IROS 2019 Papers

We proposed the soft flexible wire-driven finger mechanism with the soft skin and the multi-joint skeletal structure using two coil springs. The multi-joint skeletal structure is mainly composed of two coil springs, multiple skeletal members and a fiber wire, and the soft skin is formed outside of them. The soft skin and the multi-joint skeletal structure make it possible to distribute the contact pressure, which is necessary for not only touching a living body such as a human or an agricultural crop, but also for touching an artificial object such as pastry or an industrial product without damaging it or its package. In this paper, we describe the design of the soft flexible wire-driven finger mechanism and development of the three-fingered hand using the finger mechanism we proposed.

Generally, large knee joint torque is required when a leg is flexed. However, the large torque motor will increase the robot size and total weight. Also, a large reduction ratio gear box to realize the large torque will decrease output speed and lower backdrivability of the joint. It makes the robot difficult to perform agile and flexible motion like animals. In this paper, we propose a robot leg with a knee joint mechanism consisting of a variable reduction ratio crossed four-bar linkage mechanism (VRRCFLM) based on cruciate ligament of an animal. The VRRCFLM is to increase the reduction ratio for large knee flexion postures without greatly reducing the total backdrivability. In this paper, we developed a robot leg with a knee joint mechanism consisting of the VRRCFLM. In order to design the link parameters of the leg mechanism, optimization design aimed at maximizing the jumping height of the robot was performed. The robot model with the designed mechanism was evaluated through the dynamics simulations. Thanks to the VRRCFLM, the required torque of the knee motor at the large flexion postures was decreased. Moreover, the vertical jumping height was improved by 24.6 % comparing with a model without the mechanism. In experiments using the prototype, the required static torque was decreased as in simulation, and the jumping height was more than one leg length.

Robotic origami design allows creating meso-scale robotic systems and mechanisms not limited by degrees of freedom, miniaturization and assembly downsides of conventional transmission mechanisms. However, unlike the traditional rigid approaches, robotic origami application has been limited by the complex deformation and kinematics of the compliant joints and actuation based on active materials or conventional electric motors. To generalize their application at meso-scale requires a combination of the predictability of traditional rigid kinematics and the manufacturing flexibility of robotic origami. Here we present a study of conventional transmission mechanisms, including a slider-crank and cam-follower, made in quasi-2D form by selective machining and stacking of multiple layers of composite material with minimal assembly. Owing to a compliant design powered by low-profile piezoelectric motors, our 5.3 mm thick and lightweight mechanisms transmit rotational motion to translational movements in and out-of-plane. We develop analytic models that we validate in terms of force and motion output on our prototypes.

In this paper, we propose a new compact variable gravity compensation mechanism (CVGC). The CVGC can be used to generate gravity compensation torque by using the cam and lever mechanism and can also amplify the target gravity compensation torque by varying the pivot point of the lever. The feature of variable gravity compensation is very useful to the mobile platform, which needs to handle unstandardized tasks with a high variation of the workpiece weight. The proposed CVGC has many advantages. Most importantly, it is designed as a compact, independent one-piece structure and is lightweight, meaning it can easily be used as a mobile platform with a simple modification. The CVGC can also have a full range of compensation angle (360°), so it does not restrict any of the original workspaces of the target platform when it is installed. First, the mechanism concept and details are explained. Next, the mechanics of the prototype for force analysis are presented. Based on these mechanics and cam theory, the methodology of the cam profile design is presented. Finally, the performance of variable gravity compensation is verified through experiments that compare the designed and measured gravity compensation torque. The verification test shows adequate performance, as we had hoped, which shows potential for the development of the CVGC.

Master manipulators represent a key component of Robot-Assisted Minimally Invasive Surgery (RAMIS). In this paper, an Analytic Hierarchy Process (AHP) method is used to construct an effective workspace analysis framework, which can assist the configuration selection and design evaluation of a portable master manipulator for surgical robot control and training. The proposed framework is designed based on three criteria: 1) compactness, 2) workspace quality, and 3) mapping efficiency. A hardware prototype, called the Hamlyn Compact Robotic Master (Hamlyn CRM), is constructed following the proposed framework. Experimental verification of the platform is conducted on the da Vinci Research Kit (dVRK) with which a da Vinci robot is controlled as a slave. The proposed Hamlyn CRM is compared with Phantom Omni, a commercial portable master device, with results demonstrating the relative merits of the new platform in terms of task completion time, average control speed and number of clutching.

Motivated by a winged seed, which takes advantage of a wing with high angles of attack and its associated leading-edge vortex to boost lift, we propose a powered 13.8gram aerial robot with the maximum take-off weight of 310 mN (31.6 gram) or thrust-to-weight ratio of 2.3. The robot, consisting of two airfoils and two horizontally directed motor-driven propellers, revolves around its vertical axis to hover. To amplify the thrust production while retaining a minimal weight, we develop an optimization framework for the robot and airfoil geometries. The analysis integrates quasi-steady aerodynamic models for the airfoils and the propellers with the motor model. We fabricated the robots according to the optimized design. The prototypes are experimentally tested. The revolving-wing robot produces approximately 50% higher lift compared to conventional multirotor designs. Finally, an uncontrolled hovering flight is presented.

Glass curtain walls have been widely used in modern architecture. This makes it urgent to inspect and clean these glasses at regular intervals. Up to now, most of these work is performed by workers, which is expensive and inefficient. Therefore, a novel robot-the contact aerial manipulator system-is developed. The new designed system presents priorities in the aspects of high flexibility and easy operation. In this paper, the system mechanical structure is first introduced. Subsequently, the hybrid force/motion control framework is utilized to realize the precise and steady motion on the two-dimensional plane and maintain a certain sustained contact force, simultaneously. Finally, two flight experiments (including continuous square-wave trajectory tracking and aerial drawing task) are performed and the results indicate that the developed contact aerial manipulator works and presents good performance.

A novel compact parallel gripper with wide stroke is proposed in this paper. Conventional parallel grippers with linear guide mechanisms have been widely utilized as end effectors of robots, especially in industrial fields, because of its simple mechanism and low cost nature. However, the width of the gripper is larger than the stroke due to the mechanical structure of its linear guide mechanism. When objects with various dimensions need to be grasped by a gripper, the width of the gripper is determined by the largest distance between fingers necessary for grasping all of the objects. This makes the collision avoidance of robot difficult even when handling a small object compared to the width of the gripper. The novel compact gripper proposed in this article has three features: long finger stroke with compact dimensions, fingers’ parallel motion along a single linear trajectory, and driven by a single actuator. The finger backlash and the gripping force of the proposed mechanism are analyzed, and the parallel gripper with the mechanism is designed based on the analyses. A prototype gripper with minimum width 64 mm, whose maximum fingers stroke is 121 mm, is developed, and experimental results illustrate the performance of the developed gripper.

Hopping is a highly dynamic motion requiring precise input over brief moments of ground contact in order to achieve desired performance. While this problem has been approached from multiple perspectives, this work provides a comparative analysis of two robot models. The first model uses an actuator to store energy in a spring and release it during the ground phase, while the second uses an actuator to move an additional mass vertically to generate force on the spring. In the first model, analytic expressions are used to find the desired controllers, while trajectory optimization is used in the latter. Orbital stability of each model under the conditions of uncertain damping and poor estimation of the hop height is examined. To this end, Poincaré analysis is used to give a metric of stability in the presence of different initial conditions and parameter uncertainty. Simulations show that the first model converges quickly to a point near the desired height determined by the amount of uncertain damping present. The second model is less robust to uncertainty, but is be made to converge to a desired height with the addition of PD control around the optimal trajectory. This robustness is improved with different gains in the controller. In experiments performed on hardware for the second model, stability is observed through convergence to a periodic orbit within several hops.

Batoids use their large pectoral fins to achieve unique maneuverability and propulsive performance. In this work, the design, fabrication and characterization a soft batoid-like robot is presented. The robot is designed to mimic un-dulatory rajiform locomotion. The design is under-actuated, simple and robust and well suited for propulsive performance experiments thanks to its full autonomy, long battery life, and wireless recharging capabilities. The robot has a 180mm body length, a maximum flapping amplitude of 60deg, and reaches a peak speed of 0.93 body length per second. In order to characterize its swimming kinematics and propulsive forces a special setup was built using an instrumented holder and high speed video. Experiments were conducted to characterize propulsive force generation with varying fin flapping frequencies (from 1hz to 2. 4hz), amplitudes, and wavelengths. The results show that while both flapping frequency and amplitude influence propulsive forces, for the locomotion modes tested, flapping frequency has a stronger effect on both thrust and side forces. Furthermore, larger side forces than thrust forces are produced for the same swimming parameters.

This paper presents the design, analysis and testing of a novel three-degree-of-freedom (3-DOF) compliant parallel-kinematic active constant-force stage. The active constant-force property enables a large travel and constant driving property, which is enabled by introducing symmetrical bistable flexure hinges. The folded flexure mechanism is adopted to guide the driving input and to balance the stiffness of the stage to zero. In addition, leaf flexure hinges are employed to decouple the cross-axis motion of the 3-DOF parallel stage. Analytical modeling is conducted to evaluate the stage performance. To verify the performance of the constant-force property and motion decoupling, finite-element analysis simulation study is carried out. By minimizing the fluctuation of the constant-force value, design optimization of the stage parameters is implemented with multi-objective genetic algorithm. Moreover, a prototype is fabricated for demonstration of the proposed concept design.

We present a pedestrian tracking algorithm, DensePeds, that tracks individuals in highly dense crowds (>2 pedestrians per square meter). Our approach is designed for videos captured from front-facing or elevated cameras. We present a new motion model called Front-RVO (FRVO) for predicting pedestrian movements in dense situations using collision avoidance constraints and combine it with state-of-the-art Mask R-CNN to compute sparse feature vectors that reduce the loss of pedestrian tracks (false negatives). We evaluate DensePeds on the standard MOT benchmarks as well as a new dense crowd dataset. In practice, our approach is 4.5 × faster than prior tracking algorithms on the MOT benchmark and we are state-of-the-art in dense crowd videos by over 2.6% on the absolute scale on average.

Typically, the reconstruction problem is addressed in three independent steps: first, sensor processing techniques are used to filter and segment sensor data as required by the front end. Second, the front end builds the factor graph for the problem to obtain an accurate estimate of the robot’s full trajectory. Finally, the end product is obtained by further processing of sensor data, now re-projected from the optimized trajectory. In this paper we present an approach to model the reconstruction problem in a way that unifies the aforementioned problems under a single framework for a particular application: sonar-based inspection of underwater structures. This is achieved by formulating both the sonar segmentation and point cloud reconstruction problems as factor graphs, in tandem with the SLAM problem. We provide experimental results using data from a ship hull inspection test.

Advances in Unmanned Aerial Vehicle (UAV) opens venues for application such as tunnel inspection. Owing to its versatility to fly inside the tunnels, it can quickly identify defects and potential problems related to safety. However, long tunnels, especially with repetitive or uniform structures pose a significant problem for UAV navigation. Furthermore, post-processing visual data from the camera mounted on the UAV is required to generate useful information for the inspection task. In this work, we design a UAV with a single rotating camera to accomplish the task. Compared to other platforms, our solution can fit the stringent requirement for tunnel inspection, in terms of battery life, size and weight. While the current state-of-the-art can estimate camera pose and 3D geometry from a sequence of images, they assume large overlap, small rotational motion, and many distinct matching points between images. These assumptions severely limit their effectiveness in tunnel-like scenarios where the camera has erratic or large rotational motion, such as the one mounted on the UAV. This paper presents a novel solution which exploits Structure-from-Motion, Bundle Adjustment, and available geometry priors to robustly estimate camera pose and automatically reconstruct a fully-dense 3D scene using the least possible number of images in various challenging tunnel-like environments. We validate our system with both Virtual Reality application and experimentation with a real dataset. The results demonstrate that the proposed reconstruction along with texture mapping allows for remote navigation and inspection of tunnel-like environments, even those which are inaccessible for humans.

Lane estimation plays a central role in the context of self-driving vehicles, requiring clear road markings and a high quality infrastructure. To meet the safety requirements and manage difficult use cases (i.e. poor quality or absence of road- marking, traffic jams, severe weather conditions, etc.) without depending on cartography and lane markings, we propose a lidar-based Virtual Lanes (VL) generation system to provide a comfortable and safe ride. For this purpose, hybrid fusion based on Dempster-Shafer Theory (DST) coupled with a particle filter is introduced. Two VL strategies are merged: the first constructs virtual lanes based on independent road-borders detection, while the second uses moving objects trajectories. A novel lane similarity computation is also adopted in order to estimate an effective lane comparison. The performance is reported through extensive experiments with Valeo demo-car on highway and beltway roads. Experimental results demonstrate the accurate and robust performance of the proposed system.

Social robots are becoming an integrated part of our daily life due to their ability to provide companionship and entertainment. A subfield of robotics, Socially Assistive Robotics (SAR), is particularly suitable for expanding these benefits into the healthcare setting because of its unique ability to provide cognitive, social, and emotional support. This paper presents our recent research on developing SAR by evaluating the ability of a life-like conversational social robot, called Ryan, to administer internet-delivered cognitive behavioral therapy (iCBT) to older adults with depression. For Ryan to administer the therapy, we developed a dialogue-management system, called Program-R. Using an accredited CBT manual for the treatment of depression, we created seven hour-long iCBT dialogues and integrated them into Program-R using Artificial Intelligence Markup Language (AIML). To assess the effectiveness of Robot-based iCBT and users' likability of our approach, we conducted an HRI study with a cohort of elderly people with mild-to-moderate depression over a period of four weeks. Quantitative analyses of participant's spoken responses (e.g. word count and sentiment analysis), face-scale mood scores, and exit surveys, strongly support the notion robot-based iCBT is a viable alternative to traditional human-delivered therapy.

Simultaneous Localization and Mapping (SLAM) is commonly formulated as an optimization over a graph. A popular approach is the pose graph, which seeks to solve for robots poses that are constrained by pose-to-pose measurements, such as odometry measurements or loop closures. For range sensors, these pose-to-pose constraints can be achieved by performing scan matching techniques, such as Iterative Closest Point (ICP). However, in environments with insufficient or degenerate geometric features, the ICP solution can be unreliable and lead to significant drift in the trajectory of the graph optimization solution. In this paper, we propose a degeneracy-aware approach which has two stages: (1) a degeneracy-aware ICP algorithm and (2) a partially constrained loop closure factor to incorporate the results from (1) into the SLAM pose graph optimization. Our approach performs updates and optimizes both ICP and the pose graph in only the well constrained directions of the state space. These directions are selected on the basis of a dynamic threshold, which updates at each iteration. We apply the proposed algorithm to autonomous underwater mapping with sonar. To evaluate the performance of this algorithm, we conduct experiments in both simulation and real world scenarios, and show the method's robustness to navigational drift and ability to reject poor loop closures in degenerate environments, which would otherwise degrade the accuracy of the trajectory and the quality of the resulting map.

Self-calibration of camera intrinsics and radial distortion has a long history of research in the computer vision community. However, it remains rare to see real applications of such techniques to modern Simultaneous Localization And Mapping (SLAM) systems, especially in driving scenarios. In this paper, we revisit the geometric approach to this problem, and provide a theoretical proof that explicitly shows the ambiguity between radial distortion and scene depth when two-view geometry is used to self-calibrate the radial distortion. In view of such geometric degeneracy, we propose a learning approach that trains a convolutional neural network (CNN) on a large amount of synthetic data. We demonstrate the utility of our proposed method by applying it as a checkerboard-free calibration tool for SLAM, achieving comparable or superior performance to previous learning and hand-crafted methods.

This paper presents an self-supervised deep learning network for monocular visual inertial odometry (named DeepVIO). DeepVIO provides absolute trajectory estimation by directly merging 2D optical flow feature (OFF) and Inertial Measurement Unit (IMU) data. Specifically, it firstly estimates the depth and dense 3D point cloud of each scene by using stereo sequences, and then obtains 3D geometric constraints including 3D optical flow and 6-DoF pose as supervisory signals. Note that such 3D optical flow shows robustness and accuracy to dynamic objects and textureless environments. In DeepVIO training, 2D optical flow network is constrained by the projection of its corresponding 3D optical flow, and LSTM-style IMU preintegration network and the fusion network are learned by minimizing the loss functions from ego-motion constraints. Furthermore, we employ an IMU status update scheme to improve IMU pose estimation through updating the additional gyroscope and accelerometer bias. The experimental results on KITTI and EuRoC datasets show that DeepVIO outperforms state-of-the-art learning based methods in terms of accuracy and data adaptability. Compared to the traditional methods, DeepVIO reduces the impacts of inaccurate Camera-IMU calibrations, unsynchronized and missing data.