IROS 2019 论文列表

Inter-process communication (IPC) is one of the core functions of modern robotics middleware. We propose an efficient IPC technique called TZC (Towards Zero-Copy). As a core component of TZC, we design a novel algorithm called partial serialization. Our formulation can generate messages that can be divided into two parts. During message transmission, one part is transmitted through a socket and the other part uses shared memory. The part within shared memory is never copied or serialized during its lifetime. We have integrated TZC with ROS and ROS2 and find that TZC can be easily combined with current open-source platforms. By using TZC, the overhead of IPC remains constant when the message size grows. In particular, when the message size is 4MB (less than the size of a full HD image), TZC can reduce the overhead of ROS IPC from tens of milliseconds to hundreds of microseconds and can reduce the overhead of ROS2 IPC from hundreds of milliseconds to less than 1 millisecond. We also demonstrate the benefits of TZC by integrating it with TurtleBot2 to be used in autonomous driving scenarios. We show that by using TZC, the braking distance can be 16% shorter than with ROS.

Balancing or postural stability is one of important locomotor subfunctions in bipedal gaits. The inverted pendulum and virtual pivot point (VPP) are common modeling approaches to analyze balance control in human and robot walking. In this paper, we employ the VPP concept to investigate posture control in the frontal plane. The outcomes demonstrate that unlike posture control in the sagittal plane, the VPP in the frontal plane is place below center of mass. This finding explains a novel hybrid strategy for lateral stability in human walking. The here proposed model shows that switching between unstable inverted virtual pendulums generate stable posture control in the frontal plane. This outcome is consistent within a group of seven human subjects walking at normal and slow speeds.

Until now there have been few formalized methods for conducting systematic benchmarking aiming at reproducible results when it comes to computer vision algorithms. This is evident from lists of algorithms submitted to prominent datasets, authors of a novel method in many cases primarily state the performance of their algorithms in relation to a shallow description of the hardware system where it was evaluated. There are significant problems linked to this non-systematic approach of reporting performance, especially when comparing different approaches and when it comes to the reproducibility of claimed results. Furthermore how to conduct retrospective performance analysis such as an algorithm's suitability for embedded real-time systems over time with underlying hardware and software changes in place. This paper proposes and demonstrates a systematic way of addressing such challenges by adopting containerization of software aiming at formalization and reproducibility of benchmarks. Our results show maintainers of broadly accepted datasets in the computer vision community to strive for systematic comparison and reproducibility of submissions to increase the value and adoption of computer vision algorithms in the future.

Programming by Demonstration (PbD) lets users with little technical background program a wide variety of manipulation tasks for robots, but it should be as intuitive as possible for users while requiring as little time as possible. In this paper, we present a Programming by Demonstration system that synthesizes manipulation programs from a single observed demonstration, allowing users to program new tasks for a robot simply by performing the task once themselves. A human-in-the-loop interface helps users make corrections to the perceptual state as needed. We introduce Object Interaction Programs as a representation of multi-object, bimanual manipulation tasks and present algorithms for extracting programs from observed demonstrations and transferring programs to a robot to perform the task in a new scene. We demonstrate the expressivity and generalizability of our approach through an evaluation on a benchmark of complex tasks.

In this paper we propose a novel hybrid walking pattern generator which combines results from the virtual constraints and the preview control theories for bipedal locomotion. This choice is motivated by findings in biomechanics that show how the dynamic motion of the human walk is mainly generated by the sagittal component of the stepping. Thus, we choose the conservative preview control to generate the lateral motion while we pick a more dynamical framework such as the virtual constraints for the sagittal motion. We investigate how the time-dependent preview control and the time-independent virtual constraints approach can be integrated together in a humanoid locomotion and finally we show promising results on COMAN+, the new humanoid robot from Istituto Italiano di Tecnologia.

Reliable and accurate localization and mapping are key components of most autonomous systems. Besides geometric information about the mapped environment, the semantics plays an important role to enable intelligent navigation behaviors. In most realistic environments, this task is particularly complicated due to dynamics caused by moving objects, which can corrupt the mapping step or derail localization. In this paper, we propose an extension of a recently published surfel-based mapping approach exploiting three-dimensional laser range scans by integrating semantic information to facilitate the mapping process. The semantic information is efficiently extracted by a fully convolutional neural network and rendered on a spherical projection of the laser range data. This computed semantic segmentation results in point-wise labels for the whole scan, allowing us to build a semantically-enriched map with labeled surfels. This semantic map enables us to reliably filter moving objects, but also improve the projective scan matching via semantic constraints. Our experimental evaluation on challenging highways sequences from KITTI dataset with very few static structures and a large amount of moving cars shows the advantage of our semantic SLAM approach in comparison to a purely geometric, state-of-the-art approach.

In this paper, we investigate the relationship between the depth perception and an approaching motion toward an object. We propose the depth perception dependency test, which is the combination of tests of a motion and depth perception based on the situation that an object is placed on the human’s pathway. Firstly, as the motion test, we set the ball approach motion and asked elderly participants to approach and contact a ball by their foot, because this motion is easy to measure and requires localization skill and motion planning skill. The swing/support legs positions at the toeoff time of the swing leg just before contacting the ball was analyzed. Secondly, as the depth perception test, the pseudo3D image test was proposed. The coordinate transformation model for the calculation of the depth perception ability was also proposed. Through the proposed test and the proposed model, it was clarified that the participants who are regarded as perceiving the strong visual illusion perceive the objects closer than its real position, and their swing leg toe off positions in the ball approaching motion were significantly farther than other participants. Thus, it can be concluded that there was a relationship between the depth perception and the approaching motion that is thought to be a higher risk of stumbles.

The biomimetic approach in robotics is promising: nature has found many good solutions through millions of years of evolution. However, creating a design that enables fast and energy-efficient locomotion remains a major challenge. This paper focuses on the development of a full leg mechanism for a fast and energy-efficient 4-legged robot inspired by a cheetah morphology. In particular, we analyze how the allocation of flexible elements and their stiffness affects the cost of transport and peak power characteristics for vertical jumps and a galloping motion. The study includes the femur and full leg mechanism's locomotory behavior simulation, capturing its interaction with the ground.

Deep reinforcement learning has become a popular technique to train autonomous agents to learn control policies that enable them to accomplish complex tasks in uncertain environments. A key component of an RL algorithm is the definition of a reward function that maps each state and an action that can be taken in that state to some real-valued reward. Typically, reward functions informally capture an implicit (albeit vague) specification on the desired behavior of the agent. In this paper, we propose the use of the logical formalism of Signal Temporal Logic(STL) as a formal specification for the desired behaviors of the agent. Furthermore, we propose algorithms to locally shape rewards in each state with the goal of satisfying the high-level STL specification. We demonstrate our technique on two case studies, a cart-pole balancing problem with a discrete action space, and controlling the actuation of a simulated quadrotor for point-to-point movement.The proposed framework is agnostic to any specific RL algorithm, as locally shaped rewards can be easily used in concert with any deep RL algorithm.

Urban environments pose a significant challenge for autonomous vehicles (AVs) as they must safely navigate while in close proximity to many pedestrians. It is crucial for the AV to correctly understand and predict the future trajectories of pedestrians to avoid collision and plan a safe path. Deep neural networks (DNNs) have shown promising results in accurately predicting pedestrian trajectories, relying on large amounts of annotated real-world data to learn pedestrian behavior. However, collecting and annotating these large real-world pedestrian datasets is costly in both time and labor. This paper describes a novel method using a stochastic sampling-based simulation to train DNNs for pedestrian trajectory prediction with social interaction. Our novel simulation method can generate vast amounts of automatically-annotated, realistic, and naturalistic synthetic pedestrian trajectories based on small amounts of real annotation. We then use such synthetic trajectories to train an off-the-shelf state-of-the-art deep learning approach Social GAN (Generative Adversarial Network) to perform pedestrian trajectory prediction. Our proposed architecture, trained only using synthetic trajectories, achieves better prediction results compared to those trained on human-annotated real-world data using the same network. Our work demonstrates the effectiveness and potential of using simulation as a substitution for human annotation efforts to train high-performing prediction algorithms such as the DNNs.

Autonomous underwater gliders frequently execute extensive missions with high levels of uncertainty due to limitations of sensing, control and oceanic forecasting. Glider path planning seeks an optimal path with respect to conflicting objectives, such as travel cost and safety, that must be explicitly balanced subject to these uncertainties. In this paper, we derive a set of recursive equations for state probability and expected travel cost conditional on safety, and use them to implement a new stochastic variant of FMT* in the context of two types of objective functions that allow a glider to reach a destination region with minimum cost or maximum probability of arrival given a safety threshold. We demonstrate the framework using three simulated examples that illustrate how user-prescribed safety constraints affect the results.

Simultaneous Localization and Mapping (SLAM) is a fundamental task to mobile and aerial robotics. LiDAR based systems have proven to be superior compared to vision based systems due to its accuracy and robustness. In spite of its superiority, pure LiDAR based systems fail in certain degenerate cases like traveling through a tunnel. We propose Stereo Visual Inertial LiDAR (VIL) SLAM that performs better on these degenerate cases and has comparable performance on all other cases. VIL-SLAM accomplishes this by incorporating tightly-coupled stereo visual inertial odometry (VIO) with LiDAR mapping and LiDAR enhanced visual loop closure. The system generates loop-closure corrected 6-DOF LiDAR poses in real-time and lcm voxel dense maps near real-time. VIL-SLAM demonstrates improved accuracy and robustness compared to state-of-the-art LiDAR methods.

We introduce Recurrent State Representation Learning (RSRL) to tackle the problem of state representation learning in robotics for partially observable environments. To learn low-dimensional state representations, we combine a Long Short Term Memory network with robotic priors. RSRL introduces new priors with landmarks and combines them with existing robotics priors from the literature to train the representations. To evaluate the quality of the learned state representation, we introduce validation networks that help us better visualize and quantitatively analyze the learned state representations. We show that the learned representations are low-dimensional, locally consistent, and can approximate the underlying true state for robot localization in simulated 3D maze environments. We use the learned representations for reinforcement learning and show that we achieve similar performance as training with the true state. The learned representations are robust to landmark misclassification errors.

Pedestrian trajectory prediction is crucial for many important applications. This problem is a great challenge because of complicated interactions among pedestrians. Previous methods model only the pairwise interactions between pedestrians, which not only oversimplifies the interactions among pedestrians but also is computationally inefficient. In this paper, we propose a novel model StarNet to deal with these issues. StarNet has a star topology which includes a unique hub network and multiple host networks. The hub network takes observed trajectories of all pedestrians to produce a comprehensive description of the interpersonal interactions. Then the host networks, each of which corresponds to one pedestrian, consult the description and predict future trajectories. The star topology gives StarNet two advantages over conventional models. First, StarNet is able to consider the collective influence among all pedestrians in the hub network, making more accurate predictions. Second, StarNet is computationally efficient since the number of host network is linear to the number of pedestrians. Experiments on multiple public datasets demonstrate that StarNet outperforms multiple state-of-the-arts by a large margin in terms of both accuracy and efficiency.

This paper introduces a new gait switching approach for underactuated biped walkers. The switching condition relies on a reduced region of attraction - that of the unactuated dynamics - which is shown to be sufficient to predict falls. A gait transition is accordingly stable (as in “the robot does not fall if the biped's state is within the reduced region of attraction of the switched-in gait when switching takes place. Two- and five-link biped models complete a sequence of random gait transitions using the switching logic. The condition is also used to enlarge the full region of stability of the five-link model by embedding feedback-stabilized trajectories from a gait library in a mapping. The mapping stitches stable trajectories to the orbit of the desired gait based on each gait's reduced region of attraction. This improves the robustness of the five-link model walking on uneven terrain without ground perception regardless of the size of the gait library.

This paper presents a novel manipulator for aerial vehicles to perform grasping and manipulation tasks. The goal is to design a low-cost, relatively light but strong manipulator with a large workspace and compact storage space that can be mounted on an unmanned aerial system. A novel design solution based on the Spiral Zipper, an expanding tube, combined with tether actuators is presented. A model of the system is introduced and the control method and pose estimator are developed and tested with some experiments showing the reliable performance of the overall system. An experiment with a self-sealing suction cup gripper demonstrates manipulation while mounted on the aerial vehicle frame.

Energy-efficient simultaneous localization and mapping (SLAM) is crucial for mobile robots exploring unknown environments. The mammalian brain solves SLAM via a network of specialized neurons, exhibiting asynchronous computations and event-based communications, with very low energy consumption. We propose a brain-inspired spiking neural network (SNN) architecture that solves the unidimensional SLAM by introducing spike-based reference frame transformation, visual likelihood computation, and Bayesian inference. We integrated our neuromorphic algorithm to Intel's Loihi neuromorphic processor, a non-Von Neumann hardware that mimics the brain's computing paradigms. We performed comparative analyses for accuracy and energy-efficiency between our neuromorphic approach and the GMapping algorithm, which is widely used in small environments. Our Loihi-based SNN architecture consumes 100 times less energy than GMapping run on a CPU while having comparable accuracy in head direction localization and map-generation. These results pave the way for scaling our approach towards active-SLAM alternative solutions for Loihi-controlled autonomous robots.

We present a controller for human-robot handovers that is automatically synthesized from high-level specifications in Signal Temporal Logic (STL). In contrast to existing controllers, this approach can provide formal guarantees on the timing of each of the handover phases. Using synthesis also allows end-users to specify and dynamically change the robot's behaviors using high-level requirements of goals and constraints rather than by tuning low-level controller parameters. We illustrate the proposed approach by replicating the behavior of existing handover strategies from the literature. We also identify specification parameters that are likely to lead to successful handovers using a public database of human-human handovers.

In this paper, we study the problem of navigating quadcopters through a sequence of hoops. The specification may be given directly or indirectly via a linear temporal logic (LTL) formula. We approach this problem in three phases. First, we introduce a planner that generates a path through a given sequence of hoops. Second, we augment our planner to leverage a given specification in linear temporal logic (LTL) and generate a sequence that satisfies this specification. Third, we implement cross-entropy optimization on this planner to enhance trajectory performance where quadcopter trajectories are modified within the solution space to optimize over a cost function. We implement this planner as a novel interaction modality between users and quadcopters on the Robotarium. Simulation and experimental results are provided.