IROS 2018 论文列表

This contribution presents a motion cueing algorithm (MCA) for driving simulators using nonlinear model predictive control (MPC). The goal of the MCA is to generate a realistic motion feeling while keeping the simulator within its workspace limits. The approach relies on a realtime gradient algorithm in combination with the augmented Lagrangian method in order to directly incorporate the system constraints into the optimization. Simulation results for a reference trajectory with typical driving situations demonstrate the performance as well as the computational efficiency of the approach.

Robotic manipulators that are intended to interact with humans in their operating region are systems that need formal safety guarantees. Current solutions cannot handle both input and state constraints, have difficulties handling nonconvex constraints, or are computationally too expensive. To tackle these drawbacks, we analyzed a constrained control strategy, the Explicit Reference Governor (ERG), which can address both input and state constraints, and does not require any online optimization, thus making it computationally inexpensive. This paper presents the theory of the ERG for a general robotic manipulator and shows simulations for a specific 2DOF planar robotic manipulator. The proposed control scheme is able to steer the robot arm to the desired end-effector position, or an admissible approximation, in the presence of limited joint ranges, actuator saturations, and static obstacles. As a result, the ERG is a promising tool for the control of robotic manipulators subject to constraints.

When robot manipulators decide how to reach for an object, hand it over, or obey some task constraint, they implicitly assume a Euclidean distance metric in their configuration space. Their notion of what makes a configuration closer or further is dictated by this assumption. But different distance metrics will lead to different solutions. What is efficient under a Euclidean metric might not necessarily look the most efficient or natural to a person observing the robot. In this paper, we analyze the effect of the metric on robot behavior, examining both Euclidean, as well as non-Euclidean metrics - metrics that make certain joints cheaper, or that correlate different joints. Our user data suggests that tasks on a 3DOF arm and the Jaco 7DOF arm can typically be grouped into ones where a Euclidean metric works well, and tasks where that is no longer the case: there, surprisingly, penalizing elbow motion (and sometimes correlating the shoulder and wrist) leads to solutions that are more aligned with what users prefer.

The paper presents design, fabrication and characteristics of two kinds of conductive Knit-covered Pneumatic Artificial Muscle (it is called as k-PAM in the paper) actuators, in which two different knits are made by braiding silver-coated (conductive) yarn and spandex (non-conductive) yarn with different stitch methods. The k-PAM is able to measure the change in length of the actuator body according to the applied air pressures as well as the strain due to external force. A complete fabrication method is presented to make the actuator work for higher pressure (≥ 300[kPa]). Since the force generated by the actuator is decoupled from the external force, ultimately, it can be directly used to measure not only the length but also the force. Experimental validations are performed describing the characteristics of two different types of k-PAMs. It is expected that the k-PAM can be used directly for robotic applications in higher pressure condition, while the semi-permanent conductive knit provides the actuator with durability in high repetitive operation environment.

A topological shape analysis is proposed and utilized to learn concepts that reflect shape commonalities. Our approach is two-fold: i) a spatial topology analysis of point cloud segment constellations within objects. Therein constellations are decomposed and described in an hierarchical manner - from single segments to segment groups until a single group reflects an entire object. ii) a topology analysis of the description space in which segment decompositions are exposed in. Inspired by Persistent Homology, hidden groups of shape commonalities are revealed from object segment decompositions. Experiments show that extracted persistent groups of commonalities can represent semantically meaningful shape concepts. We also show the generalization capability of the proposed approach considering samples of external datasets.

Mobile robots have been used for various purposes with different functionalities which require them to freely move in environments containing both static and dynamic obstacles to accomplish given tasks. One of the most relevant capabilities in terms of navigating a mobile robot in such an environment is to find a safe path to a goal position. This paper shows that there exists an accurate solution to the Laplace equation which allows finding a collision-free path and that it can be efficiently calculated for a rectangular bounded domain such as a map which is represented as an image. This is accomplished by the use of the monogenic scale space resulting in a vector field which describes the attracting and repelling forces from the obstacles and the goal. The method is shown to work in reasonably convex domains and by the use of tessellation of the environment map for non-convex environments.

Although workspace is essential for the design and control of cable-driven parallel robots (CDPR) very few works have been devoted to this topic when sagging cables are considered, most probably because of the complexity of the cable model. In this paper we consider suspended CDPR with sagging cables that can support only a limited tension. We propose an algorithm to compute the border of horizontal cross-sections of the workspace for a given altitude and orientation of the platform. We show that singularities of the kinematics equations have to be taken into account for a proper determination of the border and that the workspace can be separated in several components according to the branch of the inverse kinematics on which the robot is evolving. We also compare the workspace obtained for ideal and sagging cables.

In this paper, we devise methods for the multiobjective control of humanoid robots, a.k.a. prioritized whole-body controllers, that achieve efficiency and robustness in the algorithmic computations. We use a form of whole-body controllers that is very general via incorporating centroidal momentum dynamics, operational task priorities, contact reaction forces, and internal force constraints. First, we achieve efficiency by solving a quadratic program that only involves the floating base dynamics and the reaction forces. Second, we achieve computational robustness by relaxing task accelerations such that they comply with friction cone constraints. Finally, we incorporate methods for smooth contact transitions to enhance the control of dynamic locomotion behaviors. The proposed methods are demonstrated both in simulation and in real experiments using a passive-ankle bipedal robot.

This paper presents a method for numerically computing the highest path velocity that a collaborative robot can attain, while complying with safety requirements. The safety requirements are obtained from ISO/TS 15066 that describes a collaborative method called power and force limiting, which specifies safe collisions between humans and robots. In particular, we assume that a path is given and compute the point-wise maximal path velocity that ensures a safe impact, i.e., the paper provides no considerations on the post impact safety.

For a service robot, it is not adequate to let its navigational movement be based only on a single metric, such as minimum distance path. In the environment where the robot and humans are coexisting, the robot should always perform social navigation whenever it is moving. However, to perform social navigation, the robot needs to follow certain “social norms” of the environment. Recently, deep reinforcement learning (DRL) technique is popularly applied to the robotics field; yet, it is rarely used to solve the mentioned social navigation problem, generally deemed as a high dimension complex problem. In this paper, we propose the composite reinforcement learning (CRL) framework under which the robot learns appropriate social navigation with sensor input and reward update based on human feedback. For learning the aspect of human robot interaction (HRI), we provide a method to facilitate the training of DRL in real environment by incorporating prior knowledge to the system. It turns out that our CRL system not only can incrementally learn how to set its velocity and to perform HRI but also keep collecting human feedback to synchronize the reward functions to the current social norms. The experiments show that the proposed CRL system can safely learn how to navigate in the environment and show that our system is able to perform HRI for social navigation.

We present a reinforcement learning algorithm for learning sparse non-parametric controllers in a Reproducing Kernel Hilbert Space. We improve the sample complexity of this approach by imposing a structure of the state-action function through a normalized advantage function (NAF). This representation of the policy enables efficiently composing multiple learned models without additional training samples or interaction with the environment. We demonstrate the performance of this algorithm on learning obstacle-avoidance policies in multiple simulations of a robot equipped with a laser scanner while navigating in a 2D environment. We apply the composition operation to various policy combinations and test them to show that the composed policies retain the performance of their components. We also transfer the composed policy directly to a physical platform operating in an arena with obstacles in order to demonstrate a degree of generalization.

Perception system performance can vary dramatically with contextual factors such as environmental geometry, appearance, and other phenomena. In this work we present a theoretical framework for understanding the role of context in perception and discuss three approaches for predicting probabilistic performance from observations by efficiently learning local performance models. We compare these approaches with experiments on the monocular and stereo visual odometry systems for a ground robot, and show that they can effectively predict system failures in a wide variety of environments.

Mixed reality (MR)opens up new vistas for human-robot interaction (HRI)scenarios in which a human operator can control and collaborate with co-located robots. For instance, when using a see-through head-mounted-display (HMD)such as the Microsoft HoloLens, the operator can see the real robots and additional virtual information can be superimposed over the real-world view to improve security, acceptability and predictability in HRI situations. In particular, previewing potential robot actions in-situ before they are executed has enormous potential to reduce the risks of damaging the system or injuring the human operator. In this paper, we introduce the concept and implementation of such an MR human-robot collaboration system in which a human can intuitively and naturally control a co-located industrial robot arm for pick-and-place tasks. In addition, we compared two different, multimodal HRI techniques to select the pick location on a target object using (i)head orientation (aka heading)or (ii)pointing, both in combination with speech. The results show that heading-based interaction techniques are more precise, require less time and are perceived as less physically, temporally and mentally demanding for MR-based pick-and-place scenarios. We confirmed these results in an additional usability study in a delivery-service task with a multi-robot system. The developed MR interface shows a preview of the current robot programming to the operator, e. g., pick selection or trajectory. The findings provide important implications for the design of future MR setups.

Several approaches are proposed in the literature to calculate the drag force, the electric field and the induced dielectrophoretic force. This paper analyzes the performances of various models for closed loop control of dielectrophoretic systems in comparison with experiments. This article compares their performance in terms of accuracy, computation time, and memory consumption. Four classical approaches are available to calculate the electric field. Their performances are analyzed in the paper. We have shown that combining the dipolar model of dielectrophoresis force with an anisotropic drag force (integrating the wall-effect) provides an interesting ratio precision/computation time. This paper provides an original comparison of several models described in literature whose performances have been compared with experiments.

3D reconstruction and surgical tool segmentation are necessary for several advanced tasks in robot-assisted laparoscopic surgery. These tasks include vision-based force estimation, surgical guidance, and medical image registration where pre-operative data (CT or MRI scan image slices) are overlaid on patient anatomy in real-time during surgery [1] to name a few. In this work, two main strategies were considered: (1) initialize with surgical tool segmentation from 2D images, then proceed to local 3D reconstruction near the tool-tissue interaction region by projecting the segmented result into 3D space, and (2) initialize with 3D reconstruction of the entire surgical task space, followed by surgical tool segmentation from within the 3D reconstructed model. Both methods were implemented on the Raven II surgical robot system, and accuracy and time complexity for both methods were comparatively analyzed while considering various task parameters. Finally, based on the results of this work, guidelines for selecting reconstruction and segmentation strategies and procedure for particular situations are outlined in Section V.

This paper presents an exploration approach for discovering canonical appearances in unknown environments using an autonomous ground robot equipped with a depth sensor. This approach is based on the previously proposed two-stage algorithm that alternates between local and global decision making for efficient topological mapping based on bubble space representation. Differing from it, the approach aims to identify vantage viewpoints with characterizing views for subsequent appearance-based learning as well as achieving complete coverage. This is demonstrated by a series of experiments using an outdoor benchmark data set including a comparative study with evaluation metrics including the exploration path length and number of canonical appearances discovered.

Motion planning for manipulators under task space constraints is difficult as it constrains the joint configurations to always lie on an implicitly defined manifold. It is possible to view task constrained motion planning as an optimization problem with non-linear equality constraints, which can be solved by general non-linear optimization techniques. In this paper, we present a novel custom optimizer which exploits the underlying structure present in many task constraints. At the core of our approach are some simple reformulations, which when coupled with the method of alternating projection, leads to an efficient convex optimization based routine for computing a feasible solution to the task constraints. We subsequently build on this result and use the concept of Augmented Lagrangian to guide the feasible solutions towards those that also minimize the user defined cost function. We show that the proposed optimizer is fully distributive and thus, can be easily parallelized. We validate our formulation on some common robotic benchmark problems. In particular, we show that the proposed optimizer achieves cyclic motion in the joint space corresponding to a similar nature trajectory in the task space. Furthermore, as a baseline, we compare the proposed optimizer with an off-the-shelf non-linear solver provide in open source package SciPy. We show that for similar task constraint residuals and smoothness cost, it can be upto more than three times faster than the SciPy alternative.

This paper is a preliminary exploration of how to solve the dual-manipulator path-planning problem from an energy perspective. A virtual spring is set up between the two manipulator bodies and becomes compressed as they move into the area of danger, thus producing elastic potential energy. The initial paths of the manipulators are modelled as two ends-fixed elastic ropes. The elastic potential energy stored in the virtual spring is distributed between the two elastic ropes in a certain proportion so as to deform them. In this way, the original paths of the two manipulators will deviate toward the direction of their respective bases thereby avoiding any collision crisis that may potentially occur. When the dual-manipulator moves away from the danger area, the elastic potential energy caused by the deviation of the elastic ropes will be converted back into energy stored by the virtual spring, such that the elastic ropes revert to their original state and drive the manipulators back to their initial paths. Simple but representative simulations are established and results of the simulations show the reliability of our proposed method.

We use multiple spinning micro-rafts at the air-water interface as mobile microrobot collectives and present here their collective behaviors, including navigating around anchored obstacles, and trapping and transporting floating objects. The 3D-printed micro-rafts are circular disKS of 100 μm in diameter and have parametrically defined undulating edge profile. The study of their local interactions, manifested by the pairwise interactions between micro-rafts, reveals competing magnetic and capillary interactions that keep the collectives in their dynamic state. Using collectives of 7, 19, and 36 micro-rafts and micro-channels between millimeter-sized posts, we demonstrate the effects of the size of the collectives, the size of the obstacles, and maneuver strategies on the collective navigation. Employing methods from information theory, we show that the pairwise mutual information of the collectives increases significantly during the channel-crossing as a result of the additional constraints of the channel walls on the collectives. Finally, we demonstrate the trapping of 1-mm-diameter polystyrene bead and the trapping and transporting of 600~μm-wide pm.