IROS 2017 论文列表

The constantly increasing amount of machines operating in the vicinity of humans makes it necessary to rethink the design approach for such machines to ensure that they are safe when interacting with humans. Traditional mechanisms are rigid and heavy and as such considered unsuitable, even dangerous when a controlled physical contact with humans is desired. A huge improvement in terms of safe human-robot interaction has been achieved by a radically new approach to robotics - soft material robotics. These new robots are made of compliant materials that render them safe when compared to the conventional rigid-link robots. This undeniable advantage of compliance and softness is paired with a number of drawbacks. One of them is that a complex and sophisticated controller is required to move a soft robot into the desired positions or along a desired trajectory, especially with external forces being present. In this paper we propose an improved soft fluidic rotary actuator composed of silicone rubber and fiber-based reinforcement. The actuator is cheap and easily manufactured providing near linear actuation properties when compared to pneumatic actuators presented elsewhere. The paper presents the actuator design, manufacturing process and a mathematical model of the actuator behavior as well as an experimental validation of the model. Four different actuator types are compared including a square-shaped and three differently reinforced cylindrical actuators.

Continuing research efforts in robot-assisted rehabilitation demand more adaptable and inherently soft wearable devices. A wearable rehabilitative device is required to follow the motion of the body and to provide assistive or corrective motions to restore natural movements. Providing the required level of fluidity in wearable devices becomes a challenge for rehabilitation of more sensitive and fragile body parts, such as the face. To address this challenge, we propose a soft actuation method based on a tendon-driven robotic origami (robogami) and a soft sensing method based on a strain gauge with customized stretchable mesh design. The proposed actuation and sensing methods are compatible with the requirements in a facial rehabilitative device. The conformity of robogamis originates from their multiple and redundant degrees of freedom and the controllability of the joint stiffness, which is provided by adjusting the elasticity modulus of an embedded shape memory polymer (SMP) layer. The reconfiguration of the robogami and the trajectory and directional compliance of its end-effector are controlled by modulating the temperatures, hence the stiffness, of the SMP layers. Here we demonstrate this correlation using simulation and experimental results. In this paper, we introduce a thin and highly compliant sensing method for measuring facial movements with a minimal effect on the natural motions. The measurements of the sensors on the healthy side can be used to calculate the required tendon displacement for replicating the natural motion on the paralyzed side of the face in patients suffering from facial palsy.

We present a real-time algorithm, SocioSense, for socially-aware navigation of a robot amongst pedestrians. Our approach computes time-varying behaviors of each pedestrian using Bayesian learning and Personality Trait theory. These psychological characteristics are used for long-term path prediction and generating proxemic characteristics for each pedestrian. We combine these psychological constraints with social constraints to perform human-aware robot navigation in low- to medium-density crowds. The estimation of time-varying behaviors and pedestrian personalities can improve the performance of long-term path prediction by 21%, as compared to prior interactive path prediction algorithms. We also demonstrate the benefits of our socially-aware navigation in simulated environments with tens of pedestrians.

We present a novel, data-driven framework for planning socially competent robot behaviors in crowded environments. The core of our approach is a topological model of collective navigation behaviors, based on braid groups. This model constitutes the basis for the design of a human-inspired probabilistic inference mechanism that predicts the topology of multiple agents' future trajectories, given observations of the context. We derive an approximation of this mechanism by employing a neural network learning architecture on synthetic data of collective navigation behaviors. Our planner makes use of this mechanism as a tool for interpreting the context and understanding what future behaviors are in compliance with it. The planning agent makes use of this understanding to determine a personal action that contributes to the context in the most clear way possible, while ensuring progress to its destination. Our simulations provide evidence that our planning framework results in socially competent navigation behaviors not only for the planning agent, but also for interacting naive agents. Performance benefits include (1) early conflict resolutions and (2) faster uncertainty decrease for the other agents in the scene.

For robotic vehicles to navigate safely and efficiently in pedestrian-rich environments, it is important to model subtle human behaviors and navigation rules (e.g., passing on the right). However, while instinctive to humans, socially compliant navigation is still difficult to quantify due to the stochasticity in people's behaviors. Existing works are mostly focused on using feature-matching techniques to describe and imitate human paths, but often do not generalize well since the feature values can vary from person to person, and even run to run. This work notes that while it is challenging to directly specify the details of what to do (precise mechanisms of human navigation), it is straightforward to specify what not to do (violations of social norms). Specifically, using deep reinforcement learning, this work develops a time-efficient navigation policy that respects common social norms. The proposed method is shown to enable fully autonomous navigation of a robotic vehicle moving at human walking speed in an environment with many pedestrians.

We develop a reconfigurable legged robot, named Snapbot, to emulate configuration changes and various styles of legged locomotion. The body of Snapbot houses a microcontroller and a battery for untethered operation. The body also contains connections for communication and power to the modular legs. The legs can be attached to and detached from the body using magnetic mechanical couplings. In the center of this coupling, there is a multi-pin spring-loaded electrical connector that distributes power and transmits data between the controller and leg actuators. The locomotion algorithm is implemented on the microcontroller. The algorithm enables Snapbot to locomote in various configurations with one to six legs by recognizing configuration changes and selecting the locomotion method according to the current configuration. Snapbot will be utilized for further research on legged locomotion.

Snake robots have been researched for locomotion in unstructured environments due to its unique and adaptable gaits, however, they have not been used to interact with the environment in a dexterous manner, for example to grasp or push an object. In this paper, the effect of both the configuration of the snake robot (shape) and the friction between the snake robot and ground on a wrench applied to another object are studied. It is assumed that the snake robot has anisotropic friction with the ground, a property that has been proved to be essential for locomotion. The extreme cases of no friction and ideal unbounded friction can be both studied at the same time, making the model and conclusions very versatile and possible to apply to any coefficient of friction. The model and metrics are tested in a study case.

This paper presents a novel multi-step closed-form walking trajectory generator based on the concept of Divergent Component of Motion (DCM) that guarantees smoothness of all resulting reference trajectories. Further, we introduce an analytical method for footstep adjustment to recover from strong disturbances. The DCM trajectory is adjusted to guarantee smoothness of control outputs. Additionally, we present a momentum-based disturbance observer that improves robustness w.r.t. strong continuous perturbations. The proposed methods are verified in simulations.

In this work, we address the problem of centralized scheduling and motion planning for multiple grain carts with finite capacity that serve a group of combine harvesters engaged in a harvesting operation. First, we solve a facility location problem of finding the optimal depot position to minimize the distance traveled by the grain carts. Next, we formulate the scheduling problem as a constraint satisfaction problem. For a rectangular field, we present two schemes which allow the grain carts to unload all the combine harvesters without interrupting the harvesting activity. Finally, for both schemes, we present a relation between the physical parameters of the vehicles that need to be satisfied in order to meet the scheduling constraints.

A Marangoni propulsion is exhibited by some aquatic insects such as rove beetles as a means of rapid escaping by secreting surfactant that lowers the surface tension behind them. Inspired by their locomotion, we proposed a non-tethered miniature robot that can skim and steer on the water surface without employing oscillatory movements of limbs. To make a robot afloat on water surface using the surface tension while attaining higher support force, a circular footpad with an equilateral triangular cross section was used. Other types of cross section shapes such as a circle and a square were also considered; however, their maximal support force obtained from numerical solution was less than an equilateral triangle. The robot had four circular footpads, which coated with superhydrophobic spray, and weighed 18.5 gram including on-board electronics, a battery, and two servo motors. During the experiment, the robot could successfully skim over the water surface by dripping methanol behind and change its moving direction by lowering a small oar either left or right side.

This paper proposes a single-shot approach for recognising clothing categories from 2.5D features. We propose two visual features, BSP (B-Spline Patch) and TSD (Topology Spatial Distances) for this task. The local BSP features are encoded by LLC (Locality-constrained Linear Coding) and fused with three different global features. Our visual feature is robust to deformable shapes and our approach is able to recognise the category of unknown clothing in unconstrained and random configurations. We integrated the category recognition pipeline with a stereo vision system, clothing instance detection, and dual-arm manipulators to achieve an autonomous sorting system. To verify the performance of our proposed method, we build a high-resolution RGBD clothing dataset of 50 clothing items of 5 categories sampled in random configurations (a total of 2,100 clothing samples). Experimental results show that our approach is able to reach 83.2% accuracy while classifying clothing items which were previously unseen during training. This advances beyond the previous state-of-the-art by 36.2%. Finally, we evaluate the proposed approach in an autonomous robot sorting system, in which the robot recognises a clothing item from an unconstrained pile, grasps it, and sorts it into a box according to its category. Our proposed sorting system achieves reasonable sorting success rates with single-shot perception.

In the context of coordination of mobile multi-robot/agent networked teams, we present an integrated model that simultaneously addresses two problems arising in multi-robot missions: (i) task allocation and task scheduling; and (ii) communication provisioning in the multi-hop mobile ad hoc network built by the team. The integrated model is based on a mixed integer linear programming (MILP) formulation, which is solved in a centralized mode. For the communication part, the model solution outputs data routing policies and data transmission schedules that are aimed to maximize data delivery throughput to/from control centers. The trade-off between task and network performance optimization is strategically controlled. A refinement procedure is defined that allows to further improve communications by also minimizing network delays. We report a computational analysis of the integrated MILP model and an evaluation of the impact of a number of parameters on the trade-off between computational load and quality of the output. Results show that the model is computationally affordable for reasonably sized scenarios, and can effectively balance different performance trade-offs.

Hand-eye calibration is a well-known calibration problem. The problem assumes that a camera (eye) is rigidly mounted to the gripper (hand) of a robot arm and aims to find the transformation between them. In this paper, we propose a novel pipeline for hand-eye calibration without the use of a calibration target. First we employ feature extraction and matching, followed by an initial hand-eye calibration step using 2-view matches. In an iterative process, we then alternately employ triangulation and bundle adjustment to optimize the reconstruction and the hand-eye calibration result. Unlike in structure from motion and traditional hand-eye calibration, during this process we always determine the global camera poses using the hand poses and the estimated hand-eye transformation. Synthetic-data and real-data experiments are performed to evaluate the proposed approach, and the results indicate that the accuracy of our approach is superior to state-of-the-art approaches. Moreover, the speed of our algorithm is faster than existing methods.

Robots performing manipulation tasks must operate under uncertainty about both their pose and the dynamics of the system. In order to remain robust to modeling error and shifts in payload dynamics, agents must simultaneously perform estimation and control tasks. However, the optimal estimation actions are often not the optimal actions for accomplishing the control tasks, and thus agents trade between exploration and exploitation. This work frames the problem as a Bayes-adaptive Markov decision process and solves it online using Monte Carlo tree search and an extended Kalman filter to handle Gaussian process noise and parameter uncertainty in a continuous space. MCTS selects control actions to reduce model uncertainty and reach the goal state nearly optimally. Certainty equivalent model predictive control is used as a benchmark to compare performance in simulations with varying process noise and parameter uncertainty.

In this paper we introduce a method for semiautomatic registration of 3D deformable models using 2D shape outlines (silhouettes) extracted from a monocular camera view. Our framework is based on the combination of a biomechanical model of the organ with a set of projective constraints influencing the deformation of the model. To enforce convergence towards a global minimum for this ill-posed problem we interactively provide a rough (rigid) estimation of the pose. We show that our approach allows for the estimation of the non-rigid 3D pose while relying only on 2D information. The method is evaluated experimentally on a soft silicone gel model of a liver, as well as on real surgical data, providing augmented reality of the liver and the kidney using a monocular laparoscopic camera. Results show that the final elastic registration can be obtained in just a few seconds, thus remaining compatible with clinical constraints. We also evaluate the sensitivity of our approach according to both the initial alignment of the model and the silhouette length and shape.

Robotic-assisted minimally invasive surgical systems suffer from one major limitation which is the lack of interaction forces feedback. The restricted sense of touch hinders the surgeons' performance and reduces their dexterity and precision during a procedure. In this work, we present a sensory substitution approach that relies on visual stimuli to transmit the tool-tissue interaction forces to the operating surgeon. Our approach combines a 3D diffeomorphic deformation mapping with a generative model to precisely label the force level. The main highlights of our approach are that the use of diffeomorphic transformation ensures anatomical structure preservation and the label assignment is based on a parametric form of several mixture elements. We performed experimentations on both ex-vivo and in-vivo datasets and offer careful numerical results evaluating our approach. The results show that our solution has an error measure less than 1mm in all directions and an average labeling error of 2.05%. It can also be applicable to other scenarios that require force feedback such as microsurgery, knot tying or needle-based procedures.

The strong resource limitations of unmanned aerial vehicles (UAVs) pose various challenges for UAV applications. In persistent multi-UAV surveillance, several UAVs with limited communication range and flight time have to repeatedly visit sensing locations while maintaining a multi-hop connection to the base station. In order to achieve persistence, the UAVs need to fly back to the base station in time for recharge. However, simple motion planning algorithms can result in mutual movement obstructions of UAVs caused by the constraints. We introduce two planning algorithms with different planning horizons and cooperation and compare their performance in simulation studies. It can be seen that the short horizon uncooperative strategy can outperform other strategies if a sufficient number of UAVs is used. The full horizon strategy can generate a solution visiting all sensing locations if the existence conditions for such a solution are fulfilled.

We introduce a novel approach to shape-based object classification and recognition through the use of a continuum manipulator. Noticing the fact that when a continuum manipulator wraps around an object in a whole-arm grasping, its own shape is indicative of the shape of the object, our approach enables learning and recognition of object classes based on the shapes of continuum wraps. It offers the following advantages: (1) recognition of objects that are not easily detected by vision, such as transparent objects, and (2) highly efficient recognition of such objects of varied sizes due to high-level and rich shape information in each wrap, unlike recognition based on tactile sensing via conventional grasping. Simulation and experiments demonstrate the effectiveness of our approach.

Snake like continuum robots are increasingly used for minimally invasive surgery. Most robotic devices of this sort that have been reported to date are controlled in an open loop manner. Using shape sensing to provide closed loop feedback would allow for more accurate control of the robot's position and, hence, more precise surgery. Fiber Bragg Gratings, magnetic sensors and optical reflectance sensors have all been reported for this purpose but are often limited by their cost, size, stiffness or complexity of fabrication. To address this issue, we designed, manufactured and tested a prototype two-link robot with a built-in fiber-optic shape sensor that can deliver and control the position of a CO2-laser fiber for soft tissue ablation. The shape sensing is based on optical reflectance, and the device (which has a 4 mm outer diameter) is fabricated using 3D printing. Here we present proof-of-concept results demonstrating successful shape sensing - i.e. measurement of the angular displacement of the upper link of the robot relative to the lower link - in real time with a mean measurement error of only 0.7°.