IROS 2015 Papers

Robotic exoskeletons have drawn much attention recently due to their potential ability to help the stroke and spinal cord injury patients to regain the ability of walking. However, the biggest challenge is the balancing of the exoskeleton and how it can balance is still an open question. Most of the time, patients using such exoskeleton devices require sufficient upper body strength to control upright posture and also manipulate crutches/walking frames to partially support body weight and keep balance. However, high energy cost and the high potential of falling using these devices remains a problem. In this paper, the issues are tackled by virtue of a proposed balance stabilizer mechanism which is able to provide active balance assistance for robotic exoskeletons. The design of a robotic exoskeleton together with balance stabilizer mechanism will be presented and discussed. In addition, a trajectory generation method, which can generate dynamically stable and tunable gait pattern, will also be shown. Finally, clinical trial results with a tetraplegia subject is presented and discussed.

This paper presents the design and control of a teleoperated robotic system for dexterous micromanipulation tasks at the meso-scale, specifically open microsurgery. Robotic open microsurgery is an unexplored yet potentially a high impact area of surgical robotics. Microsurgical operations, such as microanastomosis of blood vessels and reattachment of nerve fibers, require high levels of manual dexterity and accuracy that surpass human capabilities. A 3-DoF robotic wrist is designed and built based on a spherical five-bar mechanism. The wrist is attached to a 3-axis commercial off-the-shelf linear stage, achieving a fully dexterous system. Design requirements are determined using motion data collected during a simulated microanastomosis operation. The wrist design is optimized to maximize workspace and manipulability. The system is teleoperated using a haptic device, and has the required bandwidth to replicate microsurgical motions. The system was successfully used in a micromanipulation task to stack 1 mm-diameter metal spheres. The micromanipulation system presented here may improve surgical outcomes during open microsurgery by offering better accuracy and dexterity to surgeons.

Ball-balancing robots (BBRs) are endowed with rich dynamics. When properly designed and stabilized via feedback to eliminate jitter, and intuitively coordinated with a well-designed smartphone interface, BBRs exhibit a uniquely fluid and organic motion. Unlike mobile inverted pendulums (MIPs, akin to unmanned Segways), BBRs stabilize both fore/aft and left/right motions with feedback, and bank when turning. Previous research on BBRs focused on vehicles from 50cm to 2m in height; the present work is the first to build significantly smaller BBRs, with heights under 25cm. We consider the unique issues arising when miniaturizing a BBR to such a scale, which are characterized by faster time scales and reduced weight (and, thus, reduced normal force and stiction between the omniwheels and the ball). Two key patent-pending aspects of our design are (a) moving the omniwheels to contact the ball down to around 20 to 30 deg N latitude, which increases the normal force between the omniwheels and the ball, and (b) orienting the omniwheels into mutually-orthogonal planes, which improves efficiency. Design iterations were facilitated by rapid prototyping and leveraged low-cost manufacturing principles and inexpensive components. Classical successive loop closure control strategies are implemented, which prove to be remarkably effective when the BBR isn't spinning quickly, and thus the left/right and fore/aft stabilization problems decompose into two decoupled MIP problems.

According to the automobile industry's flexible manufacturing requirements, a novel flexible fixturing system for sheet metal assembly is proposed with parallel robots. A methodology of the structure synthesis is presented by taking account simultaneously several performance indices. Taking the 3UPU/UPS parallel robot in the system as an example, models of inverse kinematics, Jacobian matrix and design space are developed. Thus, the structure synthesis of the parallel robot is simplified to a two-dimensional problem. Once the performance indices (workspace, singularity and stiffness) are established, the corresponding indices atlases are expressed in the design space. The structure parameters of the parallel robot are obtained by analyzing these atlases. The prototype of the fixturing system is developed, and the relevant clamping and stiffness experiments are conducted. The experimental results generally agree well with the simulation results and satisfy the flexible fixturing systems' requirements.

Micro manipulation is essential for tissue indentation tasks in micro-scale experiment, and microscopes must be equipped with multi degree-of-freedom (DOF) stages to accurately reach the regions of interest (ROI). Current microscope stages have three main drawbacks when applied to material characterization tasks. First, most microscope stages are designed for 2- or 3-DOF manipulation, lacking the ability to rotate the specimen. Second, the average micro-positioning stage is manually adjusted, and finally most stages accomplish either a large travel range or sub-micron resolution manipulation, but not both. These limitations pose a challenge for phenotyping diseases using quantitative mechanical signatures. This research is motivated by the need for improved accuracy in breast cancer detection. The following work aims to improve upon current micro-positioning stages by allowing for 4-DOF motion (XYθ for in-plane motion and Z direction indentation) and the ability to employ a variety of devices for material characterization. The ability to rotate the specimen will provide improved accuracy in reaching the ROI and robust measurement in uncertain specimen environments. The work described herein explains the design and analysis toward an under-actuated micro-positioning stage for automated tissue indentation.

In this paper, a rotational leg-type miniature robot with an actuated middle joint and tail, called RoMiRAMT, is proposed for stable locomotion and improved climbing ability. The robot has four independently actuated rotational legs, giving it advantages of both wheel-type and leg-type locomotion. The design parameters of the rotational legs were determined by 3D simulation within the properly selected candidates. It also has unique characteristics inspired by biological structures: a spine and a tail. An actuated middle joint allows the frontal body to be lifted or lowered, which was inspired by a spine, to climb higher obstacles. Additionally, a multi-functional two degrees of freedom (2-DOF) tail was added; the tail provides instant deceleration, and minimized the likelihood of flipping, while allowing the robot to climb higher obstacles. A microcontroller was embedded in the robot, along with a micro-camera and an inertia measurement unit (IMU) sensor. By controlling the robot using the yaw angle signal measured by the IMU sensor, straight movement was enhanced. The body size of the RoMiRAMT is 155 × 80 mm without the rotational legs, and the weight is 593 g including batteries. The maximum velocity of the robot was 2.58 m/s (16.65 body lengths per second) and the maximum height of an obstacle that the robot can climb was 95 mm, which all were verified by experiments.

This paper presents the design of a novel insertable robotic capsule camera system for single incision laparoscopic surgery. This design features a unified mechanism for anchoring, navigating, and rotating an insertable camera by externally generated rotational magnetic field. The design is inspired by the spherical motor concept where the external stator generates anchoring and rotational magnetic field to control the motion of the insertable robotic capsule camera. The insertable camera body, which has no active locomotion mechanism onboard, is capsulated in a one-piece housing with two ringshaped tail-end magnets and one cylindrical central magnet embedded on-board as a rotor. The stator positioned outside an abdominal cavity consists of both permanent magnets and electromagnetic coils for generating reliable rotational magnetic field. The initial prototype results in a compact insertable camera robot with a 12.7mm diameter and a 68mm length. The design concepts are analyzed theoretically and verified experimentally. The experiments validate that the proposed capsule robot design provides reliable camera fixation and locomotion capabilities under various testing conditions.

Peg-in-tube assembly stands ahead of a more common benchmark task for industrial assembly, i.e., ‘pegin-hole’. The robot can easily be deceived to detect the actual hole while performing a ‘peg-in-tube’ task as the tube has a surrounding pocket that cannot support the peg. The paper presents a thorough geometrical analysis of the ‘peg-in-tube’ assembly process, and proposes a novel algorithm based on depth measurements of peg center to perform ‘peg-in-tube’ task. The results are demonstrated on a KUKA KR5 Arc industrial robot with a chamferless cylindrical peg and a tube having a minimum clearance of 0.10 mm.

Accurate real-time person detection is an important capability for many robot tasks, such as indoor navigation and human-robot interaction. In this paper, we introduce a depth-augmented, GPU-accelerated version of Deformable Parts Models (DPM) that uses a joint RGB+Depth feature descriptor to perform high-accuracy person detection at 5Hz while requiring less than 10 Watts on a single 2014 consumer-grade embedded chip. We provide a detailed description of the algorithm and evaluate its speed/accuracy trade-offs on an indoor person detection dataset collected from a mobile platform, showing that our RGBD approach outperforms accuracy of RGB-only DPM, depth-only DPM, and RGB HOG SVM classifier cascades. We furthermore demonstrate how reductions in model complexity and feature space dimensionality can increase speed without significantly sacrificing detector accuracy.

We consider the problem of providing surveillance to a grid area using multiple heterogeneous UAVs, named sentinels and searchers, with complementary sensing and actuation capabilities. We consider probabilistic attacks and we analyze the expected performance with respect to the team deployment. We then introduce the problem of finding minmax deployments that result in the most desirable worst case performance caused by an attack. We present an algorithm to compute deployments while trading off solution's quality and computational effort and we qualitatively and quantitatively analyze it.

This paper presents a methodology to combine information from a sequence of RGB-D spherical views acquired by a home-made multi-stereo device in order to improve the computed depth images both in terms of accuracy and completeness. This methodology is embedded in a larger visual mapping framework aiming to produce accurate and dense topometric urban maps. Our method is based on two main filtering stages. Firstly, we perform a segmentation process considering both geometric and photometric image constraints, followed by a regularization step (spatial-integration). We then proceed to a fusion stage where the geometric information is further refined by considering the depth images of nearby frames (temporal integration). This methodology can be applied to other projective models, such as perspective stereo images. Our approach is evaluated within the frameworks of image registration, localization and mapping, demonstrating higher accuracy and larger convergence domains over different datasets.

Multi-vehicle operations in the ocean are increasingly attractive as more vehicles become commercially available, however acoustics still defines the main practical channel for long-range wireless communication underwater. Long transmission times, scheduling delays and packet losses remain obstacles to implementing real-time multi-vehicle cooperative control for marine vehicles. We present a control system for dynamic pursuit with multiple vehicles exchanging time-difference-of-arrival (TDOA) measurements via underwater acoustic communications. Using data from field experiments conducted with autonomous surface vehicles, we show that a decentralized controller can track fast-moving targets and formation changes, leveraging partial information updates and decreased effective cycle times. As a comparison, we show results for trials conducted with one similar but centralized control scheme.

In this paper, we study the decentralized control problem of a two-agent system under local goal specifications given as temporal logic formulas. The agents collaboratively carry an object in a leader-follower scheme and lack means to exchange messages on-line, i.e., to communicate explicitly. Specifically, we propose a decentralized control protocol and a leader re-election strategy that secure the accomplishment of both agents' local goal specifications. The challenge herein lies in exploiting exclusively implicit inter-robot communication that is a natural outcome of the physical interaction of the robots with the object. An illustrative experiment is included clarifying and verifying the approach.

In this paper, we consider the two dimensional (2-D) predecessor-following control problem for a platoon of unicycle vehicles moving on a planar surface. More specifically, we design a decentralized kinematic control protocol, in the sense that each vehicle calculates its own control signal based solely on local information regarding its preceding vehicle, by its on-board camera, without incorporating any velocity measurements. Additionally, the transient and steady state response is a priori determined by certain designer-specified performance functions and is fully decoupled by the number of vehicles composing the platoon and the control gains selection. Moreover, collisions between successive vehicles as well as connectivity breaks, owing to the limited field of view of cameras, are provably avoided. Finally, an extensive simulation study is carried out in the WEBOTSTM realistic simulator, clarifying the proposed control scheme and verifying its effectiveness.

In this paper we present a decentralised algorithm to solve the Task Allocation problem. Given a set of tasks and a set of robots each with its own utility function, this problem concerns the non-overlapping allocation of tasks to agents maximising the sum of the robot's utility functions. Our algorithm provides a constant factor approximation of (1 - 1/e ≈ 63%) for positive-valued monotone submodular utility functions, and of (1/e ≈ 37%) for positive-valued non-monotone submodular utility functions. In our algorithm, robots only rely on local utility function and neighbour to neighbour communications to find the solution. The algorithm proceeds in two steps. Initially, the robots use Maximum Consensus-based algorithm to find a relaxed solution of the problem using the multilinear extension of their utility functions. Finally, they follow a randomised rounding procedure to exchange valuations on sets of tasks in order to round the relaxed solution. To conclude, we provide experimental evidence of our claims for both monotone and non-monotone submodular functions.

In this paper, two techniques have been presented for slippage detection. They are independent of sensor signal type and are promising for general use on tactile array sensors. The first method is based on frequency analysis of the correlation coefficient sequence of sensor array data sampled as time evolves. The main idea is that a slippage causes heavier fluctuation in the sensor signal distribution and values than a static case. The second approach employs 2-D cross correlation to detect displacements of the sliding object from tactile images, which makes it possible to estimate the slippage velocity using commercially available sensors rather than custom hardware. Experiments have been implemented to evaluate the proposed approaches. It can be seen that the first method is capable of detecting both translational and rotational slippage. Also, it works well in dynamic environments. Additionally, the ability of the second method to detect slippage velocity has been confirmed in the experiment.

This paper proposes a direct monocular SLAM algorithm that estimates a dense reconstruction of a scene in real-time on a CPU. Highly textured image areas are mapped using standard direct mapping techniques [1], that minimize the photometric error across different views. We make the assumption that homogeneous-color regions belong to approximately planar areas. Our contribution is a new algorithm for the estimation of such planar areas, based on the information of a superpixel segmentation and the semidense map from highly textured areas. We compare our approach against several alternatives using the public TUM dataset [2] and additional live experiments with a hand-held camera. We demonstrate that our proposal for piecewise planar monocular SLAM is faster, more accurate and more robust than the piecewise planar baseline [3]. In addition, our experimental results show how the depth regularization of monocular maps can damage its accuracy, being the piecewise planar assumption a reasonable option in indoor scenarios.

We consider discriminative dictionary learning in a distributed online setting, where a team of networked robots aims to jointly learn both a common basis of the feature space and a classifier over this basis from sequentially observed signals. We formulate this problem as a distributed stochastic program with a non-convex objective and present a block variant of the Arrow-Hurwicz saddle point algorithm to solve it. Only neighboring nodes in the communications network need to exchange information, and we penalize the discrepency between the individual feature basis and classifiers using Lagrange multipliers. The application we consider is for a team of robots to collaboratively recognize objects of interest in dynamic environments. As a preliminary performance benchmark, we consider the problem of learning a texture classifier across a network of robots moving around an urban setting where separate training examples are sequentially observed at each robot. Results are shown for both a standard texture dataset and a new dataset from an urban training facility, and we compare the performance of the standard centralized construction to the new distributed algorithm for the case when distinct samples from all classes are seen by the robots. These experiments yield comparable performance between the decentralized and the centralized cases, demonstrating the proposed method's practical utility.

The ability to engage human observers is a key requirement for both social robots and the arts. In this paper, we propose an approach for adapting the Intelligent Adaptive Curiosity learning algorithm to distributed interactive sculptural systems. This Curiosity-Based Learning Algorithm (CBLA) allows the system to learn about its own mechanisms and its surroundings through self-experimentation and interaction. A novel formulation using multiple agents as learning subsets of the system that communicate through shared input variables enables us to scale to a much larger system with diverse types of sensors and actuators. Experiments on a prototype interactive sculpture demonstrate the exploratory patterns of the CBLA and collective learning behaviours through the integration of multiple learning agents.