ICRA 2015 论文列表

Compared to other parallel robots, cable-driven parallel robots can achieve bigger workspace, higher dynamics with lighter parts and fewer energy while being economically competitive. However, such robots with big workspaces are affected either by cable sagging or by a very low end-effector stiffness, both causing positioning errors. With lightweight cables, the very low stiffness yields disturbing vibrations. In this paper, we address the problem of decoupled active canceling of those vibrations in the modal space. In modal space, vibrations are decomposed in orthogonal signals which allows to use simple SISO control methods to build an active vibration canceling algorithm for the end-effector. As the modal space depends on the end-effector position, better performances are achieved if the controller is tuned for the current robot position. The proposed method is tested both on a realistic simulation and on a 8-cables and 6-degrees of freedom robot.

This paper approaches from an optimal control perspective the problem of fixed-time detection of mobile radioactive sources in transit by means of a collection of mobile sensors. Under simplifying assumptions on the geometry of the source, sensors, and surrounding environment, and with knowledge of the source's trajectory, it is shown that the optimal control problem admits an intuitive, analytic closed-form solution. This is facilitated by the availability of analytic expressions for bounds on the probabilities of detection and false alarm during a Neyman-Pearson detection test. The intuition derived from this analytic solution leads to a motion control law that steers suboptimally the sensors to an arbitrarily small neighborhood of the suspected source, while navigating among stationary obstacles in their environment, making a step closer to a practical physical implementation.

This paper presents an active object recognition and pose estimation system for household objects in a highly cluttered environment. A sparse feature model, augmented with the characteristics of features when observed from different viewpoints is used for recognition and pose estimation while a dense point cloud model is used for storing geometry. This strategy makes it possible to accurately predict the expected information available during the Next-Best-View planning process as both the visibility as well as the likelihood of feature matching can be considered simultaneously. Experimental evaluations of the active object recognition and pose estimation with an RGB-D sensor mounted on a Turtlebot are presented.

We propose a novel method for robotic exploration that evaluates paths that minimize both the joint path and map entropy per meter traveled. The method uses Pose SLAM to update the path estimate, and grows an RRT* tree to generate the set of candidate paths. This action selection mechanism contrasts with previous appoaches in which the action set was built heuristically from a sparse set of candidate actions. The technique favorably compares agains the classical frontier-based exploration and other Active Pose SLAM methods in simulations in a common publicly available dataset.

We present a novel efficient algorithm for object classification. Our method is based on the active learning framework, in which training and classification are performed in loops, and new ground truth labels are queried from the supervisor in each loop. Our underlying classifier is from the family of boosting methods, but in contrast to earlier methods, our Confidence Boosting particularly focusses on misclassified samples that have a high classification confidence associated. We show that weighting these samples more than others leads to a decrease of overconfidence, for which we give a formal definition. As a result, our classifier is better suited for active learning, leading to steeper learning curves and less required label queries. We show the benefits of our approach on standard data sets from machine learning and robotics.

We consider the task of actively exploring unstructured environments to facilitate robotic grasping of occluded objects. Typically, the geometry and locations of these objects are not known a priori. We mount an RGB-D sensor on the robot gripper to maintain a 3D voxel map of the environment during exploration. The objective is to plan the motion of the sensor in order to search for feasible grasp handles that lie within occluded regions of the map. In contrast to prior work that generates exploration trajectories by sampling, we directly optimize the exploration trajectory to find grasp handles. Since it is challenging to optimize over the discrete voxel map, we encode the uncertainty of the positions of the occluded grasp handles as a mixture of Gaussians, one per occluded region. Our trajectory optimization approach encourages exploration by penalizing a measure of the uncertainty. We then plan a collision-free trajectory for the robot arm to the detected grasp handle. We evaluated our approach by actively exploring and attempting 300 grasps. Our experiments suggest that compared to the baseline method of sampling 10 trajectories, which successfully grasped 58% of the objects, our active exploration formulation with trajectory optimization successfully grasped 93% of the objects, was 1.3× faster, and had 3.2× fewer failed grasp attempts.

Being able to manipulate degrees of freedom of the environment, such as doors or drawers, is a requirement for most tasks a robot is supposed to perform. Often these external degrees of freedom depend on other ones, e.g., a drawer can only be opened if the lock is not locking the joint. We propose an approach to autonomously and efficiently explore and uncover joint dependency structures. We develop a probabilistic model for joint dependency structures which is the basis for active learning. Discontinuities in the dynamics of the joint, which often indicate key points of the joint, are used to segment the joint space into meaningful segments which then allows efficient exploration with the developed maximum cross-entropy (MaxCE) exploration strategy. Experiments in a simulated environment and on a real PR2 suggest that the proposed approach yields efficient exploration of joint dependency structures.

We propose a method for controlling multiple active drifters in the presence of external forcing induced by the ocean. Our active drifters have one actuator: they can lower and raise their drogues in depth. By exploiting the vertically stratified nature of ocean currents, we show how classical multi-robot tasks (spreading out and aggregation) can be accomplished by the multi-drifter system. Tests with a realistic simulation based on an ocean model suggest that a practical implementation of active drifters which aggregate and disperse in the coastal ocean could be realized through our control method with relatively inexpensive components. Specifically, we are able to show that over a 90 day deployment a significant fraction of drifters can be made to aggregate in few clusters suitable for recovery.

We introduce a particle filter-based approach to representing and actively reducing uncertainty over articulated motion models. The presented method provides a probabilistic model that integrates visual observations with feedback from manipulation actions to best characterize a distribution of possible articulation models. We evaluate several action selection methods to efficiently reduce the uncertainty about the articulation model. The full system is experimentally evaluated using a PR2 mobile manipulator. Our experiments demonstrate that the proposed system allows for intelligent reasoning about sparse, noisy data in a number of common manipulation scenarios.

We present a visually guided, dual-arm, industrial robot system that is capable of autonomously flattening garments by means of a novel visual perception pipeline that fully interprets high-quality RGB-D images of a clothing scene based on an active stereo robot head. A segmented clothing range map is B-Spline smoothed prior to being parsed by means of shape and topology analysis into ‘wrinkle’ structures. The length, width and height of each wrinkle is used to quantify the topology of each wrinkle and thereby rank wrinkles by size such that a greedy algorithm can identify the largest wrinkle present. A flattening plan optimised for the largest detected wrinkle is formulated based on dual-arm manipulation. We report the validation of our autonomous flattening behaviour and observe that dual-arm flattening requires significantly fewer manipulation iterations than single-arm flattening. Our experimental results also reveal that the flattening process is heavily influenced by the quality of the RGB-D sensor: use of a custom off-the-shelf high-resolution stereo-based sensor system outperformed a commercial low-resolution kinect-like camera in terms of required flattening iterations.

This paper presents a multi-robot system for autonomous vehicle extraction and transportation based on the “a-robot-for-a-wheel” concept. The developed prototype is able to extract vehicles from confined spaces with delicate handling, swiftly and in any direction. The novel lifting robots are capable of omnidirectional movement, thus they can under-ride the desired vehicle and dock to its wheels for a synchronized lifting and extraction. The overall developed system applies reasoning about available trajectory paths, wheel identification, local and undercarriage obstacle detection, in order to fully automate the process. The validity and efficiency of the AVERT robotic system is illustrated via experiments in an indoor parking lot, demonstrating successful autonomous navigation, docking, lifting and transportation of a conventional vehicle.

There is a widely recognized need for improved STEM education and increased technological literacy. Robots represent a promising educational tool with potentially large impact, due to their broad appeal and wide relevance; however, many existing educational robot platforms have cost as a barrier to widespread use. Here we present AERobot, a simple low-cost robot that can be easily used for introductory programming and robotics teaching, starting from a primary or middle school level. The hardware is open-source and can be built for ∼$10 per robot, making it possible for each student to have (and keep) their own robot, while still encompassing a rich sensor suite enabling a variety of activities. A free, open-source graphical programming environment allows students without previous programming experience to command the robot. We report on the results of three sessions of a one-week pilot course held in the summer of 2014 by STEM summer camp i2 Camp.

The needle-sized surgical tools used in arthroscopy, otolaryngology, and other surgical fields could become even more valuable to surgeons if endowed with the ability to navigate around sharp corners to manipulate or visualize tissue. We present a needle-sized wrist design that grants this ability. It can be easily interfaced with manual tools or concentric tube robots and is straightforward and inexpensive to manufacture. The wrist consists of a nitinol tube with several asymmetric cutouts, actuated by a tendon. Perhaps counter-intuitively, within this seemingly simple design concept, design optimization is challenging due to the number of parameters available and nonlinearities in material properties. In this paper, we examine a subset of possible geometries and derive kinematic and static models. Experimental results with a 1.16 mm diameter prototype validate the models. Lastly, we provide a discussion summarizing the lessons learned in our early experience designing and fabricating wrists of this type.

Learning from demonstration (LfD) is a common technique applied to many problems in robotics, such as populating grasp databases, training for reinforcement learning of high-level skill sets and bootstrapping motion planners. While such approaches are generally highly valued, they rely on the often time-consuming process of gathering user demonstrations, and hence it becomes difficult to attain a sizeable dataset. In this paper, we present a tool capable of recording large numbers of high-dimensional demonstrations of mobile manipulation tasks provided by non-experts in the field. Our tool accomplishes this via a web interface that requires no additional software to be installed beyond a web browser, as well as a scalable architecture that is capable of supporting 10 concurrent demonstrators on a single server. Our architecture employs a lightweight simulation environment to reduce unnecessary computations and improve performance. Furthermore, we show how our tool can be used to gather a large set of demonstrations of a mobile manipulation task by leveraging existing crowdsource platforms. The data set collected has been made available to the robotics community. We also present experiments in which we apply demonstrations collected through our infrastructure to teach a robot how to grasp, to teach a robot how to perform dexterous manipulation tasks such as scooping and to accelerate motion planning for full-body manipulation tasks.

Water sampling with autonomous aerial vehicles has major applications in water monitoring and chemical accident response. Currently, no robot exists that is capable of both underwater locomotion and flight. This is principally because of the major design tradeoffs for operation in both water and air. A major challenge for such an aerial-aquatic mission is the transition to flight from the water. The use of high power density jet propulsion would allow short, impulsive take-offs by Micro Air Vehicles (MAVs). In this paper, we present a high power water jet propulsion system capable of launching a 70 gram vehicle to speeds of 11m/s in 0.3s, designed to allow waterborne take off for an Aquatic Micro Air Vehicle (AquaMAV). Jumps propelled by the jet are predicted to have a range of over 20m without gliding. Propulsion is driven by a miniaturised 57 bar gas release system, with many other applications in pneumatically actuated robots. We will show the development of a theoretical model to allow designs to be tailored to specific missions, and free flying operation of the jet.

A tissue-independent model to estimate needle deflection during insertion in soft tissue is presented in this paper. A force/torque sensor is connected to the needle base in order to measure forces and moments during insertion due to needle deflection. A static mechanical model, which is based on the Euler-Bernoulli beam equation and the balance of forces applied by the tissue onto the needle takes these force and moment measurements as input. The needle tip deflection can then be calculated based on the beam model undergoing these forces. Three different needle-tissue interaction models are presented. Their estimation performance is evaluated and experimentally compared by carrying out insertion experiments into phantom tissue. The experimental results show a precise estimate of needle tip deflection for a novel virtual sensor introduced in this work. The main advantage of this virtual sensor approach is that measurements obtained from the force/torque sensor are the only necessary model inputs. Furthermore, the approach does not rely on ultrasound or other image-based needle observation techniques. This makes the virtual sensor suitable for real-time feedback of needle tip deflection.

In this paper we present an online approach to segmenting roads on large scale trajectories using only a monocular camera mounted on a car. We differ from popular 2D segmentation solutions which use single colour images and machine learning algorithms that require supervised training on huge image databases. Instead, we propose a novel approach that fuses 3D geometric data with appearance-based segmentation of 2D information in an automatic system. Our contribution is twofold: first, we propagate labels from frame to frame using depth priors of the segmented road avoiding user interaction most of the time; second, we transfer the segmented road labels to 3D laser point clouds. This reduces the complexity of state-of-the-art segmentation algorithms running on 3D Lidar data. Segmentation fails is in only 3% of the cases over a sequence of 13,600 monocular images spanning an urban trajectory of more than 10km.

We report on a model-based approach for robot design and its dynamic motion generation. A new torque-actuated dissipative spring loaded inverted pendulum model with rolling contact (TDR-SLIP) is proposed to serve as the motion template for the robot. It is a successor to a previously developed spring loaded inverted pendulum model with rolling contact (R-SLIP) model but with embedded energy flow, which has better mapping to empirical robots. The stability properties of the TDR-SLIP model are analyzed, and its stable motion trajectory is implemented on the robot as the control guidance. The robot leg is developed according to the morphology and function of the TDR-SLIP leg, which acts as the design guidance. The proposed model-based approach is experimentally evaluated and finds that the robot is able to exhibit running behavior with various speeds and leg mechanics settings.

This paper, as a generalization of our previous works, presents a unified time-optimal path-following controller for Wheeled Mobile Robots (WMRs). Unlike other path-following controllers, we solve the path-following problem for all common categories of WMRs such as car-like, differential, omnidirectional, all wheels steerable and others. We show that the insertion of our path-following controller into the kinematic and non-holonomic constraints of the wheels, simplifies the otherwise impenetrable constraints, resulting in explicit monotonic functions between the velocity of the base and that of the wheels. Based on this foundation, we present a closed-form solution that keeps all the wheels' steering and driving velocities within their corresponding pre-specified bounds. Simulation data and experimental results from executing the controller in a real-time environment demonstrate the efficacy of the method.