ICRA 2015 论文列表

We propose an intuitive user interaction system with virtual objects by directly using user's hands through a physics simulation. This system aims to provide a more direct way to manipulate and interact with virtual objects and to induce more realistic reactions of the virtual objects in the interaction than the existing methods. The key contribution of this paper is to model the deformations of the user's hands in two ways: global deformation by the hand poses and the local deformation by contacts with objects. By the two deformation models, objects can be manipulated with better stability and robustness without any predefinition of interaction methods depending on the object shapes, by solely using a physics simulation. The experiment shows that the proposed method enables a user to intuitively manipulate virtual object accurately.

We are investigating the use of photoacoustic (PA) imaging to detect critical structures, such as the carotid artery, that may be located behind the bone being drilled during robot-assisted endonasal transsphenoidal surgery. In this system, the laser is mounted on the drill (via an optical fiber) and the 2D ultrasound (US) probe is placed elsewhere on the skull. Both the drill and the US probe are tracked relative to the patient reference frame. PA imaging provides two advantages compared to conventional B-mode US: (1) the laser penetrates thin layers of bone, and (2) the PA image displays targets that are in the laser path. Thus, the laser can be used to (non-invasively) extend the drill axis, thereby enabling reliable detection of critical structures that may reside in the drill path. This setup creates a challenging alignment problem, however, because the US probe must be placed so that its image plane intersects the laser line in the neighborhood of the target anatomy (as estimated from preoperative images). This paper reports on a navigation system developed to assist with this task, and the results of phantom experiments that demonstrate that a critical structure can be detected with an accuracy of approximately 1 mm relative to the drill tip.

We present a novel algorithm to compute a gradient-continuous penetration depth (PhongPD) between two interpenetrated polygonal models. Our penetration depth (PD) formulation ensures separating the intersected models by translation, and the amount of such translation is close to an optimal motion to resolve interpenetration in most cases. In order to achieve the gradient-continuity in our algorithm, we interpolate tangent planes continuously over the contact space and then perform a projection along a normal direction defined by the interpolated tangent planes; this projection scheme is known as Phong projection. We have implemented our PhongPD algorithm and certifies its continuity using three benchmarks consisting of diverse combinatorial complexities, and show that our algorithm shows smoother PD results than a conventional Euclidean-projection-based PD method.

Operating and maneuvering in difficult terrains has remained a challenging problem in the field of legged robots. One of the major challenges arises from the high dimensionality inherent in planning foot placements coupled with center of mass motion along terrains that are multifaceted and highly diverse. Previous work has resolved these issues to an extent by constraining the center of mass to fixed trajectories or using predetermined foot placements. To deal with these challenges, this paper proposes a new set of strategies: (1) an optimized geometric Hermite curve with minimum curvature and length is used to plan the motion of the center of mass (2) single contact model dynamics for state-space approximations of center of mass behavior are used to resolve feet transitions between steps in planar environments and (3) vertical center of mass phase space trajectories are optimized to produce an overall plan with minimum energy. This framework allows us to synthesize complex maneuvers in rough terrains and to develop optimal contact transition and foot placement plans that consider the robot's configuration and constraints. Experimental results show that for any potential locations of foot contacts, our planner generates smooth and optimal trajectories for center of mass motion as well as a minimum energy plan for transitioning between foot placements.

Having accurate knowledge of the positions of people around a robot provides rich, objective and quantitative data that can be highly useful for a wide range of tasks, including autonomous person following. The primary objective of this research is to promote the development of robust, repeatable and transferable software for robots that can automatically detect, track and follow people in their environment. The work is strongly motivated by the need for such functionality onboard an intelligent power wheelchair robot designed to assist people with mobility impairments. In this paper we propose a new algorithm for robust detection, tracking and following from laser data. We show that the approach is effective in various environments, both indoor and outdoor, and on different robot platforms (the intelligent power wheelchair and a Clearpath Husky). The method has been implemented in the Robot Operating System (ROS) framework and will be publicly released as a ROS package. We also describe and will release several datasets designed to promote the standardized evaluation of similar algorithms.

Pedestrian detection is of crucial importance to autonomous driving applications. Methods based on deep learning have shown significant improvements in accuracy, which makes them particularly suitable for applications, such as pedestrian detection, where reducing the miss rate is very important. Although they are accurate, their runtime has been at best in seconds per image, which makes them not practical for onboard applications. We present a Large-Field-Of-View (LFOV) deep network for pedestrian detection, that can achieve high accuracy and is designed to make deep networks work faster for detection problems. The idea of the proposed Large-Field-of-View deep network is to learn to make classification decisions simultaneously and accurately at multiple locations. The LFOV network processes larger image areas at much faster speeds than typical deep networks have been able to, and can intrinsically reuse computations. Our pedestrian detection solution, which is a combination of a LFOV network and a standard deep network, works at 280 ms per image on GPU and achieves 35.85 average miss rate on the Caltech Pedestrian Detection Benchmark.

The cycling wheelchair is a new type of mobility aid for lower-limb-disabled individuals. Patients can drive the cycling wheelchair by pressing the front pedals with their lower limbs. The device enables patients to effectively rehabilitate their disabled legs and partake in outdoor activities. However, lower-limb-disabled individuals cannot move both legs with equal ease. Typical hemiplegia patients can voluntarily move their healthy leg, but not their disabled leg. Consequently, some of these individuals pedal the cycling wheelchair mainly with their healthy leg, leading to overuse of the limb during their daily activities. This study proposes a pedaling assistive control method that adjusts the different capabilities of the healthy and disabled legs of hemiplegia patients. The proposed control method was applied to a cycling wheelchair with a servo motor for power assist. In experiments, the proposed method reduced the work required for pedaling in two common environments; a planar floor and an upward inclined surface.

The design of finger pads for robotic and prosthetic hands is often overlooked, with relatively simple shapes and mechanical properties typically used. The finger pad geometry and mechanical properties are especially important for within-hand dexterous manipulation, and human finger usage patterns suggest extending robotic finger pad usage onto side surfaces could enable a wider range of manipulation motion. In this work, we propose a novel finger pad design that combines a ridged stiff inner structure with air gaps and a flexible outer skin to facilitate both grasp stability and versatile usage of the finger surface. The air gaps enable objects to displace the outer skin and stably settle between two adjacent ridges. During manipulation, the ridges can also serve as predictable pivot points. Experimental results comparing three ridged finger designs to a conventional solid core design show that the ridged designs consistently outperform the reference solid core design for all objects, in terms of the ability to stably manipulate objects through a large motion range without ejection (losing grip on the object). Designs with larger spaces between ridges performed better overall than designs with closer spacing, showing that larger “wells” allow objects to more stably settle into the space between ridges. We anticipate the novel finger pad designs and the analysis of their behavior will inform future robotic hand designs, especially designs which aim to incorporate side finger usage.

Highly dynamic environments pose a particular challenge for motion planning due to the need for constant evaluation or validation of plans. However, due to the wide range of applications, an algorithm to safely plan in the presence of moving obstacles is required. In this paper, we propose a novel technique that provides computationally efficient planning solutions in environments with static obstacles and several dynamic obstacles with stochastic motions. Path-Guided APF-SR works by first applying a sampling-based technique to identify a valid, collision-free path in the presence of static obstacles. Then, an artificial potential field planning method is used to safely navigate through the moving obstacles using the path as an attractive intermediate goal bias. In order to improve the safety of the artificial potential field, repulsive potential fields around moving obstacles are calculated with stochastic reachable sets, a method previously shown to significantly improve planning success in highly dynamic environments. We show that Path-Guided APF-SR outperforms other methods that have high planning success in environments with 300 stochastically moving obstacles. Furthermore, planning is achievable in environments in which previously developed methods have failed.

This paper describes a control strategy to achieve high fidelity dynamics simulation rendered on admittance controlled robotic facilities. It explores the reasons for an increasing energy found in the virtual dynamics of a free-floating satellite rendered on a six degree of freedom robot, which can lead the system to become unstable and proposes a method to cope with it. The proposed method identifies the sources of intrinsic instability provoked by time delays that are found in the computational loop of the rendered dynamics and counteracts their destabilizing effects using the passivity criteria. The performance of the system and the benefits of the method are shown in simulations and are verified experimentally.

Actuators with adjustable transmission ratios are required in a variety of robotics and automation applications, spanning from human assistive devices and mobile robots to general manipulators. Most existing self-adjustable actuators are bulky and mechanically complex, which often makes their implementation challenging. In this work, we propose a novel passively-adjustable transmission mechanism based on a twisted string actuator. Twisted string actuator is a light, cheap, and mechanically simple actuator, in which twisted strings contract as a result of twisting and therefore act as a translational gear. If one introduces a physical offset between the twisted strings, this changes a ratio between the speed and output force provided by actuator. This work introduces a kinematical model of such actuator for the configuration when a physical offset between the strings is present. In order to experimentally verify the proposed mathematical model, we designed and manufactured a twisted string actuator with variable offsets between the strings. Additionally, in this paper we also propose an idea for passively adjustable gear, controlled by twisted string actuator according to the proposed kinematical model. The main advantages of the proposed passively adjustable gear are its light weight, compliancy and mechanical simplicity, which make it attractive for implementaion in various areas of haptics, teleoperation, wearable and lightweight robotics.

The spring-mass locomotion paradigm is showing great promise as a template for agile and efficient robots. Efficiency and stability are enabled by the passive generation of locomotion patterns, rather than enforced by the control system. However, as leg designs develop more articulation and complexity, a problem arises: how do we implement a chosen set of passive dynamics in complex hardware? We present compliance and impact inertia analyses in a “visually tactile” way, allowing complex and redundant mechanism designs to be easily evaluated. The patterns shown here begin a framework for the comprehensive design of agile, highly dynamic robots.

The paper presents a model for adding stabilizers to a flying device without rotational momentum (such as quadrotors or ornithopters) that will create passively stable vehicles in hover. This model enables the design of the size and location of these stabilizers that will vary the stability and performance of the vehicle. The model is verified with nine experimental vehicles that span the stability design space. Passive stability allows the removal of costly inertial sensors and increases the robustness of the vehicle. Analysis of the cost and drag that impacts flight performance is also discussed.

We present fundamental progress on the computational universality of swarms of micro- or nano-scale robots in complex environments, controlled not by individual navigation, but by a uniform global, external force. Consider a 2D grid world, in which all obstacles and robots are unit squares, and for each actuation, robots move maximally until they collide with an obstacle or another robot. In previous work, we demonstrated components of particle computation in this world, designing obstacle configurations that implement AND and OR logic gates: by using dual-rail logic, we designed NOT, NOR, NAND, XOR, XNOR logic gates. However, we were unable to design a FAN-OUT gate, which is necessary for simulating the full range of complex interactions that are present in arbitrary digital circuits. In this work we resolve this problem by proving unit-sized robots cannot generate a FAN-OUT gate. On the positive side, we resolve the missing component with the help of 2×1 robots, which can create fan-out gates that produce multiple copies of the inputs. Using these gates we are able to establish the full range of computational universality as presented by complex digital circuits. As an example we connect our logic elements to produce a 3-bit counter. We also demonstrate how to implement a data storage element.

For systems in which control authority is shared by an autonomous controller and a human operator, it is important to find solutions that achieve a desirable system performance with a reasonable workload for the human operator. We formulate a shared autonomy system capable of capturing the interaction and switching control between an autonomous controller and a human operator, as well as the evolution of the operator's cognitive state in the working environment. To trade-off human's effort and the performance level, e.g., measured by the probability of satisfying the underlying temporal logic specification, a two-stage policy synthesis algorithm is proposed for generating Pareto efficient coordination and control policies with respect to user specified weights.

In targeted therapy, clusters of drug carriers (nanoparticles and microparticles) could be in contact with a surface such as the lumen of blood vessels and the interior of the gastrointestinal tract. We study the motion characteristics of clusters of microparticles when they slide on a surface under the influence of weak oscillating magnetic fields (less than 11 mT). The oscillating magnetic fields exert a magnetic torque on the microparticles and allow them to oscillate, and hence overcome the static friction and slide on a surface. We characterize the frequency response of clusters of microparticles by applying oscillating magnetic fields with a frequency range of 0 Hz to 55 Hz, in the presence of a constant magnetic field gradient (0.9 T/m). Clusters of 3 to 4 and 5 to 9 microparticles achieve maximum sliding speeds of 1100 µm/s and 1150 µm/s, at oscillating magnetic fields of 30 Hz. In addition, we experimentally demonstrate closed-loop motion control of the clusters with maximum position error of 20 µm. Furthermore, we show that the magnetic field gradient required to drive a cluster of microparticles (with 3 to 4 microparticles) decreases by 75% in the presence of oscillating magnetic fields from 5 Hz to 50 Hz.

For legged devices, their ability of carrying payload is a necessity for a wide range of tasks. In this paper, we present a new approach of carrying payload by using a parallel elastic mechanism, which is able to carry payloads of at least 3 times of its bodyweight. Although the robot has no sensory feedback and consists of only two rigid bodies and one spring loaded joint, it is able to achieve efficient and stable forward hopping for a wide range of attached payload. The presented payload carrier ETH Cargo is based on the further development of our platform CHIARO for the payload range between 0 and 100kg. After parameter optimization using simulations, a series of real world experiments prove stable and high efficiency hopping of the prototype over a wide range of payloads.

We have built a 4DOF tailed monoped that hops along a boom permitting free sagittal plane motion. This underactuated platform is powered by a hip motor that adjusts leg touchdown angle in flight and balance in stance, along with a tail motor that adjusts body shape in flight and drives energy into the passive leg shank spring during stance. The motor control signals arise from the application in parallel of four simple, completely decoupled 1DOF feedback laws that provably stabilize in isolation four corresponding 1DOF abstract reference plants. Each of these abstract 1DOF closed loop dynamics represents some simple but crucial specific component of the locomotion task at hand. We present a partial proof of correctness for this parallel composition of “template” reference systems along with data from the physical platform suggesting these templates are “anchored” as evidenced by the correspondence of their characteristic motions with a suitably transformed image of traces from the physical platform.

We present a novel, deeply embedded robotics middleware and programming environment. It uses a multithreaded, publish-subscribe design pattern and provides a Unix-like software interface for micro controller applications. We improve over the state of the art in deeply embedded open source systems by providing a modular and standards-oriented platform. Our system architecture is centered around a publish-subscribe object request broker on top of a POSIX application programming interface. This allows to reuse common Unix knowledge and experience, including a bash-like shell. We demonstrate with a vertical takeoff and landing (VTOL) use case that the system modularity is well suited for novel and experimental vehicle platforms. We also show how the system architecture allows a direct interface to ROS and to run individual processes either as native ROS nodes on Linux or nodes on the micro controller, maximizing interoperability. Our microcontroller-based execution environment has substantially lower latency and better hardware connectivity than a typical Robotics Linux system and is therefore well suited for fast, high rate control tasks.