ICRA 2015 论文列表

An autonomous robot requires a map of the environment for many tasks. Yet, in many cases, this map is unavailable and the robot must build one in real-time, in the so-called integrated exploration task. Several integrated exploration approaches adopt some sort of loop-closing strategy combined with an online simultaneous localization and mapping (SLAM) technique. This is important because the robot can reduce the uncertainty about its pose by revisiting known areas. One solution for environment exploration is to use the vector field computed from the numeric solution of a Boundary Value Problem (BVP). This approach is called BVP Path Planner and generates smooth and free of local minima potential fields. However, this planner cannot actively close loops and does not scale well in large scenarios. In this paper we present a technique which performs active loop closure using the BVP Path Planner. Our proposal takes advantage of the potential of unexplored regions, and induces the robot to close loops by placing dynamic barriers at the visited space. The update of the potential field is boosted using a local window charged by a Voronoi diagram of the environment containing global information. We show through experimental results the effectiveness of the technique with a thorough discussion of its characteristics.

Origami-engineering is an emerging design and manufacturing technology for creating 3-D robotic systems using 2-D designs. We present a new origami-inspired tele-micromanipulation system manufactured by printing and folding. The developed micromanipulation system consists of two miniature machines, wearable user interface devices and a multi degree-of-freedom slave manipulator. A simple hydraulic system using syringes allows motion reduction and force transmission between the master and the slave. This paper introduces how to design and rapidly prototype these devices, and, finally, demonstrate different micromanipulation abilities such as 2-D trajectory following, pick-and-place, and 3-D micro-assembly. Highlighting rapid design and fabrication of a low-cost high performance origami-based tele-micromanipulation system, this paper proposes new applications of origami-inspired printable robots to wearable robots, micro-robotics and tele-operation.

In this paper we propose a new ultrasound-based visual servoing framework, for the optimization of the positioning of an ultrasound probe manipulated by a robotic arm, in order to improve the quality of the acquired ultrasound images. To this end, we use the recent framework of ultrasound confidence map, which aims at estimating the perpixel quality of the ultrasound signal based on a model of sound propagation in soft tissues. More specifically, we treat the ultrasound confidence maps as a new modality to design a visual servoing control law for image quality optimization. The proposed framework aims at improving ultrasound imaging techniques, such as robotic tele-echography, target tracking or volume reconstruction. Here we illustrate our approach with the application of robotic tele-echography. Experiments are performed on both an ultrasound examination training phantom and ex vivo tissue samples.

Tasks with unknown dynamics and costly system interaction time present a serious challenge for reinforcement learning. If a model of the dynamics can be learned quickly, interaction time can be reduced substantially. We show that combining an optimistic exploration strategy with model-predictive control can achieve very good sample complexity for a range of nonlinear systems. Our method learns a Dirichlet process mixture of linear models using an exploration strategy based on optimism in the face of uncertainty. Trajectory optimization is used to plan paths in the learned model that both minimize the cost and perform exploration. Experimental results show that our approach achieves some of the most sample-efficient learning rates on several benchmark problems, and is able to successfully learn to control a simulated helicopter during hover and autorotation with only seconds of interaction time. The computational requirements are substantial.

This paper utilizes system observability to guide monocular SLAM. Instead of providing all measured features then performing data-driven outlier rejection (such as with RANSAC), we propose to identify only the minimal subset of features which form an optimally observable SLAM subsystem for localization. Modeling the SLAM system as a discrete time system with piece-wise linear SE〈3〉 motion, complete observability conditions are derived and a means to test the observability conditioning of candidate feature point groupings is proposed. Based on the conditioning, an efficient algorithm for picking the optimally observable feature subset is derived by incorporating the image geometric measures. The proposed monocular SLAM algorithm, called Optimally Observable and Minimal Cardinality (OOMC) SLAM is formulated as an EKF process. OOMC SLAM is first validated using a 6-DOF localization experiment; the results demonstrate accuracy comparable to the state-of-art SLAM algorithm with significantly improved computational efficiency. A longer sequence on a 620-meter trajectory is also tested. The algorithm achieves 0.9178% relative error against the GPS ground truth.

Walking capability composed of stability and efficiency is one of the most important issues in the field of humanoid robots. An effective swing of the arms is expected to enhance the walking capability under the constraints from the limited body. We propose an arm-swing method to enhance the stability and efficiency by selecting optimal arm-swing strategy depending on the walking conditions. In this research, we select the optimal strategy between the support of the center of gravity (COG) tracking for stability and the walk without arm-swing for efficiency. To support the COG tracking, we employ a predictive control. States are defined as an inverted pendulum model and inputs are given as an inertial force of arm-swing. Input and output weights in the predictive control are adjustable by a support weight introduced in this paper. Selection of the optimal support weight by a selection algorithm for locomotion (Su-SAL) switches the two strategies by adjusting the ratio of input and output (I/O) weights. Su-SAL maximizes the efficiency while keeping the stability in comparison with the case of the constant support weight.

Sequential decision making under uncertainty is studied in a mixed observability domain. The goal is to maximize the amount of information obtained on a partially observable stochastic process under constraints imposed by a fully observable internal state. An upper bound for the optimal value function is derived by relaxing constraints. We identify conditions under which the relaxed problem is a multi-armed bandit whose optimal policy is easily computable. The upper bound is applied to prune the search space in the original problem, and the effect on solution quality is assessed via simulation experiments. Empirical results show effective pruning of the search space in a target monitoring domain.

Motion planning under differential constraints is a classic problem in robotics. To date, the state of the art is represented by sampling-based techniques, with the Rapidly-exploring Random Tree algorithm as a leading example. Yet, the problem is still open in many aspects, including guarantees on the quality of the obtained solution. In this paper we provide a thorough theoretical framework to assess optimality guarantees of sampling-based algorithms for planning under differential constraints. We exploit this framework to design and analyze two novel sampling-based algorithms that are guaranteed to converge, as the number of samples increases, to an optimal solution (namely, the Differential Probabilistic RoadMap algorithm and the Differential Fast Marching Tree algorithm). Our focus is on driftless control-affine dynamical models, which accurately model a large class of robotic systems. In this paper we use the notion of convergence in probability (as opposed to convergence almost surely): the extra mathematical flexibility of this approach yields convergence rate bounds - a first in the field of optimal sampling-based motion planning under differential constraints. Numerical experiments corroborating our theoretical results are presented and discussed.

The RRT* algorithm has efficiently extended Rapidly-exploring Random Trees (RRTs) to endow it with asymptotic optimality. We propose Goal-Rooted Feedback Motion Trees (GR-FMTs) that honor state/input constraints and generate collision-free feedback policies. Given analytic solutions for optimal local steering, GR-FMTs obtain and realize safe, dynamically feasible, and asymptotically optimal trajectories toward goals. Second, for controllable linear systems with linear state/input constraints, we propose a fast method for local steering, based on polynomial basis functions and segmentation. GR-FMTs with the method obtain and realize trajectories that are collision-free, dynamically feasible under constraints, and asymptotically optimal within a set we define. The formulation includes linear or quadratic programming of small sizes, where constraints are identified by root-finding in low or medium order of polynomials and added progressively.

This paper presents algorithms for optimal selection of needle grasp, for autonomous robotic execution of the minimally invasive surgical suturing task. In order to minimize the tissue trauma during the suturing motion, the best practices of needle path planning that are used by surgeons are applied for autonomous robotic surgical suturing tasks. Once an optimal needle trajectory in a well-defined suturing scenario is chosen, another critical issue for suturing is the choice of needle grasp for the robotic system. Inappropriate needle grasp increases operating time requiring multiple regrasps to complete the desired task. The proposed methods use manipulability, dexterity and torque metrics for needle grasp selection. A simulation demonstrates the proposed methods and recommends a variety of grasps. Then a realistic demonstration compares the performances of the manipulator using different grasps.

We study the problem of planning the shortest path for a polygonal robot anchored to a fixed base point by a finite tether translating among polygonal obstacles in the plane. Specifically, we preprocess the workspace to efficiently answer queries of the following type: Given a source location of the robot and an initial configuration of the tether, compute the shortest path to reach a target location while avoiding obstacles and adhering to the tether's constraints. Our work is an extension of the recent work by Kim et al. [1] who considered the problem for a point robot. Their algorithm relies on a discretization of the workspace and is optimal with respect to this discretization. We first replace their grid-based approach with a visibility-graph based approach. This allows to improve the running time of their algorithm by several orders of magnitude. Specifically, testing on a scenario similar to one presented by Kim et al., the running time is improved by a factor of more than 500. Moreover, our approach, which plans optimal paths, is applicable to polygonal (translating) robots and can be used to plan a shortest path while ensuring a predefined clearance from the obstacles. We report on our experimental results on a variety of scenarios. In all cases the preprocessing time is less than one second on a standard-commodity laptop, and a typical query takes several tens of miliseconds.

Motor Learning is heavily governed by sensory feedback, and artificially enhancing feedback influences learning as seen in our previous works. This study provides a model-based approach in determining the optimal gain schedules for augmented feedback on a visuomotor learning task. Using Gaussian process regression, we modeled the phenomenological process of learning to operate a robot with visual rotation. We then used Pontryagin's minimum principle to achieve the optimal feedback gain schedules that yield the fastest learning, the highest post-training performance, and both at the same time. Our results reveal that the instantaneous error feedback should be doubled (×1.92) throughout the training if the fastest learning is desired. However if the highest post-training performance is desired along with the fastest learning, the feedback gain should be gradually varied from 1.92 to 1. This study explores a novel approach to optimize specific aspects of training for areas such as robotic-neuro-rehabilitation, teleoperation, sports coaching, and human-machine interactions.

An increasing number of robotic systems are using compliant actuators in which springs are placed in series with the actuator. The need for identification procedures tailored to these systems is consequently rising. When measurements of both the link side and the motor side of the spring are available the dynamic parameters can be identified independent of the spring model. The excitation of the identification procedure is optimized for the identification of the dynamical parameters to provide a rich data set by maximizing a reduced information matrix. Adopting this reduced matrix leads to a near-optimal excitation. A Fourier series is chosen as parametrization of the excitation, resulting in a periodic movement of the system under identification. This periodicity is exploited to develop an alternative weighting of the parameter estimation. This proposed weighting uses the motor torque variance at each time step of the trajectory, instead of one global torque variance. This identification procedure is demonstrated and validated on one leg of a lower limb exoskeleton. Only the dynamic model in the sagital plane (2D) is identified.

The MOBOT* project aims to develop assistive devices that support elderly patients with mobility disabilities during gait and sit-to-stand (STS) transfer. The device is supposed to help such patients to maintain an independent life through mobility as well as to regain basic life skills since self-effort is encouraged. Situation-adapted optimal assistance will be provided by the device both proactively and adaptively through behavior-based and context-aware robot control based on multimodal action recognition. The design of such a device, including the choice of actuators, which can provide the desired level of support without losing balance, is a challenging task. For these design choices, in particular STS motions have to be considered since they are more demanding for both the patient and the device. In this paper, we present a model-based optimization approach to this design task, using direct multiple shooting optimal control methods. They are applied to simultaneously determine optimal states and controls as well as the optimal mechanical design parameters of the assistive device that provide optimal STS assistance to subjects from a range of body heights and weights that represents the MOBOT target group.

This paper presents a method for shared control where real-time bursts of optimal control assistance are applied by an observer on-demand to aid a simulated figure in maintaining balance. The proposed Assistive Controller (AC) calculates the optimal burst control fast, in real time, while accounting for nonlinearities of the dynamic model. The short duration of the AC signals allows a rapid transfer of control authority between the nominal and the assistive controller. This scheme avoids prolonged loss of nominal control authority on the part of the figure while facilitating the real-time integration of an external observer's guidance through the assistive control. We demonstrate the benefits of this control scheme in simulation using the Robot Operating System (ROS), in a context where the nominal controller fails to stabilize the figure and the AC is activated intermittently to not only keep it from falling but to additionally push it back to the upright position. The example signifies the efficiency of the proposed model-based AC even in the absence of force/pressure sensors. This approach presents an opportunity for using exoskeletons in balance support, fall prevention, and therapy. In particular, our simulation results indicate that a therapist equipped with an AC interface can, with minimal effort, increase active participation on the part of the patient while ensuring their safety.

This paper addresses a trajectory planning and task allocation problem for a swarm of resource-constrained robots that cannot localize or communicate with each other and that exhibit stochasticity in their motion and task-switching policies. We model the population dynamics of the robotic swarm as a set of advection-diffusion-reaction partial differential equations (PDEs), a linear parabolic PDE model that is bilinear in the robots' velocity and task-switching rates. These parameters constitute a set of time-dependent control variables that can be optimized and broadcast to the robots prior to their deployment. The planning and allocation problem can then be formulated as a PDE-constrained optimization problem, which we solve using techniques from optimal control. Simulations of a commercial pollination scenario validate the ability of our control approach to drive a robotic swarm to achieve predefined spatial distributions of activity over a closed domain, which may contain obstacles.

The term whole-body loco-manipulation refers to the case in which a humanoid robot exploits contacts with the environment, both with the end-effectors and with its internal limbs, in order to balance, move and/or manipulate the environment. In such a situation, high degree of redundancy may not be sufficient to completely control the robot movements and/or the forces applied on the environment. This problem is tackled in this work by means of quasi-static analysis tools. The reduction of mobility and manipulability is studied introducing the Fundamental Loco-Manipulation Matrix (FLMM) and its canonical form (cFLMM). Relevant information on the system can be extracted from those, obtaining, e.g., the space of the controllable contact forces, and the controllable displacements of the center of mass. Furthermore, the best contact force distribution able to meet the friction cone constraints is demonstrated to be the solution of a convex optimization problem. The validity of the proposed methods is verified in two numerical examples, where internal contacts affects the controllability of both forces and displacements. Numerical results show that is crucial to consider the correlations between contact forces in order to exert target actions on the environment while coping with friction limits on the whole set of contacts.

Independent control of several, untethered microrobots remains a significant challenge. Although progress has been made in controlling multiple robots at scales below 100 mm, motion is dependent on external energy sources for propulsion and is often not fully decoupled. Here, we develop a method to dynamically pattern light at a resolution of 1 µm2 per pixel in a closed loop real-time system using a digital micromirror device (DMD) system. We test the system on microbiorobots - bacteria-coated microplates - and demonstrate that we can selectively stimulate pre-defined regions of bioactuators. We verify previously developed dynamic models as a step toward developing individually addressable, self-propelled microrobots.

Microbiology is mainly concerned with the study of coexistence, cooperation and interaction among groups of microorganisms. In order to study the biological systems consist of interacting groups of microorganisms, optical manipulation can be utilized as a useful tool. In fact, the ability to manipulate several groups of microorganisms is an essential step towards studying how biological microorganisms communicate and cooperate to perform a wide range of multicellular behaviours. In addition, decomposition of a group of cells into smaller groups gives capability and flexibility in manipulation tasks and remarkably increases environmental adjustability. In this paper, we aim to provide a primary result in micromanipulation of multiple groups of cells. We develop a control methodology for dynamic coordination of multiple groups of cells. A rigorous mathematical formulation is provided and the stability analysis is presented. Using the proposed method, we are able to manipulate multiple groups of microparticles to construct time-varying micro-formations. The proposed method is also useful in examination of the interactions between several groups of living organisms in the desired inter-cellular distance. Experimental results are presented to illustrate the performance of the proposed method.