraction, using SURF features. A set of omnidirectional images has been considered to test the effectiveness of this method.

17:15
How to transfer an autonomous driving model for semantic segmentation to other domains?

ABSTRACT. Semantic scene understanding is an important task for robots operating autonomously in real-world applications. Recent deep convolutional neural networks (CNNs) have demonstrated to be an effective approach for semantic image segmentation, especially for tasks where plenty of labeled data is available. However, many applications need to learn new specific classes but do not have a lot of labeled training data. This paper addresses the problem of transferring the knowledge from existing CNN models to different classes to those the model has been trained for. Our work explores the two common transfer learning approaches for the particular problem of semantic segmentation: 1) fine-tuning existing models with the new training data, following a standard pipeline; 2) training a superpixel classifier using our proposed superpixel representation, which combines local and context information. We evaluate both approaches on three varied binary segmentation use cases using small public datasets. Our experiments demonstrate the advantages and limitations from each alternative studied, showing that the proposed superpixel based strategies provide higher accuracy because they can learn better with limited amounts of labeled data.

17:30-18:00Coffee Break
18:00-19:30 Session WA2-R1: General track: Mobile Robotics Applications
Location: R1 - Sala Juan Larrañeta
18:00
Integrating an autonomous robot on a dance and new technologies festival

ABSTRACT. This paper presents the results of a project to integrate an autonomous mobile robot into a modern dance performance at a dance and new technologies festival. The main goal is to integrate a simple low cost mobile robot into the dance performance, in order to study the possibilities that this kind of platforms can offer to the artists. First, this work explains the process and design to embed the robotic platform into the choreography theme. Another contribution described in this work is the system architecture proposed and built to make the robot behaviours match the artists requirements: precise, synchronized and robust robot movements. Finally, we discuss the main issues and lessons learned for this kind of robotics and arts applications and summarize the results obtained, including the successful final live performance results.

18:15
A Perspective of Security for Mobile Service Robots

ABSTRACT. Future homes will contain Mobile Service Robots (MSR) with diverse functionality. MSRs act in close proximity to humans andhave the physical capabilities to cause serious harm to their environment. Furthermore, they have sensors that gather large amounts of data, which might contain sensitive information. A mobile service robot’s physical capabilities are controlled by networked computers that are susceptible to faults and intrusions. The proximity to humans and the possibility to physically interact with them makes it critical to think about the security issues of MSRs. In this work, we investigate possible attacks on mobile service robots. We survey adversary motivations to attack MSRs, analyse threat vectors and list different available defence mechanisms against attacks on MSRs.

18:30
Project and Trajectory Control of an Autonomous Aerator for Aquaculture

ABSTRACT. Aquaculture has resorted to the use of technology to make control of important variables and improve production. There are two problems observed in the Marine Aquaculture Station - EMA of the Federal University of Rio Grande (Brazil), more specifically related to the aeration of the culture tanks to pisciculture and carciniculture. Firstly, the high energy consumption of the paddle aerators was identified, due to the continuous need to oxygenate the water. Subsequently, it was noted that static aerators do not aerate the water evenly and its require that their placement be done manually by technicians or assistants. Thus, the article proposes a design of an autonomous aerator with electric power coming from a photovoltaic solar energy system. In the sequence, a navigation system for the aerator is proposed along the tank, based on a differential performance of the motors connected to the paddle. This paper also presents the kinematic modeling of the system and a control of trajectory simulation.

18:45
On-line adaptive side-by-side human robot companion to approach a moving person to interact

ABSTRACT. Abstract. In this paper, we present an on-line adaptive side-by-side human-robot companion to approach a moving person to interact with. Our framework makes the pair robot-human capable of overpass, in a joint way, the dynamic and static obstacles of the environment while they reach a moving goal, which is the person who wants to interact with the pair. We have defined a new moving final goal that depends on the environment, the movement of the group and the movement of the interacting person. Moreover, we modified the Extended Social Force model to include this new moving goal. The method has been validated over several situations in simulation.

19:00
A robotized dumper for debris removal intunnels under construction

ABSTRACT. Tunnels in construction exhibit many challenges for automation. In this work we address the robotization of a conventional dumper for debris removal during the construction of tunnels, in the framework of a technological transfer project. The goal is to convert a dumper into an autonomous vehicle capable of planning, navigate and localize itself. Planning and navigation techniques have been adapted to the especial kinodynamic characteristics of the vehicle. The difficulties for having a precise continuous localization in this kind of scenarios, due to the irregularities of the terrain, the changing illumination and the own scenario, have driven to develop hybrid localization techniques to integrate continuous and discrete information, coming from the navigation sensors, some semantic geometric features, and the signal strength propagation in tunnel scenarios. Simulation and real-world experiments are described, and some preliminary results are discussed.

19:15
Roboethics and Robotic Governance – a Literature Review and Research Agenda

ABSTRACT. Roboethics is an often-discussed topic in recent years. It has been addressed by science-fiction authors and Hollywood as well as by governments and scientists in various different ways. There seems to be a general consensus on the importance of the issue, but there are still many different approaches that did not achieve a great deal of awareness. Robotic Governance provides an opportunity to consolidate all these approaches and to start a discussion with all involved stakeholders in order to reach a consensus on research, manufacturing and use of technologies in the fields of robotics, automation as well as artificial intelligence. This is the first step to realize sustainable robotics and to successfully and consciously shape the future of coming generations of robotic natives without compromising our own needs.

18:00-19:30 Session WA2-R2: Agricultural robotics and field automation
Location: R2 - Salón de grados
18:00
Robot Localization System in a Hard Outdoor Environment

ABSTRACT. Localization and Mapping of autonomous robots in a hard and unstable environment is a challenging research topic. Typically, the commonly used Dead Reckoning systems can fail due to the harsh conditions of the terrain and the accurate Global Position System can be considerably noisy or not always available. One solution is using Wireless sensors in a network as landmarks. In this paper, Ultra-Wideband Time-of-flight based technology (Pozyx), that can be a cost-effective solution for application in a robotic localization system, is deeply studied and characterized. Moreover, an EKF Localization filter that fuses Odometry with the Pozyx Range measurements is implemented and compared with the default Pozyx Algorithm. The obtained results are presented and discussed and allow to present formulations for better results of Beacons Mapping Procedure (BMP) required for accurate and reliable localization systems.

18:15
Improving Dead-Reckoning Performance of Skid-Steered Wheeled Robots Using an Improved Kinematic Model

ABSTRACT. Since skid-steered platform rotations have been non-trivial, existing kinematics approaches for skid-steered wheeled robots have either ignored the relationship between the rotational speed of the robot and its wheel speeds, or approximated it with a linear function, which was sometimes estimated as a function of track instantaneous center of rotations. In this paper, we propose a new kinematic model for skid-steered wheeled platforms taking into account not only the geometric properties of the robot, but also location of its center of mass, which changes the kinematics of the robot considerably due to its effect on traction of the wheels. The proposed model was verified experimentally by varying the center of mass of the robot, and its improvements to dead-reckoning performance were also experimentally verified.

18:30
A*-Based Solution to the Coverage Path Planning Problem

ABSTRACT. Coverage Path Planning (CPP) is an essential problem in many applications of robotics, including, but not limited to autonomous de-mining and farming. In this paper we propose a solution that is based on the A* search algorithm for grid based environments. The proposed approach achieves energy efficiency in cluttered environments by reducing the number of rotations. The running time of the algorithm is kept in the order of seconds for moderate sized environments thanks to a selective branching approach and a cost function that can predict the minimum future cost of each node in the search tree.

18:45
GRAPE: Ground Robot for vineyArd Monitoring and ProtEction

ABSTRACT. Precision agriculture is a key topic in robotics. The incursion of the robots in the agriculture is becoming a reality the last years and, nowadays, robotics are not only used for crop monitoring, like aerial inspection for growth control, but is taking a key role in the daily life of the farmers and producers. Heavy tasks like pruning, seeding or even precise harvesting are the target topics of the robotics in agriculture.

In this paper we present the ongoing work of GRAPE project, founded by Echord++ program from European Commission. GRAPE consist in the automatic pheromone dispenser distribution for matting disruption in vineyards using an autonomous ground robot equipped with a robotic arm. GRAPE is focus on developing the on-board intelligence and the required algorithms, using commercial hardware. The consortium of the project is composed by Eurecat Technology Center, University Politecnico di Milano and Vitirover.

19:00
Automatic recharging system for steep-slope vineyard robots

ABSTRACT. Develop cost-effective ground robots for crop monitoring in steep slope vineyards is a complex challenge. The terrain presents harsh conditions for mobile robots and most of the time there is no one available to give support to the robots. So, a fully autonomous steep-slope robot requires an automatic recharging system. This work proposes a multilevel system to monitor the robot energy autonomy, to plan the trajectory to the near recharging point and to dock the robot to that recharging point considering visual tags. The proposed system called VineRecharge was developed to be deployed into cost-effective robots with low computational power.

18:00-19:30 Session WA2-R3a: WAF: Robots for healthcare
Location: R3a - Aula 002
18:00
Appropriate Spherical Coordinate Model for Trocar Port Constraint in Robotic Surgery

ABSTRACT. This article deals with the representation of the fulcrum point virtually created when inserting an endoscope in the human body during a surgery application. Several robotic-based surgical solutions use the well-known spherical coordinates to define the pose of the robotic endoscope tip, since it can only represents motions that satisfy this single insertion point constraint. We demonstrate in this article that the traditional spherical model may not be totally suitable for surgical applications. First of all, such model presents a representation singularity within the work space. Furthermore, control strategies based on that model may not produce optimal motions of the robotic arm holding the endoscope. We propose an adjustment of the traditional spherical model to alleviate these two issues. We also propose a very simple but efficient control task for servoing a robotic arm holding an endoscopic camera to center an instrument in the camera field of view. Simulations are provided for demonstrating the correctness and efficiency of the proposed model and control approach.

18:15
Practical aspects of deploying Robotherapy systems

ABSTRACT. The aging of the population presents new challenges that must be solved. One of them is the need to care for the elderly. Robots could help us in the not so distant future. In this paper, we present a robotherapy system, along with some practical issues that must be taken into account in its deployment. Our robotic system helps therapists in sessions of cognitive stimulation. Without taking into account aspects such as the patient's perception of the robot, or the impact of the cultural environment, the application of these systems may be doomed to failure. In this paper we describe the implantation of the system in a elderly residence in Casillas del Ángel, Fuerteventura, which is very far from where the system was originally developed, with the implications that this factor entails. We will present evidences of how the adaptation to their culture has been considered decisive in their success. We will present the results of the implementation, as well as the therapists' perception of their effectiveness and the use of healthcare technologies in elderly care.

18:30
Enhancing a Robotic Rehabilitation Model for Hand-Arm Bimanual Intensive Therapy

ABSTRACT. The NAOTherapist platform is a robotic framework that aims at developing socially-interactive rehabilitation sessions for pediatric patients with physical impairments. Although this therapeutic tool has been already assessed with the target patients in a long-term evaluation, the system is planned to participate in an Intensive Therapy Camp for Cerebral Palsy patients. This presents new challenges and requirements that must be considered to provide a better daily experience to the involved participants. This work describes how the robotic rehabilitation model has been improved for both the inclusion of new games and the individual adaptation to the users.

18:45
LifeBots I: Building the Software Infrastructure for Supporting Lifelong Technologies

ABSTRACT. The goal of the LifeBots project is the study and development of long-life mechanisms that facilitate and improve the integration of robotics platforms in smart homes to support elder and handicapped people. Specifically the system aims to design, build and validate an assistive ecosystem formed by a person living in a smart home with a social robot as her main interface to a gentler habitat. Achieving this goal requires the use and integration of different technologies and research areas, but also the development of those mechanisms in charge of providing an unified, pro-active response to the user’s needs. This paper describes some of the mechanisms implemented within the cognitive robotics architecture CORTEX that integrates deliberative and reactive agents through a common understanding and internalizing of the outer reality, that materializes in a shared representation derived from a formal graph grammar.

19:00
CLARC: A Cognitive Robot for Helping Geriatric Doctors in Real Scenarios

ABSTRACT. Comprehensive Geriatric Assessment (CGA) is an integrated clinical process to evaluate the frailty of elderly persons in order to create therapy plans that improve their quality of life. For robotizing these tests, we are designing and developing CLARC, a mobile robot able to help the physician to capture and manage data during the CGA procedures, mainly by autonomously conducting a set of predefined evaluation tests. Building around a shared internal representation of the outer world, the architecture is composed of software modules able to plan and generate a stream of actions, to execute actions emanating from the representation or to update this internal representation by including/removing items at different abstraction levels. Percepts, actions and intentions coming from all software modules are grounded within this unique representation using the same set of symbols. This allows the robot to react to unexpected events and to modify the course of action according to the dynamics of a scenario built around the interaction with the patient. The paper describes the architecture of the system as well as the preliminary user studies and evaluation to gather new user requirements.

19:15
Desarrollo de un sistema robótico de triaje rápido para situaciones de catástrofe

ABSTRACT. Los desastres naturales han sido un punto crucial de colaboración para diferentes ramas de la ciencia. La clasificación y evaluación de las víctimas se utilizan como un método de triaje fuera del hospital, para priorizar el orden de atención o eva-cuación de las víctimas. Los signos vitales y criterios médicos tratan de establecer una clasificación correcta de los pacientes. Además, se persigue un uso óptimo de los recursos. El problema se agrava en casos en los que las evaluaciones se reali-zan en espacios inaccesibles. Bajo este escenario y gracias al avance de la tecno-logía, la robótica se convierte en una herramienta colaborativa para el triaje rápido. Los robots pueden estructurarse para trabajar en espacios complejos o inaccesi-bles para los seres humanos. En consecuencia, este artículo propone un sistema para el triaje de personas que han quedado atrapadas bajo escombros y permane-cen en espacios confinados o inaccesibles al personal de rescate por motivos de seguridad; por ejemplo, puede existir riesgo contra su integridad física por resi-duos tóxicos o radiación. El sistema se basa en un sistema de navegación con de-tección de obstáculos implementado en un robot (KUKA youBot) equipado con un sistema de procesamiento de imágenes RGB-D para la detección de las perso-nas. Se aplican algoritmos para la detección sin contacto del latido cardiaco y la respiración de la persona localizada. Por lo tanto, la detección de signos vitales se realiza in situ mediante un robot de exploración que procesa toda la información utilizando el Sistema Operativo Robótico (ROS).

18:00-19:30 Session WA2-R4: Simulation in robotics
Location: R4 - Sala de Juntas
18:00
Artificial Flagellum Microrobot. Design and Simulation in COMSOL

ABSTRACT. This paper presents a study of motion in viscous environment with piezoelectric actuators. The propulsion action relies on the creation of a non-reciprocal motion along a piezoelectric layered beam divided into several segments. This requires that a voltage with the same frequency but different phases and amplitudes be applied to each segment. The motion pattern is analyzed theoretically and a control strategy in open loop is implemented to emulate a non-reciprocal motion. Simulations in COMSOL are carried out and a non-reciprocal movement is developed and with that the propulsion is verified with drag force. It was found that despite extreme size limitations using piezoelectric material is able to swim in blood flow.

18:15
Knowledge-oriented Physics-based Motion Planning for Grasping in the Presence of Uncertainty

ABSTRACT. Grasping an object in unstructured and uncertain environments is a challenging task, particularly when a collision-free trajectory does not exits. Highlevel knowledge and reasoning processes, as well as the allowing of interaction between objects, can enhance the planning efficiency in such environments. In this direction, this study proposes a knowledge-oriented physics-based motion planning approach for a hand-arm system that uses a high-level knowledge-based reasoning to partition the workspace into regions to both guide the planner and reason about the result of the dynamical interactions between rigid bodies. The proposed planner is a kinodynamic RRT that uses a region-biased state sampling strategy and a smart validity checker that takes into account uncertainty in the pose of the objects. Complex dynamical interactions along with possible physicsbased constraints such as friction and gravity are handled by the Open Dynamic Engine that is used as the RRT state propagator. The proposal is validated for different scenarios in simulation and in real environment using a 7-degree-offreedom KUKA Lightweight robot equipped with a two-finger gripper. The results show a significant improvement in the success rate of the execution of the computed plan in the presence of object pose uncertainty.

18:30
A Simulation Tool for Visualizing the Assembly Modes and Singularity Locus of 3RPR Planar Parallel Robots

ABSTRACT. This paper presents a graphical and intuitive tool for simulating the forward kinematics of planar parallel 3RPR robots with arbitrary geometric design. The proposed tool allows the user to visualize the singularity locus of the robot and the evolution of all the solutions to its forward kinematic problem in the complex plane. The user can modify all the geometric design parameters of the robot and instantaneously visualize the effect of these modifications on the singularity locus. As the presented examples illustrate, the proposed tool is especially useful for visualizing the coalescence of different solutions of the forward kinematic problem when approaching higher-order singularities, as well as for visualizing how these special singularities transform when perturbing the different geometric parameters of the robot.

18:45
On the use of mixed reality for setting up controls and coordination strategies for teams of autonomous UAV

ABSTRACT. A mixed reality simulation framework is being developed as a tool to facilitate the elaboration, testing and deployment of control and collaborative strategies for teams of UAVs. The virtual world within the framework must contain a model of the phenomenon under analysis. It has been shown that, for complex cases, the use of real UAVs in an initiation phase could serve to simplify this model while increasing its accuracy. In a second step, a subsequent intermediate phase is implemented now. In this phase the virtual model is first scaled and then is used to provide measurement data to the real planes that are equipped with virtual sensors in an augmented reality scenario. This way the cost and time of checking the coordination strategies and communications when several real planes are flying simultaneously can be greatly reduced. Once everything is tuned and adjusted within this intermediate phase, the whole system could be implemented in the full size real environment. An application on pollutant plume dispersion is used as a workbench case to show how this procedure is implemented in practice.

19:00
Flexible Work Cell Simulator using Digital Twin Methodology for Highly Complex Systems in Industry 4.0

ABSTRACT. The continuous evolution in manufacturing processes has attracted substantial interest from both scientific and research community, as well as from industry. Despite the fact that streamline manufacturing relies on automation systems, most production lines within the industrial environment lack a flexible framework that allows for evaluation and optimisation of the manufacturing process. Consequently, the development of a generic simulators able to mimic any given workflow represent a promising approach within the manufacturing industry. Recently the concept of digital twin methodology has been introduced to mimic the real world through a virtual substitute, such as, a simulator. In this paper, a solution capable of representing any industrial work cell and its properties is presented. Here we describe the key stages of such solution which has enough flexibility to be applied to different working scenarios commonly found in industrial environment.

18:00-19:30 Session WA2-R5: Vision and learning for robotics (II)
Location: R5 - Aula 009
18:00
Automatic selection of user samples for a non-collaborative face verification system

ABSTRACT. This paper describes the challenges that involve developing a software capable of capturing users' faces on mobile devices in a non-collaborative environment. The goal is to generate a set of quality training samples of the user's face for the construction of a model that can be used in a later phase of biometric identification. To this end, a supervised learning system is integrated to determine when a photo should be taken. This learning is supported by a varied input data set that contains information regarding the pose of the device, its manipulation and other environmental factors such as lighting. The software also has different ways of working with the objective of not wasting resources and be little invasive. Working modes are managed with an easy-to-maintain and scalable rules-based system. The experimental results show the robustness of the proposal.

18:15
Detecting and manipulating objects with a social robot: an ambient assisted living approach

ABSTRACT. Object grasping in domestic environments using social robots has an enormous potential to help dependant people with certain degree of disability. In this work, we made use of the well-known Pepper social robot to carry out such task. We provide an integrated solution using ROS to recognize and grasp simple objects. That system was deployed on an accelerator platform (Jetson TX1) to be able to perform object recognition in real time using RGB-D sensors attached to the robot. By using our system, we proved that the Pepper robot shows a great potential for such kind of domestic assistance tasks.

18:30
Learning leg pattern using laser lange finder in mobile robots

ABSTRACT. In spite of the advances on people detection and tracking during last years, included the skeleton based trackers, it is interesting to use different types of sensor in this task in order to achieve a more robust people detection and tracking algorithm. This work focus its attention on a laser rangefinder approach for people detection and tracking. Patterns of leg are learnt from 2d laser data using machine learning algorithms. Unlike others leg detection approaches, people can be still or moving at the surroundings of the robot. The method of leg detection is used as observation model in a particle filter to track the motion of a person.

18:45
Inertial navigation with mobile devices: A robust step count model

ABSTRACT. Navigation is an essential feature for smartphones, even indoors. Having a robust step count algorithm is the cornerstone for building an inertial navigator based on accelerometer sensors. However, accelerometer data is very sensitive to body movements, so separating noise from real steps is not a trivial issue. Our main hypothesis is that MSE measured between predicted and real signal gives a clear distinction between ideal steps, noisy steps and pure noise. In this paper we propose a combination of techniques to obtain a robust step count model for smartphones. Using the vertical component of the acceleration, Support Vector Regression (SVR) for modeling user’s activity and an algorithm that combines peak-valley detection with high Mean Squared Error (MSE) steps filtering, we achieve a computational efficient and robust model for detect steps.

19:00
Fine 3D path following of a quadcopter

ABSTRACT. This paper addresses the design and implementation of a path following controlling system for a drone which relies on 3D localization by visual markers. The program is designed only for indoor flight, special attention is paid to accuracy of the position estimation algorithm, robustness of the path following controller and real time operation. The system is composed of two components, one responsible of the image analysis and 3D pose estimation and another responsible of the drone navigation. The system has been experimentally validated both in Gazebo simulator and in a real drone.