Search

The Navlab group at Carnegie Mellon University has a long history of development of automated vehicles and intelligent systems for driver assistance. The earlier work of the group concentrated on road following, cross-country driving, and obstacle detection. The new focus is on short-range sensing, to look all around the vehicle for safe driving. The current system uses video sensing, laser rangefinders, a novel light-stripe rangefinder, software to process each sensor individually, and a map-based fusion system. The complete system has been demonstrated on the Navlab 11 vehicle for monitoring the environment of a vehicle driving through a cluttered urban environment, detecting and tracking fixed objects, moving objects, pedestrians, curbs, and roads.* [ABSTRACT FROM AUTHOR]

In this session the three participants presented a short overview of what they believe to be key issues and milestones that have brought us to the present. They then give their perspective as to the major issues for the future. Personally, I feel the major lesson from the past is that the notion of general purpose robotic devices proved to be too ambitious, and flawed as a generating principle. Instead, versatile special purpose devices have proved to be the key to successful robot development. I believe this is the major lesson from the past 50 years, and it will continue to be the case into the foreseeable future. [ABSTRACT FROM AUTHOR]

Mimesis theory is one of the primitive skill of imitative learning which is regarded as an origin of human intelligence because imitation is fundamental function for communication and symbol manipulation When the mimesis is adopted as learning method for humanoids loads for designing full body behavior would be decrease because bottom up learning approaches from robot side and top down teaching approaches from user side involved each other. Therefore we propose a behavior acquisition and understanding system for humanoids based on the mime sis theory. This system is able to abstract observed others behaviors into symbols to recognize others behavior using the symbols and to generate motion patterns using the symbols. In this paper we extend the mimesis model to geometric symbol space which contains relative distance information among symbols We also discuss how to generate complex behavior by geometric symbol manipulation in the symbol space and how to recognize novel behavior using combination of symbols by known symbols. [ABSTRACT FROM AUTHOR]

This paper discusses a comprehensive framework for modular motor control based on a recently developed theory of dynamic movement primitives (DMP). DMPs are a formulation of movement primitives with autonomous nonlinear differential equations, whose time evolution creates smooth kinematic control policies. Model-based control theory is used to convert the outputs of these policies into motor commands. By means of coupling terms, on-line modifications can be incorporated into the time evolution of the differential equations, thus providing a rather flexible and reactive framework for motor planning and execution. The linear parameterization of DMPs lends itself naturally to supervised learning from demonstration. Moreover, the temporal, scale, and translation invariance of the differential equations with respect to these parameters provides a useful means for movement recognition. A novel reinforcement learning technique based on natural stochastic policy gradients allows a general approach of improving DMPs by trial and error learning with respect to almost arbitrary optimization criteria. We demonstrate the different ingredients of the DMP approach in various examples, involving skill learning from demonstration on the humanoid robot DB, and learning biped walking from demonstration in simulation, including self-improvement of the movement patterns towards energy efficiency through resonance tuning. [ABSTRACT FROM AUTHOR]

We describe a memory-based approach to learning how to select and provide sub-goals for behavioral primitives, given an existing library of primitives. We demonstrate both learning from observation and learning from practice on a marble maze task, Labyrinth. [ABSTRACT FROM AUTHOR]

Inspired by nature's effective use of tactile feedback for rapid maneuvering, we designed a passive, highly compliant tactile sensor for Sprawlette, a hexapedal running robot. To bridge the gap between biology and design, we took initial steps toward understanding how the cockroach, Periplaneta americana, uses antenna feedback to control its orientation during a rapid wall following behavior. First, we developed a simple template model for antenna-based wall following. Second, we collected initial cockroach data that supports the idea that the rate of convergence to the wall or "tactile flow" is being used, in part, for controlling body orientation. Based on these steps, we designed and calibrated a prototype tactile sensor to measure Sprawlette's angle and distance relative to a straight wall, and employed a simple bio-inspired control law that can stabilize the template dynamics. Finally, we integrated the sensor and controller on Sprawlette and showed empirically that stabilizing Sprawlette during wall following does indeed require tactile flow, as predicted. [ABSTRACT FROM AUTHOR]

In this paper we propose Variable Stiffness actuation [1] as a viable mechanical/control co-design approach for guaranteeing control performance for robot arms that are inherently safe to humans in their environment. A new actuator under development in our Lab is then proposed, which incorporate the possibility to vary transmission stiffness during motion execution, thus allowing substantial motion speed-up while maintaining low injury risk levels. [ABSTRACT FROM AUTHOR]

This paper discusses the 100G Capturing Robot that can produce the maximum acceleration of 100G in design specification. To achieve such a high acceleration, we utilize spring energy with a light arm/gripper. The Arm/Gripper Coupling Mechanism (AGCM) is newly introduced so that we can efficiently transmit a single energy resource for finally closing the gripper. Experimental results show that the developed robot can capture a dropping ball with the maximum acceleration of 91G and the capturing time of roughly 25ms. We also discuss various technical issues to be considered when designing such a high speed capturing robot. [ABSTRACT FROM AUTHOR]

This paper presents a micro-actuator that operates free of any physically restraining tethers. We show how capacitive coupling can be used to deliver power to untethered MEMS (micro-electromechanical systems) devices, independently of their position and orientation. Our novel power delivery and actuation mechanisms are designed for use in autonomous mobile robots whose dimensions can be measured in tens to hundreds of micrometers. Test devices utilizing these mechanisms have been fabricated at scale using MEMS technology, and have been shown capable of untethered locomotion at speeds exceeding 1.5 mm/sec. The corresponding speed, scaled to a car-sized robot would be over 100 km/hr. The possibility of autonomous (untethered) microactuators and microrobots less than 80 μm in length opens the door to novel applications in distributed and parallel robotics. Scratch-drive actuators (SDAs) [1,2] are capable of both high speed and nanometer-scale precision, making them a promising choice for the actuators of mobile microrobots. We developed a novel kind of SDA that walks without the physically-restraining electrical tethers, rails, stators, or springs required in previous work. Our power delivery mechanism operates through a capacitive coupling with a silicon substrate that underlies the mobile device. This allows the device actuation mechanism to use electrostatic forces without requiring any physical connection or wire that would restrain the motion of the device. Through appropriate design of the substrate wiring and actuator geometries, the power delivered to the device can be shown to be independent of the device's position and orientation. For this reason, this power delivery mechanism is appropriate for fully two-dimensional (x, y) and (x, y, θ) mobile microrobots. [ABSTRACT FROM AUTHOR]

Human physical interaction is inherently multisensory, using vision, touch, hearing, and other senses. We discuss the constraints and opportunities for constructing computational models for multisensory interaction. Such models could be used to develop better robots and human interfaces, and to understand human interaction. [ABSTRACT FROM AUTHOR]

This paper proposes a new model for machine perception in natural outdoor environments. The goal is to find a method of describing and fusing information obtained from a diverse array of sensors, and to provide a robust interpretation of unstructured outdoor scenes, primarily for the purpose of autonomous navigation. The model has three key components: First,it is based around a high- dimensional sensor-centric description of the environment and deliberately avoids the fragile process of feature extraction from any one sensor. Second, the description is itself embedded in a probabilistic structure allowing Bayes Theorem to be used for temporal fusion of information. Finally, a task-directed process of nonlinear dimensionality reduction, abstraction or compression is used to identify low-dimensional high-contrast features embedded in the high-dimensional sensor space. Such features are the "sensors own view" of what constitutes important information for the task at hand. The paper provides a number of preliminary experimental results of applying this model to combinations of mm-wave radar, laser and night-vision sensors. [ABSTRACT FROM AUTHOR]

This paper describes anew solution for stable haptic interaction with deformable object simulations featuring low servo rates and computational delays. The proposed solution is a combination of the local model and the virtual coupling concepts proposed in the past. More specifically we propose a local model structure that adapts to the deformable object local geometry and impedance characteristics. By doing so we are able to run a haptic interaction loop at high update rates,thus maximizing the Z-width of the simulation, while obtaining overall stable behavior without using any added damping. Furthermore the local model is always computed using the current deformable object surface, thus allowing for real multi-point contact interaction. The proposed solution is presented and analyzed in a multi-rate setting. Some experimental results employing a PHANTOM haptic interface are proposed for a simple one-dimensional example. [ABSTRACT FROM AUTHOR]

We describe a new tactile display for use in dexterous telemanipulation and virtual reality. Our system renders the changing location of a remote or virtual contact by moving a tactile element along the user's fingertip. Mounted at the endpoint of a haptic mechanism, our thimble-sized device concurrently displays contact location and interaction forces. We believe such a design will enable more versatile object manipulation for haptic interactions. To evaluate this display concept, we conducted two perceptual experiments. First, human subjects judged object curvature though direct manipulation of physical models and virtual manipulation with the device. Results show similar levels of discrimination in real and virtual interactions, indicating the device can effectively portray contact information. Second, we investigated virtual interactions with rolling and anchored objects and demonstrated that users can distinguish the interaction type using our device. These experiments provide insight into the sensitivity of human perception and suggest that even a simple display of the contact centroid location may significantly enhance telerobotic and virtual grasping tasks. [ABSTRACT FROM AUTHOR]

In this paper a hierarchical multi-resolution approach allowing for high precision and distinctiveness is presented. The method combines topological and metric paradigm. The metric approach, based on the Kalman Filter, uses a new concept to avoid the problem of the drift in odometry. For the topological framework the fingerprint sequence approach is used. During the construction of the topological map, a communication between the two paradigms is established. The fingerprint used for topological navigation enables also the re-initialization of the metric localization. The experimentation section will validate the multi-resolution-representation maps approach and presents different steps of the method. [ABSTRACT FROM AUTHOR]

This paper presents our research group's latest results in autonomous force-controlled manipulation tasks: (i) advanced non-linear estimators for simultaneous parameter estimation and contact formation "map building" for 6D contact tasks (with active sensing integrated into the task planner), and (ii) the application of these results to programming by human demonstration, for tasks involving contacts. [ABSTRACT FROM AUTHOR]

We present a lazy data association algorithm for the simultaneous localization and mapping (SLAM) problem. Our approach uses a tree-structured Bayesian representation of map posteriors that makes it possible to revise data association decisions arbitrarily far into the past. We describe a criterion for detecting and repairing poor data association decisions. This technique makes it possible to acquire maps of large-scale environments with many loops, with a minimum of computational overhead for the management of multiple data association hypotheses. A empirical comparison with the popular FastSLAM algorithm shows the advantage of lazy over proactive data association. [ABSTRACT FROM AUTHOR]

This paper applies a new constant-time, consistent and convergent Simultaneous Localization and Mapping (SLAM) algorithm to synthetic aperture sonar (SAS) data acquired by an autonomous underwater vehicle (AUV). Using a novel target detection strategy, data gathered from a 40 minute survey is processed and the results compared to both a ground truth and the "gold standard" quadratic time full covariance SLAM algorithm. [ABSTRACT FROM AUTHOR]

This paper provides an alternative solution to solving SLAM's computational complexity in Inertial Navigation System (INS) application, not from the perspective of map management techniques, but by focusing on the filter's structure and model, and re-casting the SLAM algorithm into what is known as an "indirect" implementation. In doing so we provide a navigation structure which is computationally efficient in even for highly non-linear, highly dynamic systems. [ABSTRACT FROM AUTHOR]

Dexterity of human behavior arises from the ability to discover and ex-ploit so-called "knacks", or the essential reference frames (coordinates in the phase space) for successful control under unknown perturbations. This strategy differs from the traditional robot control in that it only partially specifies/controls the phase space trajectory, pushing the system into a desired natural dynamics, letting it do the job. As the first example of such a strategy, we investigated a "roll-and-rise" motion, a fully dynamic whole-body motion with unpredictable perturbations due to multiple contacts with the ground. The "knacks" of the motion are identified and analyzed by a systematic search of the phase space. A successful experiment with an adult-size whole body humanoid robot is presented; it rose from a flat-lying posture to a crouching posture within 2 seconds. [ABSTRACT FROM AUTHOR]

This paper present a humanoid robot, called HRP-2, that can walk, lie down and get up. We believe that HRP-2, 154[cm] height an 58[kg] weight, is the first humanoid robot that has the human-size and can lie down and get up. The biped walking is realized by a walking pattern generator using a preview control and by a feedback control. The lying down and getting up motions were made possible by HRP2's compact body with a waist joint and a novel motion controller. [ABSTRACT FROM AUTHOR]

Humanoid robotics hardware and control techniques have advanced rapidly during the last five years. Presently, several companies have announced the commercial availability of various humanoid robot prototypes. In order to improve the autonomy and overall functionality of these robots, reliable sensors, safety mechanisms, and general integrated software tools and techniques are needed. We believe that the development of practical motion planning algorithms and obstacle avoidance software for humanoid robots represents an important enabling technology. This paper gives an overview of some of our recent efforts to develop motion planning methods for humanoid robots for application tasks involving navigation, object grasping and manipulation, footstep placement, and dynamically-stable full-body motions. We show experimental results obtained by implementations running within a simulation environment as well as on actual humanoid robot hardware. [ABSTRACT FROM AUTHOR]

In this paper we describe the autonomous behavior control architecture of SDR-4X II, which serves to integrate multi-modal recognition and motion control technologies. We overview the entire software architecture of SDR-4X II, which is composed of perception, short and long term memory, behavior control, and motion control parts. Regarding autonomous behavior control, we further focus on issues such as spontaneous behavior generation using a homeostasis regulation mechanism, and a behavior control/selection mechanism with tree-structured situated behavior modules. In the autonomous behavior control architecture, we achieve three basic requirements, which are the concurrent evaluation of the situation of each behavior module, concurrent execution of multiple behavior modules, and preemption (behavior interruption/resume capability). Using the autonomous behavior control architecture described, we demonstrate that SDR-4X II can spontaneously and passively interact with a human. [ABSTRACT FROM AUTHOR]

Current vision systems are designed to perform in clear weather. Needless to say, in any outdoor application, there is no escape from bad weather. Ultimately, computer vision systems must include mechanisms that enable them to function (even if somewhat less reliably) in the presence of haze, fog, rain, hail and snow. We begin by studying the visual manifestations of different weather conditions. For this, we draw on what is already known about atmospheric optics, and identify effects caused by bad weather that can be turned to our advantage; we are not only interested in what bad weather does TO vision but also what it can do FOR vision. [ABSTRACT FROM AUTHOR]

This paper reviews aspects of the design and construction of an active wearable camera, and describes progress in equipping it with visual processing for reactive tasks like orientation stabilisation, slaving from head movements, and 2D tracking. The paper goes on to describe a first application of frame-rate simultaneous localisation and mapping (SLAM) to the wearable camera. Though relevant for any single camera undergoing general motion, the approach has particular benefits in wearable vision, allowing extended periods of purposive fixation followed by controlled redirection of gaze to other parts of the scene. [ABSTRACT FROM AUTHOR]

This paper presents a simultaneous localization and mapping approach based on an extended kalman filter, using only a set of non-registered stereovision image pairs. Invariant image point features are used to detect 3D landmarks and to provide local motion estimates. The 3D coordinates of each point feature are computed by stereovision, and an interest point matching algorithm solves the data association problem. The estimation of the errors on the local motion estimations, on the landmark initialisations and on the landmark observations are depicted, and results with low altitude aerial images are presented. [ABSTRACT FROM AUTHOR]

The use of visual sensors may have high impact in robotic applications where it is required to measure the pose (position and orientation) and the visual features of objects moving in unstructured environments. In this paper, the problem of real-time estimation of the position and orientation of of multiple objects is considered. Special emphasis is devoted to the case when two or more objects overlap with respect to the visual system causing occlusion. The algorithm is based on the Kalman filtering and Binary Space Partition (BSP) tree representations of the objects geometry. The real-time implementation of the algorithm is experimentally tested for the case of visual tracking of two objects using two cameras. [ABSTRACT FROM AUTHOR]

Autonomous off-road navigation is central to several important applications of unmanned ground vehicles. This requires the ability to detect obstacles in vegetation. We examine the prospects for doing so with scanning ladar and with a linear array of 2.2 GHz micro-impulse radar transceivers. For ladar, we summarize our work to date on algorithms for detecting obstacles in tall grass with single-axis ladar, then present a simple probabilistic model of the distance into tall grass that ladar-based obstacle detection is possible. This model indicates that the ladar "penetration depth" can range from on the order of 10 cm to several meters, depending on the plant type. We also present an experimental investigation of mixed pixel phenomenology for a time-of-flight, SICK ladar and discuss briefly how this bears on the problem. For radar, we show results of applying an existing algorithm for multi-frequency diffraction tomography to a set of 45 scans taken with one sensor translating laterally 4 cm/scan to mimic a linear array of transceivers. This produces a high resolution, 2-D map of scattering surfaces in front of the array and clearly reveals a large tree trunk behind over 2.5 m of thick foliage. Both types of sensor warrant further development and exploitation for this problem. [ABSTRACT FROM AUTHOR]

We present a newapproach to developing hybrid feedback policies for the control of systems with nonholonomic constraints. We extend the idea of sequential composition and use it to to switch between controllers in a state based manner, resulting in a globally convergent, pure feedback policy. Individual controllers in the palette are inspired by variable constraint control. We solve a vision guided unicycle navigation problem and present experimental results. [ABSTRACT FROM AUTHOR]

We present an algorithm for the multi-robot simultaneous localization and map-ping (SLAM) problem. Our algorithm enables teams of robots to build joint maps, even if their relative starting locations are unknown and landmarks are ambiguous—which is presently an open problem in robotics. It achieves this capability through a sparse information filter technique, which represents maps and robot poses by Gaussian Markov random fields. The alignment of local maps into a single global maps is achieved by a tree-based algorithm for searching similar-looking local landmark configurations, paired with a hill climbing algorithm that maximizes the overall likelihood by search in the space of correspondences. We report favorable results obtained with a real-world benchmark data set. [ABSTRACT FROM AUTHOR]

Future planetary exploration missions will use cooperative robots to explore and sample rough terrain. To succeed robots will need to cooperatively acquire and share data. Here a cooperative multi-agent sensing architecture is presented and applied to the mapping of a cliff surface. This algorithm efficiently repositions the systems' sensing agents using an information theoretic approach and fuses sensory information using physical models to yield a geometrically consistent environment map. This map is then distributed among the agents using an information based relevant data reduction scheme. Experimental results for cliff face mapping using the JPL Sample Return Rover (SRR) are presented. The method is shown to significantly improve mapping efficiency over conventional methods. [ABSTRACT FROM AUTHOR]

This paper introduces the application of a sensor network to navigate a flying robot. We have developed distributed algorithms and efficient geographic routing techniques to incrementally guide one or more robots to points of interest based on sensor gradient fields, or along paths defined in terms of Cartesian coordinates. The robot itself is an integral part of the localization process which establishes the positions of sensors which are not known a priori. We use this system in a large-scale outdoor experiment with Mote sensors to guide an autonomous helicopter along a path encoded in the network. [ABSTRACT FROM AUTHOR]

This paper addresses the design of simple robot behaviors that realize emergent group behaviors. We present a method to coordinate a large number of under-actuated robots by designing control laws on a small dimensional abstraction manifold, independent of the number and ordering of the robots. The abstraction manifold has a product structure consisting of elements of a Lie group that capture the position and orientation of the ensemble in the world frame, and elements of a shape manifold that provide an intrinsic description of the distribution of team members relative to one another. We design decoupled controls for regulating the group and the shape variables. The realization of the controller on each robot requires the feedback of the robot state, and the state on the abstraction manifold. We present experimental results with a team of five car-like robots equipped with omnidirectional cameras and IEEE 802.11b networking. [ABSTRACT FROM AUTHOR]

This session's papers deal with the coordination and integration of information from multiple robots to map and navigate their environments. [ABSTRACT FROM AUTHOR]

Robot-Robot and Human-Robot cooperation technologies are about to mature in commercial robotics applications. In this paper we outline KUKA's cooperating robot technology and show its potential benefit in automotive production. A major step in demonstrating human-robot coexistence has been achieved by the KUKA RoboCoaster, a robotic entertainment system. Further steps are discussed by showing potential technologies for safeguarded human-robot interaction in production. Keywords robot-robot cooperation, human-robot cooperation, entertainment robotics, robotics safety, capacitive proximity sensors, haptic devices. [ABSTRACT FROM AUTHOR]

This paper introduces a novel approach for face recognition using multiple face patterns obtained in various views for robot vision. A face pattern may change dramatically due to changes in the relation between the positions of a robot, a subject and light sources. As a robot is not generally able to ascertain such changes by itself, face recognition in robot vision must be robust against variations caused by the changes. Conventional methods using a single face pattern are not capable of dealing with the variations of face pattern. In order to overcome the problem, we have developed a face recognition method based on the constrained mutual subspace method (CMSM) using multi-viewpoint face patterns attributable to the movement of a robot or a subject. The effectiveness of our method for robot vision is demonstrated by means of a preliminary experiment. [ABSTRACT FROM AUTHOR]

Unlike conventional haptic devices, an encountered-type device is not held by a user all the time. Instead, the device stays at the location of a virtual object and waits for the user to encounter it. In this paper, we extend this concept to fingertip contacts and design an encountered-type haptic display for multiple fingertip contacts to simulate tasks of grasping an object with any shape and size. Before designing the device, we intensively observed human grasping behaviors. This observation was very helpful to determine the mechanism of the device. An encountered-type device for three-fingered grasping was actually prototyped based on our design. [ABSTRACT FROM AUTHOR]

One of the biggest obstacles facing humans and robots is the lack of means for natural and meaningful interaction. Robots find it difficult to understand human intentions since our way of communication is different from the way machines exchange their information. Our aim is to remove this barrier by creating systems that react and respond to natural human actions. In this research, we present a robotic system that identifies and picks up an arbitrary object in 3D space that a person is looking at. It is done through an active vision system that is able to understand the focus of attention of a user. Whenever the person is close enough, the gaze direction is determined and used to find the object of interest which is of unknown size, shape and color. A robot arm responds by picking up this object and handing it over to the person. [ABSTRACT FROM AUTHOR]

We describe our current progress in developing Human-Machine Collaborative Systems (HMCSs) for microsurgical applications such as vitreo-retinal eye surgery. Three specific problems considered here are (1) developing of systems tools for describing and implementing an HMCS, (2) segmentation of complex tasks into logical components given sensor traces of a human performing the task, and (3) measuring HMCS performance. Our goal is to integrate these into a full microsurgical workstation with the ability to automatically "parse" traces of user execution into a task model which is then loaded into the execution environment, providing the user with assistance using online recognition of task state. The major contributions of our work to date include an XML task graph modeling framework and execution engine, an algorithm for real-time segmentation of user actions using continuous Hidden Markov Models, and validation techniques for analyzing the performance of HMCSs. [ABSTRACT FROM AUTHOR]

Grenoble was one of the first places worldwide where a robot was used in clinical routine. In 1989, the first robot-assisted procedure on a patient took place in the neurosurgery department of the Grenoble University Hospital. An industrial robot modified to satisfy to the clinical constraints was used as a positioning device for guiding minimally invasively the surgical tool to a planned target. Based on that experience and on our knowledge of the clinical domain, we progressively re-directed our activity towards the design of specific robotic devices answering important issues among which safety, interactivity and clinical suitability. In this paper, we describe in more details our view of what surgical robotics is and should be and we illustrate our approach by the description and discussion of research in progress at TIMC: the PADyC arm, a passive device constraining the surgeon's motions in function of a pre-planned surgical protocol - the TER system a non rigid and portable robot for tele-echography - and LER a portable endoscope holder. We will discuss the specificities of medical robotics. Finally, we will draw the perspectives that we foresee for this domain. [ABSTRACT FROM AUTHOR]

The execution of teleoperated tasks is based on the visual feedback of the working scene through TV cameras located adequately in the working environment. Common problems in many applications are the lack of visibility produced by the task conditions and poor dexterity. With the aim of increasing the users' perception capabilities, as well as, their precision ability, a set of teleoperation aids have been studied and developed. These aids are based on visual information and other sensor data or on some knowledge about the working environment. [ABSTRACT FROM AUTHOR]

Many modern applications of technology to medicine, such as robotic surgery, non-rigid registration, virtual reality operation planning and surgeon training systems, and surgical simulators, require knowledge of mechanical properties of very soft tissues. In this paper we describe and identify a model of mechanical properties of brain tissue aimed in particular at integration with interactive brain atlases. A non-linear, viscoelastic model based on the generalization of the Ogden strain energy hyperelastic constitutive equation is proposed. The material parameters are identified using in-vitro experimental results. The model accounts well for brain tissue deformation behavior in both tension and compression (natural strain $\in\langle -0.3, 0.2\rangle$) for strain rates typical for surgical procedures. It can be immediately applied in large-scale finite element simulations and, therefore, offers the possibility of incorporating mechanics into surgical planning and training systems such as NeuroPlanner and BrainBench. Finally we show that the brain model identified based on in-vitro experiments can be applied in the more realistic in-vivo setting. [ABSTRACT FROM AUTHOR]

Recent advances in molecular biology such as cloning demonstrate that increasingly complex micromanipulation strategies for manipulating individual biological cells are required. From a robotics standpoint, the manipulation of biological cells, sometimes referred to as biomanipulation, presents several interesting research issues that extend well beyond cell manipulation. Biological cells are highly deformable objects, and the material properties of these objects are not well quantified, so developing strategies for manipulating deformable objects must be addressed. Most biological cells are between 1μm and 100μm in diameter, depending on the cell type, so micromanipulation issues must be explored, including the appropriate use of high resolution, low depth-of-field vision feedback and very low magnitude multi-axis force feedback. By pursuing robotic manipulation of biological cells, many interesting robotics research avenues in micromanipulation, deformable object handling, multi-sensor integration, and force and vision feedback assimilation must be explored. This paper explores the visual tracking of biological cells using physics-based models and the measurement of applied force fields using a new cell deformation model with visual feedback. A multi-axis MEMS-based force sensor is used to determine applied forces and develop models of cell deformation. Robust tracking of cell deformation is shown and real-time determination of applied force fields is demonstrated. In addition, the system developed has been used to quantitate for the first time a phenomenon known as "zona hardening" during mouse oocyte fertilization. [ABSTRACT FROM AUTHOR]

We propose an original approach to integrate symbolic task planning, and geometric motion and manipulation planning. We focus more particularly on one key aspect: the relation between the symbolic positions and their geometric counterparts. Indeed, we have developed an instantiation process that is able to propagate incrementally task-dependent as well as 3D environment-dependent constraints and to guide efficiently the search until valid geometric configurations are found that satisfy the plan at both levels. The overall process is discussed and illustrated through an implemented example. [ABSTRACT FROM AUTHOR]

Many motion planning, artificial intelligence, and mobile robot planning techniques exploit topology, as encoded in the free space, to aid in generating concise maps and efficient planners. Prior work of the authors has used the generalized Voronoi diagram, which is a topological map that has a natural embedding into the free space. Topological maps have been used in the AI community, starting with Ben Kuipers insightful work on distinctive places and edges that connected them (although he used a Voronoi diagram-like structure also). In this paper, we put forth a more rigorous definition of a topological map, and then overview problems that exploit a topological map to achieve tasks including path planning, SLAM, and hybrid controls. [ABSTRACT FROM AUTHOR]

We propose the combination of techniques that solve multiple queries for motion planning problems with single query planners in a motion planning framework that can be efficiently parallelized. In multiple query motion planning, a data structure is built during a preprocessing phase in order to quickly respond to on-line queries. Alternatively, in single query planning, there is no preprocessing phase and all computations occur during query resolution. This paper shows how to effectively combine a powerful sample-based method primarily designed for multiple query planning (the Probabilistic Roadmap Method - PRM) with sample-based tree methods that were primarily designed for single query planning (such as Expansive Space Trees, Rapidly Exploring Random Trees, and others). Our planner, which we call the Probabilistic Roadmap of Trees (PRT), uses a tree algorithm as a subroutine for PRM. The nodes of the PRM roadmap are now trees. We take advantage of the very powerful sampling schemes of recent tree planners to populate our roadmaps. The combined sampling scheme is in the spirit of the non-uniform sampling and refinement techniques employed in earlier work on PRM. PRT not only achieves a smooth spectrum between multiple query and single query planning but it combines advantages of both. We present experiments which show that PRT is capable of solving problems that cannot be addressed efficiently with PRM or single-query planners. A key advantage of PRT is that it is significantly more decoupled than PRM and sample-based tree planners. Using this property, we designed and implemented a parallel version of PRT. Our experiments show that PRT distributes well and can easily solve high dimensional problems that exhaust resources available to single machines. [ABSTRACT FROM AUTHOR]

Many instances of the mobile robot guidance mapping problem exhibit a topology that can be represented as a graph of nodes (observations) connected by edges (poses). We show that a cycle basis of this pose network can be used to generate the independent constraint equations in a natural constrained optimization formulation of the mapping problem. Explicit reasoning about the loop topology of the network can automatically generate such a cycle basis in linear time. Furthermore, in many practical cases, the pose network has sparse structure and the associated equations can then be solved in time linear in the number of images. This approach can be used to construct globally consistent maps on very large scales in very limited computation. While the technique is applicable to mapbuilding in general, and even optimization in general, it is illustrated here for batch processing of 2D ladar scans into a mobile robot guidance map. [ABSTRACT FROM AUTHOR]

We present an application of hierarchical Bayesian estimation to robot map building. The revisiting problem occurs when a robot has to decide whether it is seeing a previously-built portion of a map, or is exploring new territory. This is a difficult decision problem, requiring the probability of being outside of the current known map. To estimate this probability, we model the structure of a "typical" environment as a hidden Markov model that generates sequences of views observed by a robot navigating through the environment. A Dirichlet prior over structural models is learned from previously explored environments. Whenever a robot explores a new environment, the posterior over the model is estimated using Dirichlet hyperparameters. Our approach is implemented and tested in the context of multi-robot map merging, a particularly difficult instance of the revisiting problem. Experiments with robot data show that the technique yields strong improvements over alternative methods. [ABSTRACT FROM AUTHOR]