IROS 2017 论文列表

Physical human-UAV interaction (PHUI) with Unmanned Aerial Vehicles (UAV) has many possible applications, and a few recent works [1], [2] have shown that it is possible to perform PHUI with a standard quadrotor. However, the intrinsic underactuation of quadrotors may hinder the interaction task and also cause safety issues. In this paper, we present an admittance control-based scheme to perform PHUI with a fully actuated UAV. The system also benefits from the robustness provided by our fully actuated UAV implementation of the adaptive super twisting controller (ASTC). While validating our system in simulation, we also show the superior performance of the fully actuated UAV with respect to a standard quadcopter.

Robotic coverage path planning describes the problem of determining a configuration space trajectory for successively covering a specified workspace target area with the robot's end-effector. Performing coverage path planning for mobile robots further requires solving the problem of robot placement, i.e. determining of suitable robot base positions to perform the task. Finding an optimal solution is hard as both problems cannot be solved independently. Combined robot placement and coverage planning is particularly interesting if repositioning of the robot is costly or if simultaneous repositioning and end-effector motion is not desired. In this paper, we present a general approach for combined robot placement and coverage path planning that takes constraints like collision avoidance and static stability into account. In contrast to related approaches, we focus on mobile manipulation tasks that require a fixed placement for executing coverage trajectory segments. The approach is evaluated in two scenarios that exemplify the broad range of possible applications: The coverage of a building facade using a robotic manlift and the coverage of an industrial conveyer belt for maintenance tasks using the humanoid robot ARMAR-III.

THIS paper presents the development of a bidirectional fabric-based soft pneumatic actuator requiring low fluid pressurization for actuation, which is incorporated into a soft robotic gripper to demonstrate its utility. The bidirectional soft fabric-based actuator is able to provide both flexion and extension. Fabrication of the fabric actuators is simple as compared to the steps involved in traditional silicone-based approach. In addition, the fabric actuators are able to generate comparably larger vertical grip resistive force at lower operating pressure than elastomeric actuators and 3D-printed actuators, being able to generate resistive grip force up to 20N at 120 kPa. Five of the bidirectional soft fabric-based actuators are deployed within a five-fingered soft robotic gripper, complete with five casings and a base. It is capable of grasping a variety of objects with maximum width or diameter closer to its bending curvature. A cutting task involved bimanual manipulation was demonstrated successfully with the gripper. To incorporate intelligent control for such a task, a soft force made completely of compliant material was attached to the gripper, which allows determination of whether the cutting task is completed. To the authors' knowledge, this work is the first study which incorporates two soft robotic grippers for bimanual manipulation with one of the grippers sensorized to provide closed loop control.

Finding the source of gaseous compounds released in the air with robots finds several applications in various critical situations, such as search and rescue. While the distribution of gas in the air is inherently a 3D phenomenon, most of the previous works have downgraded the problem into 2D search, using only ground robots. In this paper, we have designed a bio-inspired 3D algorithm involving cross-wind Lévy Walk, spiralling and upwind surge. The algorithm has been validated using high-fidelity simulations, and evaluated in a wind tunnel which represents a realistic controlled environment, under different conditions in terms of wind speed, source release rates and odor threshold. Studying success rate and execution time, the results show that the proposed method outperforms its 2D counterpart and is robust to the various setup conditions, especially to the source release rate and the odor threshold.

This paper presents a 2-point algorithm for relative pose estimation of a monocular camera mounted on a ground vehicle. Specifically, the geometric model is developed by combining two-view geometry with planar motion constraint. Based on the proposed model, the relative camera pose information can be estimated efficiently with at least 2 feature correspondences. Besides, the proposed algorithm can recover rotation independently with translation so that pure rotation can be computed stably under planar motion. Both experiments on synthetic data and in real world are presented. In the synthetic data experiments, the proposed algorithm is compared with other well-known algorithms to illustrate its high accuracy and robustness to image noise. In addition, the great robustness of the proposed algorithm to nonplanar motion is demonstrated as well. Furthermore, experiment are carried out based on KITTI Vision Benchmark, and the result indicates that the proposed algorithm has better performance than 5-point algorithm for monocular visual odometry and has a wide range of applicability.

This paper presents a COM trajectory generation method for 3D walking and skating motion by nonlinear optimization. In our method, we solve the following problems: (1) dealing with both walking and skating motion in the same framework, (2) generating center of mass (COM) trajectory faster than execution time, (3) executing motion with large acceleration. For solving (1) and (2), we calculate the COM trajectory at every step and introduce frictional constraints to the Divergent Component of Motion as terminal conditions. By changing the terminal condition, we can generate both skating and walking motion. Besides, the nonlinear constrained optimization using Gauss Pseudospectral Method is introduced for solving (2) and (3). Thanks to this method, we generate the 3D COM trajectory considering contact constraints and kinematic constraints faster than execution time. Finally, the walking and skating experiment were carried out to confirm the effectiveness of our method using life-sized humanoid HRP-2. Applying the proposed method, HRP-2 could successfully walk at 0.4 [m/s] and skate at 1.0 [m/s].

Suturing is an essential requirement for surgical robots. Current designs of needle drivers require significant forces to be applied so that the needle will not flip or slip between the jaws when inserted into the tissue. The required force implies that most designs are based on rigid instruments with a straight shaft that induces less dexterity. Flexible robotics provides great dexterity to achieve the task, although often it lacks the capability to apply large forces and therefore hold the surgical needle firmly. Solutions can be found in how the needle driver's jaws are designed. This paper presents improved designs for needle grasping tools for flexible robotic instruments, to be produced directly with selective laser melting. Features on the gripping surface were embedded to increase the gripping effectiveness even in the presence of limited grasping force. A detailed characterisation has been performed for each design showing improvements with respect to the state of the art.

In this paper, we address the problem of 3D object instance recognition and pose estimation of localized objects in cluttered environments using convolutional neural networks. Inspired by the descriptor learning approach of Wohlhart et al. [1], we propose a method that introduces the dynamic margin in the manifold learning triplet loss function. Such a loss function is designed to map images of different objects under different poses to a lower-dimensional, similarity-preserving descriptor space on which efficient nearest neighbor search algorithms can be applied. Introducing the dynamic margin allows for faster training times and better accuracy of the resulting low-dimensional manifolds. Furthermore, we contribute the following: adding in-plane rotations (ignored by the baseline method) to the training, proposing new background noise types that help to better mimic realistic scenarios and improve accuracy with respect to clutter, adding surface normals as another powerful image modality representing an object surface leading to better performance than merely depth, and finally implementing an efficient online batch generation that allows for better variability during the training phase. We perform an exhaustive evaluation to demonstrate the effects of our contributions. Additionally, we assess the performance of the algorithm on the large BigBIRD dataset [2] to demonstrate good scalability properties of the pipeline with respect to the number of models.

Recognizing objects from the point cloud captured by modern 3D sensors is an important task for robots operating autonomously in real-world environments. However, the existing well-performing approaches typically suffer from a trade-off between resolution of representation and computational efficiency. In this paper, raw point cloud normals are fed into the Point Convolution Network (PCN) without any other representation converts. The point cloud set disordered and unstructured problems are tackled by Kd-tree-based local permutation and spatial commutative pooling strategies proposed in this paper. Experiments on ModelNet illustrate that our method has two orders of magnitude less floating point computation in each non-linear mapping layer while it contributes to significant classification accuracy improvement. Compared to some of the state-of-the-art methods using the 3D volumetric image convolution, the PCN method also yields comparable classification accuracy.

This paper presents a novel approach to build consistent 3D maps for multi robot cooperation in USAR environments. The sensor streams from unmanned aerial vehicles (UAVs) and ground robots (UGV) are fused in one consistent map. The UAV camera data are used to generate 3D point clouds that are fused with the 3D point clouds generated by a rolling 2D laser scanner at the UGV. The registration method is based on the matching of corresponding planar segments that are extracted from the point clouds. Based on the registration, an approach for a globally optimized localization is presented. Apart from the structural information of the point clouds, it is important to mention that no further information is required for the localization. Two examples show the performance of the overall registration.

2D fully convolutional network has been recently successfully applied to the object detection problem on images. In this paper, we extend the fully convolutional network based detection techniques to 3D and apply it to point cloud data. The proposed approach is verified on the task of vehicle detection from lidar point cloud for autonomous driving. Experiments on the KITTI dataset shows significant performance improvement over the previous point cloud based detection approaches.

Previously, we developed a 3D path following algorithm to improve the microrobot performances to overcome the modeling errors and environmental disturbances in order to perform real tasks, tested experimentally in a scaled-up helical microswimmer. In this paper, we show that adapting the propulsion mode, the general path following algorithm can be used for any microswimmer behaving as a nonholonomic system. For that purpose, we study a magnetic robot with a flexible tail that mimics the spermatozoa locomotion mechanism. A frequency and amplitude characterization of the flexible swimmer using an oscillating magnetic field is shown. To adapt the propulsion mode, we develop a 3D magnetic field control based on the steering angular velocities which are computed from the path following algorithm in order to propel and steer the flexible robot to reach a desired location in space and achieve for the first time a 3D closed-loop motion control using a swimmer with flexible flagella.

This paper establishes the feasibility of robotically 3D printing biomaterials such as alginate hydrogels onto moving human anatomy and a stationary plane. The alginate hydrogels used are in-vivo compatible and a proven biomaterial for tissue scaffolds. We developed a control scheme for precision material deposition via piezo microjetting while tracking in real-time to continuously sense anatomy location and deposits material in a predefined trajectory derived from two pre-selected target geometries. We show that multilayer 3D structures can be created on a moving human hand with 1.6 mm average error and 87.8 % overall accuracy.

This paper reports an ultra-fast manipulation on micro-scale objects in a microfluidic channel. The objects are manipulated in simple harmonic motions and the manipulation speed has been achieved up to 3000 Hz. While the maximum speed was only about 130 Hz in previous works, the speed is advanced with an additional order in this work. Based on the experimental results, we find that the gain decreases along the line of -20 dB/decade in a logarithmic plot. The transfer function based on a theoretical model was employed for interpreting the results. The experimental gain and theoretical gain were well matched. Finally, experiments with red blood cells were performed and it supports the feasibility of applying such ultra-fast manipulation to cell evaluations.

This paper addresses an omnidirectional locomotion control to make a biped robot walk back and forth, turn to any direction, move sideways and carry out them seamlessly. This is achieved by integrating smooth-path-tracking and lateral walking controllers that have been proposed by the authors, which is not straightforward because the path tracking is designed to converge to a referential path while the lateral walking aims to go away from the path. It can be avoided by extending the path-tracking controller to consecutively redraw the orbit that passes the middle of the feet. The proposed controller is designed completely in the robot-centric frame, namely, all the references and the manipulated variables are represented from the robot's viewpoint so that the operator can issue the referential values in the first-person view. Additionally, as the robot is ruled by neither a plan of footsteps nor time-dependent trajectories, it can flexibly respond to various perturbations. The validity of the proposed controller is investigated through computer simulations and an application to navigation of the robot is also shown.

This paper deals with the open-loop characterization of a micromanipulation system actuated by thermocapillary convective flows. Micrometric size objects placed at the air/liquid interface are actuated by heating the surface of the liquid using a laser. The heat generates a surface tension gradient at the interface which induces thermocapillary convective flows that are used to move the objects. In this paper, the performances of this approach are analyzed based on open-loop experiments. Several actuation strategies are proposed and discussed. The experimental results highlight the potential of this approach since velocities up to several millimeters per second are obtained. However the precision of the positioning is not ensured by open-loop actuation, so closed-loop control will be necessary in future works. As a first step towards closed-loop control, this paper proposes a model of the system. This model is based on the open-loop experimental results, but the proposed methodology can be applied to any setup that use thermocapillary convective flows for particle manipulation.

This paper describes a high-frame-rate (HFR) vision system that can detect multiple objects in an image of 512 × 512 pixels at 12,000 frames per seconds (fps). An optimized algorithm is proposed based on conventional Histograms of Oriented Gradient (HOG) descriptor and Support Vector Machine (SVM) classifier algorithms for hardware implementation. By implementing the proposed algorithm on a field-programmable gate array (FPGA) of a high-speed vision platform, multi-object in an image can be detected at 12,000 fps under complex background. In hardware implementation, 64 pixels were processed in parallel with 80 MHz camera clock. Source image and detection results can be transferred to personal computer (PC) in real-time for recording or post-processing. Our developed HFR multi-object detection system was verified by performing several evaluations.