IROS 2015 Papers

This paper gives an analytical description for the ideal and the real workspace of a 6-RSS parallel robot; that is, by considering its geometrical and its mechanical constraints. The ideal workspace is described as a set of inequalities representing a 3D volume that encloses valid attitudes, whose frontier can be plotted as a 3D surface by using Constructive Solid Geometry and an adaptive polygonization method. The real workspace represents a subset of the ideal one, determined by the presence of joint limits at the spherical joints. These ones arise from the fact that these spherical joints are not ideal, but implemented as spherical plain bearings. However, although it is not straight forward to express these joint limits in an analytical way, it was possible to find an analytical expression to describe the real workspace, represented by additional inequalities that must be met by a desired attitude of the platform.

Despite having achieved unconstrained stable flight, the insect-scale flapping-wing robot is still tethered for power and control. Towards the goal of operating a biologically-inspired robot autonomously outside of laboratory conditions. In this paper, we simulate outdoor disturbances in the laboratory setting and investigate the effects of wind gusts on the flight dynamics of a millimeter-scale flapping wing robot. Simplified models describing the disturbance effects on the robot's dynamics are proposed, together with two disturbance rejection schemes capable of estimating and compensating for the disturbances. The proposed methods are experimentally verified. The results show that they reduced the root mean square position errors by approximately 50% when the robot was subject to 60 cm·s−1 horizontal wind.

Controlling the robot with a permanently-updated optimal trajectory, also known as model predictive control, is the Holy Grail of whole-body motion generation. Before obtaining it, several challenges should be faced: computation cost, non-linear local minima, algorithm stability, etc. In this paper, we address the problem of applying the updated optimal control in real-time on the physical robot. In particular, we focus on the problems raised by the delays due to computation and by the differences between the real robot and the simulated model. Based on the optimal-control solver MuJoCo, we implemented a complete model-predictive controller and we applied it in real-time on the physical HRP-2 robot. It is the first time that such a whole-body model predictive controller is applied in real-time on a complex dynamic robot. Aside from the technical contributions cited above, the main contribution of this paper is to report the experimental results of this première implementation.

Whole-body holding manipulation is effective for carrying the handleless large object. In order to keep the object stability, the dexterous transition motion is necessary. From geometric and physical conditions of object manipulation, we propose the general method of generating the transition graph, which represents the object pose and grasp contact. By searching the path on the graph, the transition motion is planned automatically with considering the object motion and contact switching simultaneously. By generating and modifying the whole-body holding motion, the planned object motion is achieved stably. We show the effectiveness of the proposed method by the experiments, in which robot lifts up a large object with whole-body contact by the planned transition motion.

Gecko-inspired structured adhesives will be valuable for novel climbing and space robots. Robots also provide useful evaluation platforms for these adhesives. Climbing robots need to be lightweight, and thus many designs use multiple feet on a single rotating wheel-leg. Generally, such designs have not been able to walk robustly on steeper than vertical substrates. In this work, we use an improved version of our previous Mushroom-Shaped Adhesive MicroStructured (MSAMS) tape to support a power-autonomous robot reliably walking inverted on glass ceilings. The resulting speeds are greater than one body/length per second, faster than other adhesion-based climbing prototypes. The printed robot design is also a contribution toward future robotic designs and will have future applications in testing new adhesives for robotic feet.

Robust object recognition is a crucial skill for robots operating autonomously in real world environments. Range sensors such as LiDAR and RGBD cameras are increasingly found in modern robotic systems, providing a rich source of 3D information that can aid in this task. However, many current systems do not fully utilize this information and have trouble efficiently dealing with large amounts of point cloud data. In this paper, we propose VoxNet, an architecture to tackle this problem by integrating a volumetric Occupancy Grid representation with a supervised 3D Convolutional Neural Network (3D CNN). We evaluate our approach on publicly available benchmarks using LiDAR, RGBD, and CAD data. VoxNet achieves accuracy beyond the state of the art while labeling hundreds of instances per second.

A visually guided shock absorber for high-speed catching without rebound is designed and realized. This visual shock absorber is based on the Maxwell model, which uses plastic deformation to suppress rebounding. The shock absorber is constructed from a spring-based passive elastic body and a servo control-based software damper connected in series. The visual shock absorber is realized in a robot with two degrees of freedom and high-speed vision. The robot receives the impact of a rolling object without repelling it.

The problem of image based visual servoing for robots working in a cluttered dynamic environment is addressed in this paper. It is assumed that the environment is observed by depth sensors which allow to measure the distance between any moving obstacle and the robot. Also an eye-in-hand camera is used to extract image features. The main idea is to control suitable image moments and to relax a certain number of robot's degrees of freedom during the interaction phase. If an obstacle approaches the robot, the main visual servoing task is relaxed partially or completely, while the image features are kept in the camera field of view by controlling the image moments. Fuzzy rules are used to set the desired values of the image moments. Beside that, the relaxed redundancy of the robot is exploited to avoid collisions. After removing the risk of collision, the main visual servoing task is resumed. The effectiveness of the algorithm is shown by several case studies on a KUKA LWR 4 robot arm.

To increase an industrial production rate in a robotic operation, automated assembly tasks on production lines require the operation being performed while a target is still moving. Due to complicated factory environments and missions, unexpected vision occlusion, switching grippers, and grasping the target in motion are demanded. Therefore, a vision guided robotic system consisting of a joint space adaptive controller and a visual target motion estimator is proposed. With the visual target motion estimator, current target velocities and positions are estimated. Regarding the current target velocities and positions, desired robotic joint velocities and positions are computed even if vision is occluded. Moreover, by tracking the desired joint velocities and positions with the adaptive controller, a robotic end-effector is controlled to catch the target even if target and gripper parameters are uncertain. In experiments, grasping the moving target at different speeds with the different grippers for different lengths of occlusion time by using our proposed system were presented for performance evaluations according to its efficiency and robustness.

The performance of any cooperative task that involves two or more robots will be determined by their capacity to recognize common information of the environment. Vision sensors are very effective for this particular goal, but the cost of transmitting the visual information represents a real issue, even more if communication must be performed in narrow bandwidth networks and/or over a multi-hop path. Visual vocabularies provide a dimensionality reduction that has been effectively used in computer vision to reduce the computational load of performing searches in large volumes of data. In this paper we propose to exploit the same technique to decrease the volume of information that is exchanged in the network. This way, robots do not need to send the full descriptors associated to the features they observe, but only the word indices of the corresponding features in the vocabulary. Experiments with a wide variety of vocabularies are used to evaluate the quality of the association given by the algorithm. Finally, real experiments in a wireless network with a limited bandwidth are reported, showing the advantages of the proposed method compared to the communication of full images or feature descriptors.

Concentric tube robots have shown promise for minimally invasive surgical (MIS) tasks that require navigation via tortuous anatomical paths. Despite extensive research on their kinematic and dynamic modelling, however, inaccuracies and deformations of their shape due to unknown loads and collisions with the anatomy make intraoperative shape sensing a requirement. This paper presents a vision-based shape-sensing algorithm for concentric tube robots. The proposed algorithm fuses information extracted from a standard imaging modality, monoplane X-ray fluoroscopy, with the kinematics model of the concentric tube robot, to achieve automatic, real-time, accurate and continuous robot-shape estimations despite kinematics' noise and unmodelled forces. Fusion is performed by a fast 2D/3D non-rigid registration, which combines kinematics and intraoperative tracking of the robot. Extensive simulations with a range of noise models and virtual loads acting on the robot, and experimental evaluation in air and in a skull phantom, demonstrate the clinical value of the proposed technique1.

In this system paper we present a novel approach for autonomous, mobile manipulation for agricultural robots. Our target application is mechanical weed control which for example is needed in organic farming. Today, this task is often performed by field workers, whose availability is declining and the quality of their work differs greatly. In addition, manual weeding is a very tedious task with adverse health effects. Hence, mechanical weed control is considered as a prime example for automation. One of the challenges in order to perform the task efficiently is the need for fast plant detection and precise treatment of single plants with a mobile manipulator. The unstructured environment and rough field conditions raise additional challenges. We introduce a first system concept and control design known to handle these issues. In particular, high speed image processing and visual servoing are applied to precisely position a specially designed weeding tool. To evaluate the performance of the mobile manipulator we run experiments with a mobile test vehicle in field-like conditions and demonstrate that our system can treat single plants in less than 1s as it is required to be competitive.

The vast majority of existing methods for vision-aided inertial navigation rely on the detection and tracking of point features in the images. However, in several man-made environments, such as indoor office spaces, straight line features are prevalent, while point features may be sparse. Therefore, developing methods that will enable the use of straight-line features for vision-aided inertial navigation can lead to improved performance. While limited prior work on the subject exists, it assumes the use of a global-shutter camera, i.e., a camera in which all image pixels are captured simultaneously. Most low-cost cameras, however, use rolling-shutter (RS) image capture, which renders the existing methods inapplicable. To address these limitations, we here present an algorithm for vision-aided inertial navigation that employs both point and line features, and is capable of operation with RS cameras. The two key contributions of this work are (i) a novel parameterization for 3D lines, which is shown to exhibit better linearity properties than existing ones, and (ii) a novel approach for the use of line observations in images. This approach forgoes line-fitting and does not assume that a straight line in 3D projects to a straight line in the image, and is thus suitable for use with RS cameras. Our results demonstrate that our proposed estimator formulation leads to improved precision, in point-feature-poor environments.

For mobile robots and position-based services, such as healthcare service, precise localization is the most fundamental capability while low-cost localization solutions are with increasing need and potentially have a wide market. A low-cost localization solution based on a novel Visible Light Communication (VLC) system for indoor environments is proposed in this paper. A number of modulated LED lights are used as beacons to aid indoor localization additional to illumination. A Gaussian Process(GP) is used to model the intensity distributions of the light sources. A Bayesian localization framework is constructed using the results of the GP, leading to precise localization. Path-planning is hereby feasible by only using the GP variance field, rather than using a metric map. Dijkstra's algorithm-based path-planner is adopted to cope with the practical situations. We demonstrate our localization system by real-time experiments performed on a tablet PC in an indoor environment.

We study the problem of planning paths for a team of robots motivated by coverage, persistent monitoring and surveillance applications. The input is a set of target points in a polygonal environment that must be monitored using robots with omni-directional cameras. The goal is to compute paths for all robots such that every target is visible from at least one path. The cost of a path is given by the weighted combination of the length of the path (travel time) and the number of viewpoints along the path (measurement time). The overall cost is given by the maximum cost over all robot paths and the objective is to minimize the maximum cost. In its general form, this problem is NP-hard. In this paper, we present an optimal algorithm and a constant factor approximation for two special versions of the problem. In both cases, the paths are restricted to lie on a pre-defined curve in the polygon. We show that if the curve satisfies a special property, termed chain-visibility, then there exists an optimal algorithm for monitoring a given set of target locations. Furthermore, if we restrict the input polygon to the class of street polygons, then we present a constant-factor approximation which is applicable even if the set of target locations is the entire polygon. In addition to theoretical proofs, we also present results from simulation studies.

This paper describes the indoor snowfall simulation chamber and its application for evaluating visibility performance of vision-based sensors affected by environmental conditions. First, vision-based safety sensors are introduced and outdoor environmental requirements for these sensors are discussed. And also, the relationship between the visibility and the performance of vision based sensors is discussed and the current problems for evaluating visibility performance are described. Next, the indoor snowfall simulation method using expanded polystyrene beads (EPB) is explained and the spectral transmission measurement equipment for evaluating the visibility reduction is presented. Finally, the visibility reduction caused by simulated and artificial snowfall are measured for verifying the proposed indoor snowfall simulation chamber. Preliminary experiments are also carried out in both snowfall systems in order to evaluate various kinds of vision-based safety sensors for personal care robots.

Whole-Body Control methods offer the potential to execute several tasks on highly redundant robots, such as humanoids. Unfortunately, task combinations often result in incompatibilities which generate undesirable behaviors. Prioritization techniques can prevent tasks from perturbing one another but often to the detriment of the lower precedence tasks. For many tasks, static prioritization is not necessary or even appropriate because tasks can often be achieved in variable ways, as in reaching. In this paper, we show that such task variability can be used to modulate task priorities during execution, to temporarily deviate certain tasks as needed, in the presence of incompatibilities. We first present a method for mapping from task variance to task priority and then provide an approach for computing task variance. Through three common conflict scenarios, we demonstrate that mapping from task variance to priorities reactively solves a number of task incompatibilities.

In this paper a model-free approach for damping control of Variable Stiffness Actuators is proposed. The idea is to take advantage of the possibility to change the stiffness of the actuators in controlling the damping. The problem of minimizing the terminal energy for a one degree of freedom spring-mass model with controlled stiffness is first considered. The optimal bang-bang control law uses a maximum stiffness when the link gets away from the desired position, i.e. the link velocity is decreasing, and a minimum one when the link is going towards it, i.e. the link velocity is increasing. Based on Lyapunov stability theorems the obtained law has been proved to be stable for a multi-DoF system. Finally, the proposed control law has been tested and validated through experimental tests.

A novel variable stiffness actuator composed of a dielectric elastomer actuator (DEA) and a low-melting-point-alloy (LMPA) embedded silicone substrate is demonstrated. The device which we call variable stiffness dielectric elastomer actuator (VSDEA) enables functional soft robots with a simplified structure, where the DEA generates a bending actuation and the LMPA provides controllable stiffness between soft and rigid states by Joule heating. The entire structure of VSDEA is made of soft silicones with an elastic modulus of less than 1 MPa providing a high compliance when the LMPA is active. The device has the dimension of 40 mm length × 10 mm width × 1 mm thickness, with mass of ∼1 g. We characterize VSDEA in terms of the actuation stroke angle, the blocked force, and the reaction force against a forced displacement. The results show the controllable actuation angle and the blocked force up to 23.7 ° and 2.4 mN in the soft state, and 0.6 ° and 2.1 mN in the rigid state. Compared to an actuator without the LMPA, VSDEA exhibits ∼90× higher rigidity. We develop a VSDEA gripper where the mass of active parts is ∼2 g, which is able to successfully hold an object mass of 11 g, exhibiting the high performance of the actuator.