Wednesday, June 29th
The goal of the RoboCup initiative is that “By the mid-21st century, a team of fully autonomous humanoid soccer players shall win the soccer game, comply with the official rule of the FIFA, against the winner of the most recent World Cup”. As the IAV 2016 is co-located with the 20th RoboCup, it is time to review what has been achieved in the area of humanoid robots that autonomously play soccer in teams.
This paper describes the development of an optimized path planning algorithm for automated vehicles in urban environments. This path planning is developed on the basis of urban environments, where Cybernetic Transportation Systems (CTS) will operate. Our approach is mainly affected by vehicle's kinematics and physical road constraints. Based on this assumptions, computational time for path planning can be significantly reduced by creating an off-line database that already optimized all the potential trajectories in each curve the CTS can carry out. Therefore, this algorithm generates a database of smooth and continuous curves considering a big set of different intersection scenarios, taking into account the constraints of the infrastructure and the physical limitations of the vehicle. According to the real scenario, the local planner selects from the database the appropriate curves from searching for the ones that fit with the intersections defined on it. The path planning algorithm has been tested in simulation using the previous control architecture. The results obtained show path generation improvements in terms of smoothness and continuity. Finally, the proposed algorithm was compared with previous path planning algorithms for its assessment.
Thanks to the continuously increasing computational power of CPUs, nowadays Model Predictive Control (MPC), initially designed for multi-variable control of chemical plants, is adopted in a large number of different applications, covering a wide range of process dynamics ranging from slow to even fast time scales. As a consequence, a renewed interest in numerical optimization tools, especially for nonlinear systems, arose. This work aims at demonstrating the feasibility of a novel methodology, based on a Linear Fractional Transform (LFT) formulation of the system dynamics, that allows to efficiently solve nonlinear MPC problems. An application example, concerning the tracking control of an autonomous vehicle, shows the effectiveness of the proposal.
In the last few years the number of applications of autonomous mobile robots to outdoor and off-road tasks, like border surveillance and monitoring, search and rescue, agriculture, driveless mobility has been rapidly increasing. Among the huge number of functionalities that are required to make a vehicle an autonomous robot, localisation, path planning and trajectory tracking are the most important. This paper proposes a novel approach to solve the trajectory tracking problem of an Ackermann steering vehicle. A multi-body dynamic model of the vehicle is proposed as a mean to tune and validate the tracking controller. The proposal is supported by an experimental validation, that shows the effectiveness of the trajectory tracking controller and its performance in comparison with the accuracy of the localisation algorithm.
This paper presents a nonlinear path tracking controller for autonomous bi-steerable (four-wheel steering or 4WS) vehicles, allowing high precision tracking even when the reference path proposes fast varying curvature. Indeed, such paths are very common as soon as vehicles have to avoid obstacles in cluttered environments. Considering the well-known bicycle model, the sole reference path usually defined for the rear wheel is replaced by two synchronized paths, introducing a new way to calculate the expected yaw rate of the vehicle without numerical derivatives. Equations describing the motion of the vehicle with respect to this dual-path are presented and used to design the proposed control law. Then, simulations and experiments with the electrical public transport vehicle EZ10 demonstrate the controller ability to precisely follow complex trajectories.
In this work we address simultaneous pose tracking and sensor parameter self-calibration by applying the pose-graph optimization approach. A factor-graph is employed to store robot pose estimates at different time instants and calibration parameters such as magnetometer hard and soft iron distortion and gyroscope bias. Specific factors are developed in this paper to handle Ackermann kinematic readings, inertial measurement units, magnetometers and global positioning systems in the pose-based factor-graph. An experimental evaluation supports the viability of the approach considering an autonomous all-terrain vehicle, for which we perform calibration and real-time pose tracking during navigation.
An energy efficient local path planner for navigation of omnidirectional battery-powered mobile robots in dynamic environments. The proposed algorithm extends the Dynamic Window Approach (DWA) to incorporate a cost function based on energy consumption. The estimated energy consumption during planning is predicted using a linear regression model that is learned on the fly using a variable learning rate. Empirical results are presented on a mobile robot platform that show a 9.79\% decrease in energy consumption in comparison to the DWA approach.
Inaccurate actuator responses affect the behavior of an autonomous driving system. A disadvantageous combination of such inaccuracies might lead to a collision, but is hard to test in advance due to the exponentially large number of possible combinations. This paper introduces STARVEC, a tool to test autonomous driving systems for undesired behaviors in the presence of sensor and actuator inaccuracies in a simulation environment. It stores intermediate states of the simulation and uses these states to efficiently explore the space of possible behaviors. Each step continues with the execution of the state with the highest distance to its neighbors. Thus, the potentially large space of reachable states is covered fast and increasingly dense. The approach is applied to an autonomous parking system with inaccurate actuators and its performance is compared to a Monte-Carlo algorithm and a previous prototype.
In the development of software-intensive systems in a vehicle, like an autonomous driving system, defects are often only recognized during trials on the physical vehicle. In contrast to a simulation environment, a physically executed maneuver does not offer the possibility to pause and debug critical code sections or to reproduce and repeat faulty trials. Furthermore, development space and capacities are limited inside the car. Therefore, it is best practice to analyze faults observed during a physical execution offline and to reproduce faulty trials in a simulation environment. The repetition in a simulation environment is a time consuming effort but necessary for pushing the software component towards a state in which it showed the faulty behavior. This paper shows an approach for executing the faulty state again in a simulation environment by serializing the exact state of the software system and summarizes practical experience gained by this approach.
The work presented in this paper focuses on the comparison of well-known and new techniques for designing robust fault diagnosis schemes in the robot domain. Correctly identifying and handling faults is an inherent characteristic that all autonomous mobile agents should possess, as none of the hardware and software parts used by robots are perfect; instead, they are often error-prone and able to introduce serious problems that might endanger both robots and their environment. Based on a study of literature covering model-based fault-diagnosis algorithms, we selected four of these methods based on both linear and non-linear models. We analyzed and implemented them in a mathematical model, representing a four-wheel-OMNI mobile robot. Numerical examples were used to test the ability of the algorithms to detect and identify abnormal behavior and to optimise the model parameters for the given training data. The final goal was to point out the strengths of each algorithm and to figure out which method would best suit the demands of fault diagnosis for a particular mobile robot.
The verification of safety is expected to be one of the largest challenges in the commercialization of autonomous vehicles. Using traditional methods would require infeasible time and resources. Recent research has shown the possibility of using near-collisions in order to estimate the frequency of actual collisions using Extreme Value Theory. However, little research has been done on how the measure for determining the closeness to a collision affect the result of the estimation. This paper compares a time-based measure against one that relates to an inevitable collision state. The result shows that the latter one is the more robust choice and that more research needs to be made into measure of collision proximity.
In this paper we studied a system fault detection and isolation in a networked Multi-UAVs formation flight set-up using a Cubature Kalman Filter (CKF). Both actuator and sensor faults of a UAV are considered as an agent node fault on the system of UAVs in the formation flight. The CKF based fault detection scheme developed is used in order to detect a system wide fault in the formation flight. Furthermore, the graph theoretic approach used for modeling the multi agent UAV's communication is exploited to isolate the faulty UAV (node) from the flight formation. A numerical simulation is presented to confirm the proposed fault detection and isolation (FDI) performance.
The development of precise and robust navigation strategies for Autonomous Underwater Vehicles (AUVs) is fundamental to reach the high level of performance required by complex underwater tasks, often including more than one AUV. One of the main factors affecting the accuracy of AUVs navigation systems is the algorithm used to estimate the vehicle motion, usually based on kinematic vehicle models and linear estimators. In this paper, the authors present a navigation strategy specifically designed for AUVs, based on the Unscented Kalman Filter (UKF). The algorithm proves to be effective if applied to this class of vehicles and allows to achieve a satisfying accuracy improvement compared to standard navigation algorithms. The proposed strategy has been experimentally validated in suitable sea tests performed near the Cala Minnola wreck (Levanzo, Aegadian Islands, Sicily, Italy). The vehicles involved are the Typhoon AUVs, developed and built by the Department of Industrial Engineering of the University of Florence during the THESAURUS Tuscany Region project and the European ARROWS project for exploration and surveillance of underwater archaeological sites. The proposed algorithm has been implemented online on the AUVs and tested. The validation of the proposed strategy provided accurate results in estimating the vehicle dynamic behavior, better than those obtained through standard navigation algorithms.
Strategies of AUV navigation for achieving the near-bottom survey of a steep terrain are addressed. Unlike a flat terrain, a steep terrain can disrupt an altitude-based bottom-following dive of a cruising AUV. During its near-bottom dive over a steep terrain, a cruising AUV possibly experience longitudinal motion instability accompanied by severely fluctuating heave and pitch which may lead to bottom collision of the vehicle. Based on dive simulations, it is shown that the lost bottom lock of a sonar altimeter and the altitude overestimation are two major sources of the longitudinal motion instability. Underwater acoustic physics interacting with the sea water and bottom, as well as the coupled 3-d.o.f. longitudinal vehicle dynamics are incorporated into our simulation model. As alternative strategies for pursuing more reliable AUV-based near-bottom survey, the slope-following navigation and the pseudo bottom-following navigation have been investigated.
The paper addresses the computation of the information theoretic inspired empowerment quantity for a simplified vertical plane dynamic model of a Folaga autonomous underwater vehicle. Online empowerment computation can be exploited within complex autonomous robotics missions. In fact empowerment can be used to measure how much influence a vehicle has on its environment, and it identifies desirable states of the vehicle, hence it can act as an intrinsic cost function to use during a mission. In particular, the proposed approach is being developed in the framework of and H2020 underwater robotics research project (Widely scalable Mobile Underwater Sonar Technology) addressing the use of autonomous underwater vehicles for acoustic seismic applications.
This paper presents the initial stage of the development of an underwater localization system suitable for a flexible number of users. Multiple AUVs can work as a team and cooperate with other teams of AUVs without costly and even acoustically active components, which saves energy and allows AUVs to remain silent. The main building blocks for such a system are: spiral wavefront beacon in conjunction with a standard (circular) acoustic modem, Chip Scale Atomic Clocks (CSAC), acoustic modems, state-of-the-art adaptive underwater networking and Cooperative Localization (CL) algorithms. Using the difference in time of arrival between the spiral wavefront and the modem circular wavefront, receivers will be able to determine the bearing to the source using only one hydrophone. Synchronizing vehicles Chip Scale Atomic Clocks (CSAC) with the beacon at the beginning of the mission and during the longer missions will ensure the vehicles ability to also calculate their distance from the beacon upon every message reception.
Thursday, June 30th
This paper presents an autonomous navigation architecture for a robot using stereo vision-based localisation. The main contribution is the prediction of the quality of future localisation of the system in order to detect and avoid areas where vision-based localisation may fail, due to lack of texture in the scene. A criterion based on the estimation of future visible landmarks, considering uncertainties on landmarks and camera positions, is integrated in a Model Predictive Control loop to compute safe trajectories with respect to the visual localisation. The system was tested on a mobile robot and the obtained results demonstrate the effectiveness of our method.
Robust text detection and recognition on an arbitrarily distributed, unrestricted image areas is a difficult problem, e.g. when interpreting traffic panels outdoors during autonomous driving. Most previous work in Text detection only considers single script, usually Latin, and it is not able to detect text with multiple scripts. This contribution combines an established technique -Maximum Stable Extremal Regions- with a histogram of stroke width feature (HSW) and a Support Vector Machine classifier. We combined characters into groups by ray casting and merge aligned groups into lines of text that can also be verified by using the HSW. We evaluated our detection pipeline on our own dataset of Autobahn road scenes, and show how the character classifier stage can be trained with one script and be successfully tested on a different one. While precision and recall at least match to state of the art solution, a unique characteristic of the HSW feature is that it can learn and detect multiple scripts, which yields true script independence.
In this work we present a method for visual odometry that allows robust 3DoF trajectory estimation for wheeled robots using a downward facing RGBD-camera. Assuming that the robot moves on a ground plane while the environment itself can have arbitrary geometry allows us to estimate the frame to frame motion from orthographic projections of the RGBD- data. Instead of directly aligning these projections, the reference frame is split into blocks, which are individually registered using ESM, and thus create several estimates of the current motion. These estimates are combined using an outlier rejection scheme to create a robust estimate of the actual motion even under challenging conditions. We compare the results of our method to results of several state-of-the-art methods to show its accuracy and robustness.
In this paper, we present a system for autonomous object search and exploration in cluttered environments. The system shortens the average time needed to complete search tasks by continually planning multiple perception actions ahead of time using probabilistic prior knowledge. Useful sensing actions are found using a frontier-based view sampling technique in a continuously built 3D map. We demonstrate the system on real hardware, investigate the planner’s performance in three experiments in simulation, and show that our approach achieves shorter overall run times of search tasks compared to a greedy strategy.
This work proposes a full pipeline for a robot to autonomously explore, model and segment an apartment. Viewpoints are found offline and then visited by the robot to create a 3-D model of the environment. This model is segmented in order to find the various rooms and how they are linked (windows, doors, walls) yielding a topological map. Moreover areas of interest are also segmented, in this case furniture's planar surfaces. The method is validated on a realistic three rooms apartment. Results show that, despite occlusion, autonomous exploration and modelling covers 95\% of the appartment. For the segmentation part, 1 link out of 14 is wrongly classified while all the existing areas of interest are found.
In this paper we present an extension of Large Scale Kinect Fusion to compute optimized triangle meshes on-the-fly by removing redundant triangles on planar surfaces. The optimization is integrated into the reconstruction pipeline to compute the optimized meshes asynchronously to the reconstruction process in real time. The computed reconstructions can be extracted directly without the need of any post processing.
For about 80 years, people have been dreaming of cars that are able to drive by themselves. These days, this vision is starting to become reality. For the first time, cars found their way over a long distance in the Darpa Grand Challenge in 2005. Two years later, the famous Darpa Urban Challenge took place. In both events, all finalists based their systems on active sensors, and Google also started their impressive work with a high-end laser scanner accompanied by radars. In 2013, we let a new S-class vehicle with close to production cameras und radars drive itself from Mannheim to Pforzheim, following the route that Bertha Benz took 125 years ago. Many lessons have been learned from that experiment; above all that (computer) vision will become a key to autonomous driving, because we need a deep understanding of the scene if we want cars to drive safely in complex urban traffic. This can only be obtained through vision. In my talk, I will first summarize Bertha’s drive and the lessons that we learned from it. In the second part, I will present recent work towards achieving the necessary scene understanding. Based on modern CNNs, we are able to assign object labels to each pixel and to derive a rich but compact representation of the scene that serves as the basis of our work on autonomous cars. Within just a few years these new approaches radically changed the way we look at image understanding – more a revolution than an evolution.
Besides longitudinal control, advanced driving assistance functions additionally require lateral control. Up to now many different control approaches have been proposed and documented in literature dealing with lateral control. In contrast to this, there are only very few publications describing the handling of comfortable handover from manual to assisted driving controlling lateral car dynamics. The presented work aims to enhance handover at the activation of lateral control by significantly reducing lateral acceleration and jerk, resulting in smooth and comfortable control handover. First a state of the art model-based control is outlined. Based on this application a flatness-based approach for bumpless transfer from manual to controlled mode is proposed. Both, model-based control and bumpless transfer extension, allow straight forward tuning and execution on standard automotive hardware, due their low computational effort. Results are discussed on simulation results.
The knowledge of tire-ground interaction forces is interesting for intelligent vehicles. However, tire forces transducers are expensive and not suitable for ordinary passengers cars. An alternative is to estimate these forces using common sensors. This paper presents an estimator structure capable of reconstruct tire-ground interaction forces in all directions. A delayed interconnected cascade-observer structure is proposed to eliminate mutual dependence between estimators. Observers are developed based on nonlinear vehicle dynamic models with the Extended Kalman Filter algorithm. The estimator is validated with experimental results. The results are also compared with an existent estimator of the literature.
The availability of an efficient and reliable path planning strategy is a great benefit to mobile robots. Being able to intelligently connect a series of waypoints is a crucial requirement for the execution of autonomous navigation tasks. Several performance indices can be used to evaluate the goodness of a path, including its length and smoothness. In this paper, the authors focus on planar path planning for mobile robots; Bézier curves are employed, optimizing the computed path with respect to length and curvature, the latter used as a measure of smoothness. Due to the complexity of the objective function, optimization is performed by means of a direct search method. The proposed approach aims at generating paths offering advantages to mobile robots navigation in terms of controllability and reducing the related power consumption. The performed tests show that the presented method allows to achieve interesting results, suggesting its viability as a suitable path planning strategy for different kinds of mobile robots.
This paper documents the study on car-following behaviors under different environmental conditions and with different leading vehicle types on freeway. Unlike previous studies, in which filed data were collected at limited segments or simulation data were used, car-following time series covering large amount of freeway scenarios were extracted from the up-to-date Strategic Highway Research Program (SHRP2) Naturalistic Driving Study database. Differences and distributions of following headway and kinetic patterns were compared for different following conditions. New findings were concluded and compared with previous studies. The results can be used as references for microscopic traffic modeling and intelligent transportation systems.
Recent advances in autonomous driving and vehicle connectivity help to ensure safety and comfort in various driving conditions. These trends have widenened the system boundary conditions for hybrid powertrain operation with driving trajectory planning hence offering potential to improve powertrain operational efficiency. This paper presents an energy management (EM) controller for a plug-in hybrid vehicle exploiting predicted velocity trajectory together with its integration in both autonomous longitudinal guidance and driver-aware scenarios. The driver-aware scenario uses markov chain based stochastic modelling of driving characteristics. The proposed EM controller solves online, a discretized version of the fuel consumption minimization problem using direct methods transcribing the problem into a finite dimensional mixed boolean quadratic problem with polytopic constraints. Convex part of the resulting problem is solved using an active set method. Simulation results from different driving situations based on standard driving cycle and real world driving scenarios demonstrate the functionality of the controller and its flexbility to handle varying control objectives.
The article presents a tool which assesses the risk of automotive accidents taking into account the interactions between environment, driver and vehicle. The evaluated risk is composed of two parts: one concerns the impending risk (i.e. risk of a clearly identified danger and which is present in a short time horizon) and the other one, the latent risk (i.e. risky behavior of the driver which can lead to an accident). The developed tool uses information present in the CAN bus, additional sensors and car communication for shared sensing. With the collected information and estimated variables (e.g. grip and reaction time), it infers a probability of risk with a Bayesian Network. The tool can also be used for evaluating autonomous car driving and driver decisions.
This contribution proposes an application of an optimisation algorithm to be implemented in an intelligent drive assistant. A vehicle dynamic model is introduced and after the characterisation of its considered nonlinearities some elements of the background concerning "Dynamic Programming" are given together with the proposed algorithm. A two-point boundary value optimization problem in term of velocity is proposed with its solution using the well-known "Dynamic Programming" method. Because of the nonlinearity of the considered problem, using numerical methods seems to be the only way solving it. Using "Dynamic programming" such kind of a complex problem is solved by dividing it into a collection of simpler subproblems. At the end, an optimal feedback controller for the angular position of the throttle valve of the engine is proposed. Simulation results are discussed together with possible application aspects such as "Curse of Dimensionality" and an explicit analysis concerning the calculation effort.
This paper presents a compact and fast data association which can be used in tracking-by-detection based multiple pedestrian tracking approaches. The goal is to make the data association simple and robust so that it can be really useful in compact computing environment. This is realized by replacing computationally heavy Bayesian based filter with the compact Median Flow tracker. Two layers of data association are proposed for fulfilling different requirements of tracking tasks. Afterwards, we assess the performance of the algorithm by evaluating it on a standard dataset. Experimental results show that it improves the processing speed upon state-of-art methods tremendously with only trading off limited performance.
In order to improve the longitudinal control behavior of automated vehicles, a predictive control scheme with an adaptive vehicle state and parameter observer is proposed. The underlying nonlinear model of vehicle and powertrain dynamics makes use of the estimated torque signal which is calculated in the engine management system, as well as of vehicle speed and acceleration measurements. An Extended Kalman Filter is implemented to both estimate filtered vehicle states and the vehicle mass. Simulation results show good convergence of the parameter estimate. The contributions of this paper build the foundation to further examine the potential of improvement in fuel savings, planning accuracy and passenger comfort.
Collision avoidance systems demand sophisticated control algorithms to ensure driving on a safe preplanned path. Model predictive control (MPC) is suitable for this task as different influences can be considered by the algorithm. However MPC requires a model of the plant to guarantee good control characteristics. Especially the task to generate a simple steering model covering the considered range of vehicle dynamics is challenging. In this work a concept for an adaptive steering model is presented. The adaptive steering model is integrated in a model predictive control concept for collision avoidance maneuvers. Simulation results show the effectiveness for different maneuvers.
A nonlinear model predictive control (NMPC) approach for steering an autonomous mobile robot is presented. The vehicle dynamics with a counter steering system is described by a nonlinear bicycle model. The NMPC problem is formulated taking into account the obstacles description as inequality constraints which will be updated at each sampling time based on a laser scanner detection. The nonlinear optimal control problem (NOCP) is efficiently solved by a combined multiple-shooting and collocation method. Experimentation results illustrate the viability of our approach for active autonomous steering in avoiding spontaneous obstacles.
Recent investigations on the longitudinal and lateral
control of wheeled autonomous vehicles are reported. A
model-based design, which employs flatness-based techniques
is first introduced via a simplified model. It depends on
some physical parameters, like cornering stiffness
coefficients of the tires, friction coefficient of the
road, \dots, which are notoriously difficult to identify.
Then a model-free control strategy, which exploits the flat
outputs, is proposed. Those outputs also depend on physical
parameters which are poorly known, \textit{i.e.}, the
vehicle mass and inertia and the position of the center of
gravity. A totally model-free control law is therefore
adopted. It uses natural output variables, namely the
longitudinal velocity and the lateral deviation of the
vehicle. This last method, which is easily understandable
and implementable, ensures a robust trajectory tracking
problem in both longitudinal and lateral dynamics.
Friday, July 1st
We research on autonomous mobile robots with a seamless integration of perception, cognition, and action. In this talk, I will first introduce our CoBot service robots and their novel localization and symbiotic autonomy, which enable them to consistently move in our buildings, now for more than 1,000km. I will then introduce the CoBot robots as novel mobile data collectors of vital information of our buildings, and present their data representation, their active data gathering algorithm, and the particular use of the gathered WiFi data by CoBot. I will further present an overview of multiple human-robot interaction contributions, and detail the use and planning for language-based complex commands. I will then conclude with some philosophical and technical points on my view on the future of autonomous robots in our environments. The presented work is joint with my CORAL research group, and in particular refers to the past PhD theses of Joydeep Biswas, Stephanie Rosenthal, and Richard Wang, and recent work of Vittorio Perera.
Abstract—Simultaneous localization and mapping (SLAM) is vital for autonomous robot navigation. The robot must build a map of its environment while tracking its own motion through that map. There are many ways to approach the problem, mostly based on the sequential probabilistic approach, based around extended Kalman filter (EKF) or the Rao-Blackwellized particle filter. In order to improve the SLAM solution and to overcome some of the (EKF) and (PF) limitations, especially when the process and observation models contain uncertain parameters, we propose to use a robust approach to solve the SLAM problem based on variable structure theory. The new alternative called Smooth Variable Structure Filter (SVSF) is a predictor corrector estimator based on sliding mode control and estimation concepts. It has been demonstrated that the (SVSF) is stable and very robust face modeling uncertainties and noises. Visual SVSF-SLAM is implemented, validated and compared with EKF-SLAM filter. The comparison proofs the efficient and the robustness of localization and mapping using SVSF-SLAM.
We present a 3D mesh surface navigation system for mobile robots. This system uses a 3D point cloud to reconstruct a triangle mesh of the environment in real time that is enriched with a graph structure to represent local connectivity. This Navigation Mesh is then analyzed for roughness and trafficability and used for online path planning. The presented approach is evaluated with a VolksBot XT platform in a real life outdoor environment.
Automatic surface reconstruction from point cloud data is an active field of research in robotics, as polygonal representations are compact and geometrically precise. Standard meshing algorithms produce many redundant triangles. Therefore methods for optimization are required. In this paper we present and evaluate a mesh optimization algorithm for robotic applications that was specially designed to exploit the planar structure of typical indoor environments.
Topological maps have many applications in robotics. Matching two topological maps from the same environment can be used for map merging, place detection, map evaluation and other purposes. In this paper we present an approach to match two corresponding edges from two Topology Graphs to each other based on the actual path with which the vertices of the edges are connected in the underlying 2D grid maps. We perform experiments with two artificial maps as well as with four maps from the RoboCup Rescue WorldCup 2010.
This paper presents a large-scale 3D environment mapping solution for mobile robots that is based on hybrid metric-topological maps. If a robot performs simultaneous localization and mapping (SLAM) while exploring an unknown environment, sometimes loops are closed and the whole SLAM graph has to be optimized. If a conventional occupancy grid mapping algorithm using a monolithic map is used, the whole occupancy grid map has to be rebuilt after optimization, which for large maps can easily become too time consuming for real-time operation. In the same way, path planning on very large occupancy grid maps can become computationally too expensive. Hybrid metric-topological maps can solve both problems. The global metric map is divided into sub-maps and a global topological graph is formed on the map. This allows recomputing only isolated areas of the map, whereas others can remain unchanged. The topological graph allows efficient path planning on the hybrid map. We show that the hybrid mapping approach presented here is considerably faster than conventional methods. Within the same time, it can generate more detailed maps of large environments. The computation time for maps with identical level of detail can be improved by up to two orders of magnitude.
As robots leave the simple and static environments to more complex and dynamic ones, they will have to improve their localisation abilities and to deal with heterogeneous and imprecise data. In this paper, we present a general cooperative framework designed to localize in an absolute way a fleet of heterogeneous vehicles. Depending on the sensors it embeds, each vehicle localize itself using a GNSS system (typically GPS), an orientation system (a compass for instance), the detection of the others robots in the neighbourhood (typically with a LIDAR) and the detection of visible geo-referenced features in the map (eg. wall, poles, etc...). These map features are often imprecise (as is typically the case with collaborative public maps such as OpenStreetMap). Our approach allows to update these features positions in the same framework. We first present the filtering approach we developed to solve the classical over-convergence problem using the SCI (Split Covariance Intersection) filter. Map feature relative detection being simultaneously the main information source as well as compute-time expensive, we show how in the same framework we optimize resource usage thanks to an entropy optimization strategy which avoids all sensor data fusion and instead selects the best one at each time step.
This paper applies Posterior Cramer-Rao Bound theory to the SLAM problem to measure the information supplied by different sensor modalities over time. Range-only, bearing-only and full range-bearing sensors were considered, as well as the gains in information achieved by using multiple sensors in centralised co-operative SLAM. An efficient recursive formula was used to compute the bound for a variety of simulated scenarios, and its validity verified by comparing the bound with the second-order error performance of FastSLAM 2.0 and the EKF.
In nuclear facilities, such as the experimental fusion reactor ITER, the cargo transfer operations are performed by autonomous guided vehicles under remote supervision. In ITER, these vehicles can reach up to 100 tons and, with a rhombic-like configuration, have to move in cluttered scenarios. In case of failure, the vehicles have to be manually guided. This paper presents three solutions for the teleoperation of rhombic-like vehicles. Two set of devices were used to test each solution: one is based on a gamepad and the other is based on a joystick with a rotational disc specially designed for this purpose. The solutions were experimented by the developer and by 12 users without prior experience on rhombic-like vehicles. The experiments were performed in a software simulator that provides 2D maps of the test facility and simulates the kinematics of the vehicles in real time. The main conclusions are reported.
Carsharing programs provide an alternative to private vehicle ownership. Combining carsharing programs with autonomous vehicles would improve user access to vehicles thereby removing one of the main challenges these programs face for widescale adoption. While the ability to easily move cars to meet demand would be significant for carsharing programs, if implemented incorrectly it could lead to worse system performance. In this paper, we seek to improve the performance of a fleet of shared autonomous vehicles through improved matching of vehicles to passengers requesting rides. We consider carsharing with autonomous vehicles as an assignment problem and examine four different methods for matching cars to users in a dynamic setting. We show how applying a recent algorithm (SCRAM) for minimizing the maximal edge in a perfect matching can result in a more efficient, reliable, and fair carsharing system. Our results highlight some of the problems with greedy or decentralized approaches. Introducing a centralized system creates the possibility for users to strategically mis-report their locations and improve their expected wait time so we provide a proof demonstrating that cancellation fees can be applied to eliminate the incentive to mis-report location.
Road roughness detection on the highway networks is considered to be vital in order to ensure driving comfort and safety. High-speed profilometers are developed in order to be able to monitor road roughness on the highways without affecting the normal traffic flow. This paper focuses on the improvement and the implementation of TRRL-type high-speed laser profilometers. In the presented work, the original TRRL design is made more compact in order to allow faster operating speeds. The error analysis seen in the original work is extended with the changed geometric design. It is observed that the factor that affects the measurement accuracy the most is surface texture due to its randomness. Therefore the paper proposes a new method which eliminates the texture-caused errors by modelling them with quadratic functions. The performance of the presented profilometer is evaluated by conducting experiments on a road with a known true profile. The accuracy and the repeatability of the system show that the presented profilometer can be used for measuring true profiles with some further improvement.
Mobile manipulators are intended to be deployed in domestic and industrial environments where they will carry out tasks that require physical interaction with the surrounding world, for example, picking or handing over fragile objects. In-hand slippage, i.e. a grasped object moving within the robot's grasp, is inherent to many of these tasks and thus, a robot's ability to detect a slippage is vital for executing a manipulation task successfully. In this paper, we develop a slip detection approach which is based on the robot's tactile sensors, a force/torque sensor and a combination thereof. The evaluation of our approach, carried out on the Care-O-bot 3 platform, highly suggests that the actions and motions performed by the robot during grasping should be taken into account during slip detection for improved performance. Based on this insight, we propose an in-hand slip detection architecture that is able to adapt to the current robot's actions at run time.
Through its Technical Committees the IFAC spans the entire breadth and depth of both theory and application of control systems and system identification. The Technical Committees are grouped into 9 different clusters. CC1 and CC2 are mainly focused on theory whilst the others CC3 to CC9 derive their existence from a technology domain. CC1 is in the first instance about the analysis of signals and systems, CC2 considers the generic issues of synthesis. Each of the clusters is home to a number of specific Technical Committees.
Through the Technical Committees the IFAC harnesses and guides the volunteering effort of active researchers and practitioners in the broad field of control and automation. Over 2000 volunteers support the various IFAC activities. Each of the TCs help organize workshops and conferences, and in particular support the triennial IFAC World Congress.
The 9th IFAC Symposium on Intelligent Autonomous Vehicles is partially sponsored by Bonn Rhein-Sieg University of Applied Sciences and the IEEE Robotics and Automation Society.
Contact IAV Organiser
Bonn-Rhein-Sieg University of Applied Sciences
M.Sc. Anastassia Küstenmacher
Grantham-Allee 20
Phone: +49 2241 865 743
Email: anastassia.kuestenmacher@h-brs.de
2015 © IAV 2016 - 9th Symposium on Intelligent Autonomous Vehicles, All Rights Reserved. Picture of Leipzig's cityscape © Michael Bader, Picture of Leipziger Messe © Leipziger Messe/Grubitzsch Impressum.