ICRA 2015 论文列表

In this paper, we propose a novel method to recover the 3D trajectory of a moving person from a monocular camera mounted on a quadrotor micro aerial vehicle (MAV). The key contribution is an integrated approach that simultaneously performs visual odometry (VO) and persistent tracking of a person automatically detected in the scene. All computation pertaining to VO, detection and tracking runs onboard the MAV from a front-facing monocular RGB camera. Given the gravity direction from an inertial sensor and the knowledge of the individual's height, a complete 3D trajectory of the person within the reconstructed scene can be estimated. When the ground plane is detected from the triangulated 3D points, the absolute metric scale of the trajectory and the 3D map is also recovered. Our extensive indoor and outdoor experiments show that the system can localize a person moving naturally within a large area. The system runs at 17 frames per second on the onboard computer. A walking person was successfully tracked for two minutes and an accurate trajectory was recovered over a distance of 140 meters with our system running onboard.

This paper considers a kinematic model that captures the speed and heading angle of a modular snake robot performing side-winding gaits. Modular snake robot locomotion is controlled in joint space. Consequently, the motion is achieved by continuous changes of the robot's body shape. In light of this behavior, this work describes a method to determine the position and orientation of a floating frame of reference to capture the robot's attitude in real-time, while executing side-winding gaits. A simple velocity model based on ground static contacts that determine the magnitude and direction of the robot's motion is proposed based on that frame. This model is validated by comparisons with experiments. The benefit of the proposed floating frame formulation and the velocity model for side-winding gaits depends on the fact that their calculation relied only in the gait control parameters, hence they could be determined analytically on-line.

It is well known that anthropomorphic manipulators, such as the PA-10, are intolerant to a single locked joint failure of the elbow. This is because the elbow is the only joint that can change the distance between the spherical shoulder joint and the spherical wrist. In this work, it is shown how such arms can be made significantly more fault tolerant by a minor modification to the kinematic structure of the arm. We quantify the degree of fault tolerance to locked joint failures as the minimum of the smallest singular value of the resulting seven Jacobians over all possible single failures. The DH parameters for the modified arm are designed so that the corresponding fault tolerant properties are close to those of a robot with an optimally failure tolerant Jacobian. The fault tolerance of the designed robot is evaluated for two different classes of applications, i.e., point-to-point motions and specified end-effector trajectories.

Optical tweezers has become a powerful tool in automated cell transportation control and has been used in a variety of biological applications. The use of optical tweezers for cell surgery has great potential for various biomedical applications such as microinjection, organelle extraction and modification, and preimplantation genetic diagnosis (PGD). In these cell surgical manipulation tasks, the cell of interest must be oriented properly such that the desired component, e.g., the polar-body or organelles, can be visualized by optical microscopy; thus cell rotation becomes a necessary procedure. Currently, cell rotational control can be carried out by laser tools that are usually handled by skilled people. The open-loop manual operation cannot be readily used for applications requiring precise and high throughput cell rotational control. This highlights the need of developing an automated controlled robot manipulator to rotate biological cells more accurately and efficiently. In this paper, we propose a cell surgery system that utilizes two optical traps, generated by robotically controlled holographic optical tweezers (HOT), to manipulate the cell for rotation, where the optical tweezers functions as special robot manipulators. Through dynamic modeling using T-matrix approach, the relationship between the applied torques and the spherical coordinates of the optical tweezers is characterized. A rotational controller is further developed to rotate the cell to the pre-desired orientation accurately. Experiments are performed to demonstrate the effectiveness of the proposed approach.

Mobility of a hexarotor UAV in its standard configuration is limited, since all the propeller force vectors are parallel and they achieve only 4-DoF actuation, similar, e.g., to quadrotors. As a consequence, the hexarotor pose cannot track an arbitrary trajectory while the center of mass is tracking a position trajectory. In this paper, we consider a different hexarotor architecture where propellers are tilted, without the need of any additional hardware. In this way, the hexarotor gains a 6-DoF actuation which allows to independently reach positions and orientations in free space and to be able to exert forces on the environment to resist any wrench for aerial manipulation tasks. After deriving the dynamical model of the proposed hexarotor, we discuss the controllability and the tilt angle optimization to reduce the control effort for the specific task. An exact feedback linearization and decoupling control law is proposed based on the input-output mapping, considering the Jacobian and task acceleration, for non-linear trajectory tracking. The capabilities of our approach are shown by simulation results.

Successful deployment of a snake robot in search and rescue tasks requires the capability of generating controls which can adapt to unknown environments in real-time. However, available motion generation techniques can be computationally expensive and lack the ability to adapt to the surroundings. This work considers modeling the rolling motion of a snake robot by applying the Bellows model with computation reduction techniques. One benefit of this is that controllers are defined with physically meaningful parameters, which in turn allows for higher level control of the robot. Another benefit is that it allows controllers to be defined by “composing shapes”, which enables developing controllers that can adapt to the surroundings. Using shape composition, we implemented a novel gait, named rolling hump, which forms a contour-fitting hump to negotiate obstacles. The efficacy of a snake robot climbing over obstacles by using the rolling hump is experimentally evaluated. An autonomous control strategy is presented and realized in simulation.

We propose an efficient Human Robot Interaction approach to efficiently model the appearance of all relevant objects in robot's environment. Given an input video stream recorded while the robot is navigating, the user just needs to annotate a very small number of frames to build specific classifiers for each of the objects of interest. At the core of the method, there are several random ferns classifiers that share the same features and are updated online. The resulting methodology is fast (runs at 8 fps), versatile (it can be applied to unconstrained scenarios), scalable (real experiments show we can model up to 30 different object classes), and minimizes the amount of human intervention by leveraging the uncertainty measures associated to each classifier. We thoroughly validate the approach on synthetic data and on real sequences acquired with a mobile platform in outdoor and challenging scenarios containing a multitude of different objects. We show that the human can, with minimal effort, provide the robot with a detailed model of the objects in the scene.

This paper introduces a model-driven approach for engineering complex movement control architectures based on motion primitives, which in recent years have been a central development towards adaptive and flexible control of complex and compliant robots. We consider rich motor skills realized through the composition of motion primitives as our domain. In this domain we analyze the control architectures of representative example systems to identify common abstractions. It turns out that the introduced notion of motion primitives implemented as dynamical systems with machine learning capabilities, provide the computational building block for a large class of such control architectures. Building on the identified concepts, we introduce domain-specific languages that allow the compact specification of movement control architectures based on motion primitives and their coordination respectively. Using a proper tool chain, we show how to employ this model-driven approach in a case study for the real world example of automatic laundry grasping with the KUKA LWR-IV, where executable source-code is automatically generated from the domain-specific language specification.

In this paper we present a mathematical framework to describe the interaction between compliant hands and environmental constraints during grasping tasks. In the proposed model, we considered compliance at wrist, joint and contact level. We modeled the general case in which the hand is in contact with the object and the surrounding environment. All the other contact cases can be derived from the proposed system of equations. We performed several numerical simulation using the SynGrasp Matlab Toolbox to prove the consistency of the proposed model. We tested different combinations of compliance as well as different reference inputs for the hand/arm system considered. This work has to be intended as a tool for compliant hand designer since it allows to tune compliance at different levels before the real hand realization. Furthermore, the same framework can be used for compliant hand simulation in order to study the interaction with the environmental constrains and to plan complex manipulation tasks.

Needle insertions are common during surgical procedures. Accurately delivering the needle at a specific location in the human body is of importance for the clinical outcome of the procedure. Studies have already shown that robotically inserting traditional needles with a bevel tip can improve targeting accuracy. However, steering of such needles requires spinning the needle, which may lead to additional tissue damage. Therefore, we propose a novel design consisting of a flexible needle with a tendon-driven actuated-tip. Changing the orientation of the actuated-tip allows to control the steering direction of the needle and the amount of deflection. We derive the kinematic model which describes the needle path given the actuated-tip orientation based on nonholonomic kinematics. We present a method for steering the needle towards a target location in soft tissue. This method incorporates online parameter estimation in order to adapt for changes in tissue stiffness. Needle insertion experiments are performed in soft-tissue simulants, made from porcine gelatin. Needle tip pose is measured during insertion using Fiber Bragg Grating (FBG) based shape reconstruction. Results show that the needle can be steered towards targets located at 20 mm from the initial insertion axis, at a depth of 100 mm with a mean targeting error of 2.02 mm.

Precise manipulation of microparticles has received considerable attention for its great potential applications to clinical medicine. Among the existing manipulation techniques, the method of magnetic force based manipulation exhibits great advantages for its minimally-invasive feature and insensitivity to biological substance, making it ideally suitable to in vivo environment. On the other hand, increasing demand for accurate and high throughput magnetic manipulation highlights the need of incorporating automation technology in the manipulation. In this paper, we propose an automated control approach to manipulating a magnetic microparticle (bead) with a home-designed electromagnetic coil system. A simplified two-order dynamic model for a microparticle suspended in fluidic environment is established first. A closed-loop controller with utilizing visual feedback is then developed based on input-to-state stability and backstepping methodology. The proposed controller guarantees that the microparticle follows the desired trajectory even in presence of environmental uncertainties and disturbances. Experiments are performed to demonstrate the effectiveness of the proposed approach.

Fast and exact motions of continuum robots are hardly seen so far. Partly this is caused by physical constraints, e.g. small available actuation forces. Another reason is the dynamic coupling between the actuators that cannot be neglected during fast motions. Therefore, a model-based MIMO controller in actuator space was developed, that is based on a spatial dynamic model with one mass point per section. Using feedback linearization, the actuators can be decoupled and feedforward control in combination with linear controllers can be applied. Measurements of an example manipulator show the good tracking result of pure feedforward action with feedback linearization. Adding a linear PD-controller increases the robustness against disturbances without reducing the possibility of fast motions.

This paper describes an automated sewing system that focuses on sewing of curved edge segments. The sewing cell includes an industrial sewing machine as well as three industrial robots that handle the parts during the sewing process. Based on experiences from previous work, several improvements are presented that address issues with previous implementations.

We report on the model-based approach to dynamic trotting and pronking gait generation for a leg-wheel transformable robot. The rolling spring-loaded inverted pendulum (R-SLIP) model served as the template for robot locomotion by programming the robot's motion according to the stable fixed-point trajectories of the model. Two strategies are developed to match the robot leg motions to the virtual model leg: First, using the two active degrees of freedom on each leg to simulate the passive spring effect of the R-SLIP model leg. Second, installing a torsion spring on the leg-wheel module to render the leg-wheel morphology identical to that of the model leg. This model-based approach to dynamic behavior generation for the robot is experimentally evaluated. The robot can successfully generate an R-SLIP-like stable pronking gait with a flight phase.

The objective of this paper is to present the model-based 3D object recognition and fetching by a 7-DoF robot with online obstacle avoidance for factory automation. The robot can fetch the random type of 3D objects with arbitrary pose using a 3D visual camera. Object recognition pipeline based on different descriptors are compared in order to decide the best choice for this scenario. A correct grasp is taught by human operator using programming by touch. As a result, the time of configuring and setting up the robot drastically decreases and thus saves a large amount of preparation time before the robot can actually get to work. Considering the safety issue of the human operators that cooperate with the robot, as well as the uncertainty of the environment such that a moving obstacle might present, or the environment might change for some reason, an online obstacle avoidance algorithm is also integrated into the system. Experimental proof of principles has been successfully implemented.

This paper addresses the problem of controlling underactuated bipedal walking robots in the presence of actuator torque saturation. The proposed method synthesizes elements of the Human-Inspired Control (HIC) approach for generating provably-stable walking controllers, rapidly exponentially stabilizing control Lyapunov functions (RES-CLFs) and standard model predictive control (MPC). Specifically, the proposed controller uses feedback linearization to construct a linear control system describing the dynamics of the walking outputs. The input to this linear system is designed to be the solution of a MPC-based Quadratic Program which minimizes the sum of the values of a RES-CLF-describing the walking control objectives-over a finite-time horizon. Future values of the torque constraints are mapped into the linear control system using the Hybrid Zero Dynamics property of HIC and subsequently incorporated in the Quadratic Program. The proposed method is implemented in a rigid-body dynamics simulation and initial experiments with the Durus robot.

The goal of this paper is to present a real-time controller for an omnidirectional wheeled humanoid robot which can be strongly disturbed and tilt around its wheels. It is based on two linear model predictive controllers, managed by a tilt supervisor, which detects changes of the dynamic model caused by the tilt of the robot. Experimental results are proposed on the robot Pepper made by Aldebaran, showing good performance in term of stability an robustness.

Assembly of large structures requires large fixtures, often referred to as monuments. Their cost and massive size limit flexibility and scalability of the manufacturing process. Numerous small mobile robots can replace these large structures and, therefore, replicate the efficiency of the assembly line with far more flexibility. An assembly line made up of mobile manipulators can easily and rapidly be reconfigured to support scalability and a varied product mix, while allowing for near optimal resource assignment. The challenge to using small robots in place of monuments is making their joint behavior precise enough to accomplish the task and efficient enough to execute subtasks in a reasonable period of time. In this paper, we describe a set of techniques that we combine to achieve the necessary precision and overall efficiency to build a large structure. We describe and demonstrate these techniques in the context of a testbed we implemented for assembling a wing ladder.

In this paper, we present a trajectory planning method to navigate a quadrotor with a cable-suspended payload through known obstacle-filled environments. We model the system as a hybrid dynamical system and formulate the trajectory generation problem as a Mixed Integer Quadratic Program (MIQP). Specifically, we address two novel challenges. First, we plan for a multi-body system, and obstacle avoidance must be guaranteed for the quadrotor, load, and the cable. Second, our method accommodates transitions between subsystems of the hybrid dynamical system, allowing for maneuvers that would otherwise be infeasible if the cable were constrained to remain taut. Numerical and experimental results validate the proposed approach for the full hybrid system.