IROS 2016 Papers

The quality of holes has important influence on the mechanical strength, assembly quality and life of aircraft, and inferior quality holes may even cause airplane crash and casualties. The perpendicular accuracy of holes during drilling is crucial one of all the factors influencing the quality of holes. In order to improve the perpendicular accuracy during drilling, this paper presented an attitude adjustment mechanism and attitude adjustment algorithm for the end effector. For the mechanism, it adjusts the drill bit attitude by using double eccentric discs and a spherical pair. After the surface normal vector at drilling point is obtained, double eccentric discs will rotate define angels calculated by using attitude adjustment algorithm based on the shortest time principle to achieve the coincidence of the surface normal vector and the drill bit axis vector if the angle between the normal vector and drill bit axis vector is greater than 0.5°. To demonstrate the method mentioned above, simulation and experiment are carried out on the aeronautical drilling robot. The results show that the attitude adjustment mechanism and attitude adjustment algorithm can increase the perpendicular accuracy of holes during drilling and satisfy the demand in automatic aircraft assembly.

We present an omnidirectional ball wheel drive design that utilizes a permanent magnet as the drive roller to generate the contact force. Particularly interesting for novel human-mobile robot interaction scenarios where the users are expected to physically interact with many palm-sized robots, our design combines simplicity, low cost and compactness. We first detail our design and explain its key parameters. Then, we present our implementation and compare it with an omniwheel drive built with identical conditions and similar cost. Finally, we elaborate on the main advantages and drawbacks of our design.

This work presents a comparison of the five observability indices used for robot calibration. The comparison is realized in order to determine the most appropriate observability index, which allows for the best parameter identification of a calibrated robot, and therefore leading to the best improvement of the robot accuracy. In this study, the accuracy analysis is based on the robot end-effector errors, which are expressed in term of Euclidean errors. The parameter identification process is based on minimizing the residual of the position errors. The actual values of these positions are usually measured by an external measurement device and have measurement noise. The position residuals are calculated in all the calibration configurations, which are selected by using observability indices. An optimal set of configurations is the one reducing the impact of the measurement noise on the parameter identification efficacy. Our study is carried out for the calibration of four robots: two degrees of freedom (DOF) and 6-DOF serial robots, and 2-DOF and 3-DOF planar parallel robots. The comparison of the observability indices was achieved through a Monte Carlo simulation, using 100 different cases for each of the four robots considered. The position measurement noise was assumed to be within a range of ± 200 μm. Investigations led to conclude that there is a specific index that may be considered the best observability index for robot calibration. Finally, an experimental study has been applied to a LR Mate 200ic FANUC robot and confirms the simulated results.

Despite the existence of powerful formal languages for writing robot controllers, most existing functional modules are written using standard programming languages. The existence of such a code base raises critical challenges: 1. How to enable automated analysis, monitoring, and reuse of existing code given that reasoning directly about code fragments is impractical. 2. How to convey to users the expected level of performance of an autonomous robot? 3. Perhaps most crucial: how to quickly identify abnormal behavior of autonomous robots? This is a key impediment to the deployment of such platforms in open environments. We address these issues through the use of performance-level profiles (PLPs), a formal, yet intuitive, language for specifying the expected properties of functional modules, designed with the above aims in mind. PLPs are motivated by action specification languages, such as PDDL 2.1, but add novel elements important for robotic applications, such as update frequency, run-time statistics, progress measures, and trigger conditions, and take into account the different roles modules can play. PLPs have been used to support monitoring in two projects: an autonomous compact track loader, and a service robot. Additionally, we developed a number of tools for automated monitoring-code generation from PLPs.

Classical control approaches to robot force control have been extensively addressed by research in the last decades and are now considered a paradigm when dealing with force control for industrial robots. With this respect, the present paper exploits the capability of state-of-the-art Quadratic Programming (QP) solvers to specify a simple and intuitive constraint-based optimization strategy aiming at improving closed-loop performance of a classical force controller, such as the implicit force control with pure integral action for a position-controlled manipulator in contact with a compliant environment. The effectiveness of the proposed control strategy is experimentally validated on an industrial robot equipped with a force sensor.

No quantitative procedure currently exists to evaluate the obstacle avoidance capabilities of robotic systems. Such an evaluation method is not only needed to compare different avoidance methods, but also to determine the operational limits of autonomous systems. This work proposes an evaluation framework which can find such limits. The framework comprises two types of tests: detection tests and avoidance tests. For each type, both environment and performance metrics need to be defined. For detection tests such metrics are well known, but for avoidance tests such metrics are not readily available. Therefore a new set of metrics is proposed. The framework is applied to a UAV that uses stereo vision to detect obstacles. Three different avoidance methods are compared in environments of varying difficulty.

This paper presents initial results of a set of experiments to compare time-delayed teleoperation performance of selected architectures based on two popular passivity based methods, namely, Wave Variables Transformation (WVT) and Time Domain Passivity Approach (TDPA). Five different architectures widely used and available in literature are selected: 2-channel and 3-channel for both WV and TDPA, and 4-channel based on TDPA. An empirical approach for measuring transparency is proposed, which presents three main qualities: 1) Controller independency 2) Frequency dependency of performance and 3) Removal of the human operator out of the loop. In order to achieve this, the human operator is replaced by a linear actuator equipped with a force sensor. Deterministic interactions between the simulated operator and the rest of the system can thus be easily implemented and registered. Therefore, this approach allows to isolate pure effects of any control method and perform a proper analysis in terms of selected attributes for a desired metrics. In this article, these are chosen to be effective impedance and position tracking error, both in the frequency domain. In addition to the quantitative results, a qualitative analysis is presented which underlines practical implications of each architecture and control method. The comparison sample presented contains five different delays, varying from 0 to 200 milliseconds, five different environmental impedances and 5 bilateral control architectures along with an analysis of the results obtained.

In this work we address the problem of trajectory planning in Graph SLAM. We propose the use of Expected Value of the Final Uncertainty, which summarizes all the possible uncertainties that should be considered. In fully explored environments, this is used to determine the most reliable path to the final position. In partially explored environments, this criteria quantifies the reliability of the path planned in the free space. Tests demonstrate its ability to avoid unreliable paths in fully explored environments as compared to other uncertainty based criteria. In the exploration scenario, potential paths not present in the original graph are proposed using Voronoi Diagram of the space ahead observed by the sensors. A cost function is proposed considering the length of the path as well as the expected final uncertainty, thus including potentially shorter, but still reliable paths. Tests demonstrate the shortest path is preferred as long as it contains loop closures with low uncertainty.

This paper presents a control algorithm that steers a robotic walking assistant along a planned path using electromechanical brakes. The device is modeled as a Dubins' car, i.e., a wheeled vehicle that moves only forward in the plane and has a limited turning radius. In order to reduce the cost of the hardware, no force sensor is employed. This feature hampers the application of control algorithms based on a modulated braking action. A viable solution is based on the application of on/off braking action, thus forcing the vehicle to turn with a fixed turning radius. In order to avoid the annoying chattering behaviour, which is the inevitable companion of all bang-bang solutions, we propose a hybrid controller based on three discrete states that rule the application of the braking action. The resulting feedback controller secures a gentle convergence of the user toward the planned path and his/her steady progress towards the destination. This is obtained by using two independent hystereses thresholds, the first one associated with the approaching phase and the second with the following phases. The system convergence toward the path is formally proved. Simulations and experiments show the effectiveness of the proposed approach and the good level of comfort for the user.

This paper addresses the design of a partitioned vision-guided scheme for repetitive optical biopsies. More precisely, our approach uses two image modalities to perform 6 degrees of freedom (DOF) positioning task. The development aims to partition the control into 3 DOF controlled by the B-scan images acquired with an optical coherence tomography (OCT) system and the remaining 3 DOF controlled by the white light images provided by a CCD camera. Moreover, for the control and instead of conventional visual features (e.g., points, lines, moments, etc.) extracted using algorithms combined with visual tracking approaches, our visual servoing method uses the multiresolution wavelet coefficients. The developed method was experimentally validated using a parallel kinematic structure equipped with a Telesto-II OCT benchtop. The validation task consisted of an automatic spatial repositioning of the robotic structure to precisely retrieve the position of an initial optical biopsy. Several tests are achieved, which clearly demonstrate the reliability of the proposed controller.

With respect to classical visual servoing (VS) technics based on geometrical features, the main drawback of direct visual servoing is its limited convergence area. In this paper we propose a new direct visual servoing control law that relies on a particle filter to achieve non-local and non-linear optimization in order to increase this convergence area. Thanks to multi-view geometry and image transfer techniques, a set of particles (which correspond to potential camera velocities) are drawn and evaluated in order to evaluate the best camera trajectory. This new control law is validated on a 6 DOF positioning task performed on a real gantry robot and statistical comparisons are also provided from simulation results.

We provide a framework for 6 DOF pose estimation in seam-tracking applications using particle filtering. The particle filter algorithm developed incorporates measurements from both a 2 DOF laser seam tracker and the robot forward kinematics under an assumed external force. Special attention is paid to modeling of disturbances in the respective measurements, and methods are developed to assist the selection of sensor configurations for optimal estimation performance. The developed estimation algorithm and simulation environment are provided as an open-source, extendable package, written with an intended balance between readability and performance.

This paper addresses sound source separation and identification for noise-contaminated acoustic signals recorded with a microphone array embedded in an Unmanned Aerial Vehicle (UAV), aiming at people's voice detection quickly and widely in a disaster situation. The key approach to achieve this is Deep Neural Network (DNN), but it is well known that training a DNN needs a huge dataset to improve its performance. In a practical application, building such a dataset is not often realistic owing to the cost of manual data annotation. Therefore, we propose a Partially-Shared Deep Neural Network (PS-DNN) which can learn multiple tasks at the same time with a small amount of annotated data. Preliminary results show that the PS-DNN outperforms conventional DNN-based approaches which require fully-annotated data in training in terms of identification accuracy. In addition, it maintains performance even when noise-suppressed signals are used for sound source separation training, and partially annotated data is used for sound source identification training.

We present POMP (Pareto Optimal Motion Planner), an anytime algorithm for geometric path planning on roadmaps. For robots with several degrees of freedom, collision checks are computationally expensive and often dominate planning time. Our goal is to minimize the number of collision checks for obtaining the first feasible path and successively shorter feasible paths. We assume that the roadmaps we search over are embedded in a continuous ambient space, where nearby points tend to share the same collision state. This enables us to formulate a probabilistic model that computes the probability of unevaluated configurations being collision-free. We update the model over time as more checks are performed. This model lets us define a weighting function for roadmap edges that is related to the probability of the edge being in collision. Our approach is to trade off between these two weights, gradually prioritizing edge length over collision likelihood. We also show that this tradeoff is approximately equivalent to minimizing the expected path length, with a penalty of being in collision. Our experiments demonstrate that POMP performs comparably with RRTConnect and LazyPRM for the first feasible path, and BIT* for anytime performance, both in terms of collision checks and total planning time.

Binary descriptors allow fast detection and matching algorithms in computer vision problems. Though binary descriptors can be computed at almost two orders of magnitude faster than traditional gradient based descriptors, they suffer from poor matching accuracy in challenging conditions. In this paper we propose three improvements for binary descriptors in their computation and matching that enhance their performance in comparison to traditional binary and non-binary descriptors without compromising their speed. This is achieved by learning some weights and threshold parameters that allow customized matching under some variations such as lighting and viewpoint. Our suggested improvements can be easily applied to any binary descriptor. We demonstrate our approach on the ORB (Oriented FAST and Rotated BRIEF) descriptor and compare its performance with the traditional ORB and SIFT descriptors on a wide variety of datasets. In all instances, our enhancements outperform standard ORB and are comparable to SIFT.

Most approaches to visual odometry estimates the camera motion based on point features, consequently, their performance deteriorates in low-textured scenes where it is difficult to find a reliable set of them. This paper extends a popular semi-direct approach to monocular visual odometry known as SVO [1] to work with line segments, hence obtaining a more robust system capable of dealing with both textured and structured environments. The proposed odometry system allows for the fast tracking of line segments since it eliminates the necessity of continuously extracting and matching features between subsequent frames. The method, of course, has a higher computational burden than the original SVO, but it still runs with frequencies of 60Hz on a personal computer while performing robustly in a wider variety of scenarios.

Human motion is fast and hard to predict. To implement a provably safe collision-avoidance strategy for robots in collaborative spaces with humans, an overapproximative prediction of the occupancy of the human is required, which needs to be calculated faster than real time. We present a method for computing volumes containing the entire possible future occupancy of the human, given its state, faster than real time. The dynamic model of the human is built from analysing a set of archetypal movements performed by test subjects. The occupancy prediction is tested on a publicly available database of motion capture data, and shown to be overapproximative for all movements relating to everyday activities, sport and dance. Our novel algorithm is useful to guarantee safety in human-robot collaboration scenarios.

Because of their simplicity, wave variables have become almost a standard strategy for stabilizing delayed bilateral teleoperation systems. However, the price to pay for a stable behavior is a degradation in the performance of the teleoperation system. Recently, more flexible and transparency oriented bilateral architectures have been proposed (e.g. TDPN, PSPM, Two-Layer approach) but they are complex to implement and to tune. In [1], a strategy for blending the high performance of the new control methodologies with the simplicity of wave based bilateral teleoperation has been proposed. Nevertheless, while appealing in terms of simplicity, this method is conservative in terms of the transparency that can be achieved. In this paper, we extend the architecture in [1] in order to optimize the use of the energy and for achieving a coupling that is as close as possible to the desired one while preserving the passivity of the overall system.

The planar and energetically conservative Spring-Loaded Inverted Pendulum (SLIP) model with a linear spring has been modified in recent years to include an articulated knee, friction and contact losses, and energy thrusts during stance to stabilize forward velocity and hopping height. The work presented here advances the SLIP template toward a biological anchor by adding a third parallel articulating joint, which provides redundancy in the inverse kinematics solution space. Instead of one actuator into which an energy thrust can be added, the total thrust can be distributed between two actuators operating on the lower two leg joints (the hip joint remains passive during stance as in the original SLIP model). It is shown that a careless distribution of the thrust can lead to significant energy loss, so an offline search protocol is presented as a general approach for resolving the optimal thrust distribution ratio between these joints. The results of the offline search are then approximated for rapid computation of this ratio during hopping tests in a high-fidelity simulation engine.