ICRA 2015 Papers

We propose a method for recognizing rigid object instances in RGB-D point clouds by accumulating low-level information from keypoint correspondences over multiple observations. Compared to existing multi-view approaches, we make fewer assumptions on the recognition problem, dealing with cluttered and partially dynamic environments as well as covering a wide range of objects. Evaluation on the publicly available TUW and Willow datasets showed that our method achieves state-of-the-art recognition performance for challenging sequences of static environments and a significant improvement for environments partially changing during the observation.

During robotic teleoperation, the dynamics of the master manipulators and the control of remote-side instruments impose challenges on the motor system of the human operator, and may impact performance and learning. In teleoperated robot-assisted minimally invasive surgery, there is a clear correlation between patient outcomes and the surgeon's case experience. However, the effect of the teleoperator on human motor skills and the relationship between these motor skills and patient outcomes are unknown. We used the da Vinci Research Kit, a custom research version of the da Vinci Surgical System, to compare teleoperated and open needle-driving movements of experienced da Vinci surgeons and novices. The experimental protocol consisted of structured but unconstrained needle driving trials repeated 80 times to allow for computational modeling of movement coordination and learning. Kinematic analysis showed that teleoperation increases trial time but reduces path length, that the trial times and path lengths of experienced surgeons are smaller than those of novices. In addition, there are significant differences in learning between experienced surgeons and novice users. Modeling of the movements and learning processes of experienced and novice surgeons may be used in the design of novel controllers that will expand robotic surgery capabilities and improve robot-assisted surgical skill acquisition.

Personal robots need to manipulate a variety of articulated mechanisms as part of day-to-day tasks. These tasks are often specific, goal-driven, and permit very little bootstrap time for learning the articulation type. In this work, we address the problem of purposefully manipulating an articulated object, with uncertainty in the type of articulation. To this end, we provide two primary contributions: first, an efficient planning algorithm that, given a set of candidate articulation models, is able to correctly identify the underlying model and simultaneously complete a task; and second, a representation for articulated objects called the Generalized Kinematic Graph (GK-Graph), that allows for modeling complex mechanisms whose articulation varies as a function of the state space. Finally, we provide a practical method to auto-generate candidate articulation models from RGB-D data and present extensive results on the PR2 robot to demonstrate the utility of our representation and the efficiency of our planner.

This paper presents an effective method for planning and optimizing robot motions in joint space using via-points. The via-point formulation allows for a sparse movement representation which is inherently smooth according to a minimum jerk criterion. In this research we focus on two aspects. First, we present an initialization method to find a feasible number of via-points and their distribution in time. This initialization takes the difficulty of the planning problem into account and finds a tradeoff between representation complexity (number of via-points) and optimization time. Second, we propose a two-step optimization approach to improve the via-point parameters to satisfy constraints over the whole motion duration. We show that our methods generate feasible and smooth robot motions with a high success rate for tasks of varying complexity. The methods are evaluated in simulation using four different scenarios with two different robots, a 7-DOF robot arm and a 16-DOF humanoid upper body. Additionally, the experiments with the robot arm are shown in a real-world experiment.

This paper addresses the motion planning problem in the presence of obstacles for underactuated robots that are assigned a geometric task. It is assumed that the robot is subject to kinematic (joint limits, joint velocity bounds) as well as dynamic (torque bounds) constraints. Building on our previous work on task-constrained motion planning, we describe a randomized planner that works directly at the torque level and generates solutions by separating geometric motions from time history. The effectiveness of the proposed approach is shown by planning collision-free swing-up maneuvers for a Pendubot system.

Grasp planning in multi-robot systems is usually studied in a centralized setting with all robots sharing common knowledge about the overall system. Relaxing this assumption would allow multiple mobile manipulators to cooperate even without strict and precise coordination. Moreover, most typical tasks for cooperative settings, such as transporting heavy objects, require certain forces/torques to be exerted along/around particular directions, for instance, compensating for the weight of the transported object. In this paper, we propose task specific multi-robot grasp planning strategies that allow decentralized planning. Each agent plans its own actions without precise information about the other's plans. The approach is based on analysing a task specific grasp quality metric in a probabilistic context, compensating thus for the incomplete knowledge. Results from simulation experiments demonstrate that task independent planning is clearly inferior when task characteristics are known and thus task specific quality measures should be used. Furthermore, the proposed decentralized planning approaches clearly outperform the baseline and show close to globally optimal performance.

We present a novel task scheduling algorithm for use on mobile robots in real environments. The scheduling problem is formalised as mixed integer program, which is a standard approach in the scheduling community. Our contribution is the use of Allen's interval algebra to prune the search to be performed by the mixed integer program. This significantly speeds up the algorithm. The proposed algorithm has been used on several mobile robots in long-term autonomy scenarios, where it schedules large sets containing a variety of tasks. The proposed algorithm outperforms the state of the art by at least one order of magnitude on both these real tasks and synthetic datasets.

Fall prevention is one of the most important functions of walking assistance devices for user's safety. It is preferable that these devices obviates factors which induce falling over rather than helping them recovering from falling motion. Tandem stance, where both legs form a line along walking direction, is a factor of falling. It is often observed in turning motion because a swing leg moves in lateral direction as well as forward. Generally, upper body's turning of a person proceeds before lower limb's turning during walking, and the behavior of the lower limb is constrained by upper body turning. It is thus possible to control behavior of the lower limb for tandem stance prevention by constraining behavior of the upper body. This paper therefore introduces a tandem stance prevention method for safer walk of the elderly or physically challenged person. The method adjusts admittance of cane robot's turning motion according to the positions of a support leg and a swing leg. The cane robot measures user's leg position and estimates a walk phase of the user. By adjusting an admittance model of the cane robot based on the walk phase, the cane robot resists to turn while support leg is on the same side of the turning direction. As a result, the tandem stance is avoided. Through experiments, we confirmed that the cane robot successfully controls user's upper body's behavior to prevent the user from being in tandem stance.

Stroke can lead to loss or impairment of somatosensory sensation (i.e. proprioception), that reduces functional control of limb movements. Here we examine the possibility of providing artificial feedback to make up for lost sensory information following stroke. However, it is not clear whether this kind of sensory substitution is even possible due to stroke-related loss of central processing pathways that subserve somatosensation. In this paper we address this issue in a small cohort of stroke survivors using a tracking task that emulates many activities of daily living. Artificial proprioceptive information was provided to the subjects in the form of vibrotactile cues. The goal was to assist participants in guiding their arm towards a moving target on the screen. Our experiment indicates reliable tracking accuracy under the effect of vibrotactile proprioceptive feedback, even in subjects with impaired natural proprioception. This result is promising and can create new directions in rehabilitation robotics with augmented somatosensory feedback.

This paper presents the Tact hand—an anthropomorphic, open-source, myoelectric prosthetic hand that was designed for use by people with transradial amputations in developing countries. This hand matches or exceeds the performance of other state-of-the-art myoelectric prosthetic hands, but costs two orders of magnitude less ($250) and is easy to manufacture with a 3D printer and off-the-shelf parts. We describe our design process, evaluate the Tact hand with both qualitative and quantitative measures of performance, and show examples of using this hand to grasp household objects.

Tendon driven continuum robots have attracted a significant amount of attention in the last decade. As a result several discrete and continuous modeling approaches have been explored in order to understand their behavior. The following article proposes an analytical model describing the influence of frictional effects between the tendon and its conduit in tendon driven continuum robotic manipulators. Existing constant curvature models ignore internal device friction completely and treat it as a negligible effect. This results in an assumption that the tendon tension is always constant along the length of the continuum manipulator. While this assumption is reasonable for small articulation angles, internal device friction can significantly change internal tension distribution deteriorating performance of constant curvature models at high articulation angles. The proposed model departs from the constant tension assumption and predicts a tension distribution along the tendon. From the tension distribution a curvature distribution is determined. It is observed that the proposed nonlinear friction model performs better than the existing linear elastic constant curvature model. Following this, a consequence of internal device friction: hysteresis effects are predicted. An algorithm that estimates tendon tension distribution given applied tendon tension history is proposed.

The Spherical Underactuated Planetary Exploration Robot ball (SUPERball) is an ongoing project within NASA Ames Research Center's Intelligent Robotics Group and the Dynamic Tensegrity Robotics Lab (DTRL). The current SUPERball is the first full prototype of this tensegrity robot platform, eventually destined for space exploration missions. This work, building on prior published discussions of individual components, presents the fully-constructed robot. Various design improvements are discussed, as well as testing results of the sensors and actuators that illustrate system performance. Basic low-level motor position controls are implemented and validated against sensor data, which show SUPERball to be uniquely suited for highly dynamic state trajectory tracking. Finally, SUPERball is shown in a simple example of locomotion. This implementation of a basic motion primitive shows SUPERball in untethered control.

This paper computes all the continuous point-transformation symmetries for a planar multi-agent dynamical control system. The relative configuration among the agents is often the most important aspect of this problem, and this paper presents a method to “factor out” the location and orientation on the plane of a multi-agent formation problem. This is useful in the case where only the relative positioning of the agents is of concern. The method to compute the symmetries and determine the reduced coordinates is well-established. The contributions of this paper are (1) computing the symmetries for a multi-agent system, (2) highlighting the utility of the reduction for the formation control application such as formation stability and (3) analyzing the symmetries to determine a relatively broad class of systems which have the same symmetries and hence identical coordinates for reduction. Although it has been addressed in various ways, essentially all methods to determine the stability properties of the formation control problem need the dynamics to be formulated on a reduced, relative space. Hence, stability (and stability-like notions such as that which follow from LaSalle's Principle) analyses will be the main beneficiary of this work.

Force sensing and control are of paramount importance in robotic micromanipulation. A contact force regulator capable of accurately applying mechanical stimuli to a live Drosophila larva could greatly facilitate mechanobiology research on Drosophila and may eventually lead to novel discoveries in mechanotransduction of neuron circuitry. In this paper, we present a novel contact force scheme implemented in an automated Drosophila larvae micromanipulation system, featuring a switched fuzzy-PD controller and a noise-insensitive extended high gain observer. The switched fuzzy-PD controller inherits the fast convergence from fuzzy controller and overcomes the drawbacks (overshoot and oscillation) of a conventional fuzzy controller. The observer provides precise estimation to compensate for system modeling errors despite force measurement noise, which overcomes the disadvantage of traditional high gain observer. Force control experiments show that, compared to an conventional PID controller, this new controller-observer scheme has significantly enhanced dynamic performance in terms of rising time, overshoot, and oscillation. The developed robotic system and the force control scheme will be applied to mechanical stimulation and fluorescence imaging of Drosophila larvae for identifying new mechanotransduction mechanisms.

Transportation manipulation of biological cells, where cells are required to move into a fixed or moving region, has recently attracted increasing attention in bioscience and nanomedicine. Currently, the multicell transportation in practical applications is implemented manually, with low precision and efficiency. This paper presents a swarm-inspired approach to automated transportation of multiple cells using robotically controlled optical tweezers. A swarming controller, where holographic optical tweezers function as end-effectors to manipulate the cells, was developed. To ensure that the cells do not escape from the optical traps, the controller was designed by incorporating a saturation control of the cell offset to the laser center. Because the optical tweezers can only be position-controlled, oscillation may easily occur. This problem was solved by integrating artificial first-order kinematics of the optical tweezers into the controller design. Experiments of transporting multiple yeast cells were performed to verify the effectiveness of the proposed approach.

Surgical tool pose estimation has been proven to be useful for high- and low- level feedback tasks including safety-enhancement, semantic feedback and surgical skill assessment. Tool pose estimation using monocular camera input is a well-studied research problem as the monocular camera is one of the ubiquitous sensor across the spectrum of robotic devices. Current state-of-the art methods for visual tool pose estimation are computationally expensive and require elaborate geometric and appearance models of surgical tools. We propose a visual tool pose estimation method that maps the visual bounding box to the 3D tool pose without any explicit knowledge of tool geometry using Gaussian process regression. The proposed approach can be generalized to any surgical tool and provides tool pose estimates with a variance estimate in real-time. We demonstrate rigorous evaluation of the method under various conditions that might effect the estimation process. In order to evaluate the algorithm, we have instrumented a standard box trainer kit with two laparoscopic tools to get simultaneous ground truth pose and a video feed.

Using Unmanned Aerial Vehicles (UAVs) to deploy sensor networks promises an autonomous and useful method of installation in remote or hard to access locations. Some sensors, such as soil moisture sensors, must be physically installed in soft soil, yet UAVs cannot easily determine soil softness with remote sensors. In this paper, we use data from an onboard accelerometer measured during UAV landings to determine the softness of the ground. We collect and analyze over 200 data sets gathered from 8 different materials: foam, carpet, wood, tile, grass, dirt, concrete, and woodchips. Based on this analysis, we examine a number of features from the accelerometer and four classification algorithms: LDA, QDA, SVM, and binary decision trees. The decision tree performs well and is simple to implement onboard the UAV. We implement this in our UAV control system and perform experiments to verify that the UAV can accurately classify the softness of the surface with 90% accuracy. This lays the groundwork for our future work on developing a UAV capable of installing sensors in soft soil.

Most contemporary manufacturing systems appear as complex event-driven automation facilities. Supervisor synthesis and simplification are fundamental in automated manufacturing systems (AMSs). From design and implementation standpoints, it is preferable to decrease supervisor scales so as to mitigate their realization expense. Ordinary and general Petri nets are considered as powerful formalisms to describe, analyze, and control AMSs. The former is concise while the latter is compact. By considering both cases, our approach shows some properties so as to disclose their rationales with regard to liveness enforcement. Many control methodologies are developed in the context of the ordinary ones. They are straightforward but might be unadaptable to the general ones. This work develops new theory in this perspective. Some comparison and contrast survey has been demonstrated in order to clarify their common theoretical basis and distinct supervisory effect. In accordance to a thorough study, a novel simplification method is developed so as to unify both cases and facilitate their application. Experimental results show that the developed strategy is effective and efficient when simplifying supervisors.

Colored Petri nets are widely used to model Automated Manufacturing Systems thanks to their compactness to describe complex networked systems. Compared to general Petri nets, they allow many folding techniques so as to condense the system model. With them, many control synthesis problems are reduced to the solution of matrix operations. Supervisory control techniques can be realized in a computationally efficient way. In this work, these matrices are further distinguished by independent and dependent ones. A method is proposed to identify these independent ones while removing those dependent ones. As a result, a supervisor can be significantly simplified. In practice, this implies the significant reduction of cost, increase of reliability, and improvement of performance. Examples are presented to illustrate the effectiveness and efficiency of this method.