IROS 2021 Papers

In this paper, we explore the challenge of generating animal-like walking motions for legged robots. To this end, we propose a versatile and robust control pipeline that combines a state-of-the-art model-based controller with a data-driven technique that is commonly used in computer animation. We demonstrate the efficacy of our control framework on a variety of quadrupedal robots in simulation. We show, in particular, that our approach can automatically reproduce key characteristics of animal motions, including speed-specific gaits, unscripted footfall patterns for nonperiodic motions, and natural small variations in overall body movements.

Radar SLAM has attracted wide attention due to its all-day and all-weather working characteristics in the last decade. The existing radar SLAM systems mainly adopt mechanically pivoting radar with simple principle and high resolution, but this kind of radar has disadvantages such as low frame rate, distortion of the radar image, and high cost. Although array snapshot radar has the advantages of high frame rate and low cost, its low azimuth resolution, multipath reflection, and angular glint limit its application in SLAM. This paper proposes a SLAM system developed on array snapshot radar. The system realizes angular super-resolution radar imaging through compressed sensing, which effectively solves the problems of poor azimuth resolution and multipath reflection of array snapshot radar. We also propose the corresponding point cloud extraction method and scan matching method, this method performs a centroid iterative closest point algorithm between the submaps, thereby effectively improving the interference of noise and angular glint. Experimental results show that our proposed array snapshot radar SLAM system can reduce the mean absolute trajectory error by more than 3 times compared with the existing system, and can show accuracy and robustness in various environments.

Tip force estimation on continuum arms is of crucial clinical importance for catheter-based procedures, i.e., catheter-based ablation therapies. In this study, an analytical solution for force estimation based on inverse Cosserat rod modeling was proposed and validated. Initially, a previously validated Bezier-based shape interpolation was used to parameterize the deformation and the kinematics and balance equations of the catheter were derived thereof. The tip force estimation problem was formulated as an inverse problem with a functional minimization technique and was solved analytically. In the end, the proposed method was experimentally tested for accuracy and computation efficiency through a series of simulations and experiments. The results showed that the estimated forces were in agreement with reference measurement with a mean-absolute error of 0.024 ± 0.020 N and a computation time of 7 ± 5 ms per frame. The exhibited performance was comparable to other studies and was in compliance with the requirements of catheter-based procedures.

In this paper, we propose a simple analytical model for pneu-net soft actuator. The model is based on Euler– Bernoulli finite strain hyperelastic thin cantilever beam theory. The deformation of the air chambers is modelled using infinitesimal strain membrane theory. The proposed theoretical model estimates the deformation and force characteristics of the actuator. The developed model accounts the axial stretch and forces applied to the actuator. The theoretical deformation and force characteristics of different actuators are compared with finite element (FE) model and experimental characteristics. The theoretically estimated deformation and force of the actuator are similar to the FE model, but the theoretical model computation time is less than 1% of the FE model.

The spherical joint is an effective solution to design parallel micro-robotic systems with rotation capabilities in the three-dimensional space. This type of joint has however some non-linear characteristics, such as the clearance, which affect the positioning accuracy in micro-robotic tasks. The starting point of this study lies in experimental observations of rotation errors from a 3-PPPS 6-DOF parallel micro-robotic systems operating inside a scanning electron microscope. The objective of the paper is to assess the role of the spherical joints in the rotation errors and to evaluate whether the joints non-linearities can cause errors with the same order of magnitude as those observed experimentally. To this end, the first part of the study addresses the modeling of 3-PPPS 6-DOF parallel micro-robotic systems with spherical joints including the clearance. This model allows for analysing the effect of the clearance on position and rotation accuracies of the micro-robotic system. It is found by simulations that the same positioning behavior as in the experiments occurs when the clearance of the spherical joint is included in the model, supporting the hypothesis. Therefore, it is concluded that clearance in spherical joints has a significant effect on the precision of parallel type micro-robotic systems which opens new challenges in the control of poly-articulated micro-robotic systems with clearance compensation.

This paper considers how the motions of a telepresence robot moving autonomously affect a person immersed in the robot through a head-mounted display. In particular, we explore the preference, comfort, and naturalness of elements of piecewise linear paths compared to the same elements on a smooth path. In a user study, thirty-six subjects watched panoramic videos of three different paths through a simulated museum in virtual reality and responded to questionnaires regarding each path. Preference for a particular path was influenced the most by comfort, forward speed, and characteristics of the turns. Preference was also strongly associated with the users’ perceived naturalness, which was primarily determined by the ability to see salient objects, the distance to the walls and objects, as well as the turns. Participants favored the paths that had a one meter per second forward speed and rated the path with the least amount of turns as the most comfortable.

We propose a novel theoretical approach for solving a Stochastic Security Game using augmented Markov Decison Processes and an experimental evaluation. Most of the previous works mentioned in the literature focus on Linear Programming techniques seeking Strong Stackelberg Equilibria through the defender and attacker’s strategy spaces. Although effective, these techniques are computationally expensive and tend to not scale well to very large problems. By fixing the set of the possible defense strategies, our approach is able to use the well-known augmented MDP formalism to compute an optimal policy for an attacker facing a defender patrolling. Experimental results on fully observable cases validate our approach and show good performances in comparison with optimistic and pessimistic approaches. However, these results also highlight the need of scalability improvements and of handling the partial observability cases.

Current rigid linkage grippers are limited in flexibility, and gripper design optimality relies on expertise, experiments, or arbitrary parameters. Our proposed rigid gripper can accommodate irregular and off-center objects through a whippletree mechanism, improving adaptability. We present a whippletree-based rigid under-actuated gripper and its parametric design multi-objective optimization for a one-wall climbing task. Our proposed objective function considers kinematics and grasping forces simultaneously with a mathematical metric based on a model of an object environment. Our multi-objective problem is formulated as a single kinematic objective function with auto-tuning force-based weight. Our results indicate that our proposed objective function determines optimal parameters and kinematic ranges for our under-actuated gripper in the task environment with sufficient grasping forces.

This paper reports a novel optical localization method, including both the hardware design and algorithm design, to track mobile Unmanned Aerial Vehicles (UAVs). The method relies on a circle-shaped blinking LED marker installed on the UAV and uses a single Dynamic Vision Sensing (DVS) camera to sense the temporal difference of the video streams. A temporal-filtering algorithm processes the video stream and detects the target marker by filtering out the background image. The triangulation-based spatial localization algorithm captures the trace of the target with the help of the prior knowledge of the physical size of the marker. The proposed system was evaluated in flight tests and compared with ground truth data provided by a motion capture system. The proposed system provides a simple and accurate localization solution for UAV tracking with a low computing overhead.

Underwater visual localization is an essential technique for the autonomous operation of underwater robots. However, the unique underwater image characteristics, including refraction, sparse features, and severe noise, pose an enormous challenge to it. For addressing these issues, this paper proposes an open-source fiducial-based underwater stereo visual-inertial localization method under the extended Kalman filter (EKF) framework, which is called FBUS-EKF. First, the refraction is corrected by the refractive camera model and akin triangulation. Second, the fiducial marker and a novel marker pose estimation method are applied to alleviate the adverse effect of sparse features. Third, the EKF is utilized to fuse the inertial and visual information so as to reject the serious noise. Finally, extensive experiments on a test bench demonstrate the effectiveness of the FBUS-EKF method, where the typical localization error is less than 3%, namely, the average error is lower than 3 cm within one meter. The obtained results reveal that the FBUS-EKF method has the prospect to be applied in the precise short-range operation and the localization for underwater robots, which offers a valuable insight for further autonomous underwater task.

To realize effective heterogeneous multi-robot teams, researchers must leverage individual robots’ relative strengths and coordinate their individual behaviors. Specifically, heterogeneous multi-robot systems must answer three important questions: who (task allocation), when (scheduling), and how (motion planning). While specific variants of each of these problems are known to be NP-Hard, their interdependence only exacerbates the challenges involved in solving them together. In this paper, we present a novel framework that interleaves task allocation, scheduling, and motion planning. We introduce a search-based approach for trait-based time-extended task allocation named Incremental Task Allocation Graph Search (ITAGS). In contrast to approaches that solve the three problems in sequence, ITAGS’s interleaved approach enables efficient search for allocations while simultaneously satisfying scheduling constraints and accounting for the time taken to execute motion plans. To enable effective interleaving, we develop a convex combination of two search heuristics that optimizes the satisfaction of task requirements as well as the makespan of the associated schedule. We demonstrate the efficacy of ITAGS using detailed ablation studies and comparisons against two state-of-the-art algorithms in a simulated emergency response domain.

Acceptance of social robots in human-robot collaborative environments depends on the robots’ sensitivity to human moral and social norms. Robot behavior that violates norms may decrease trust and lead human interactants to blame the robot and view it negatively. Hence, for long-term acceptance, social robots need to detect possible norm violations in their action plans and refuse to perform such plans. This paper integrates the Distributed, Integrated, Affect, Reflection, Cognition (DIARC) robot architecture (implemented in the Agent Development Environment (ADE)) with a novel place recognition module and a norm-aware task planner to achieve context-sensitive moral reasoning. This will allow the robot to reject inappropriate commands and comply with context-sensitive norms. In a validation scenario, our results show that the robot would not comply with a human command to violate a privacy norm in a private context.

The design of an industrial pipe inspection robot for 4" firewater piping inspection and mapping with some new and unreported features is presented. The robot consists of a compact 3D MEMs Lidar with non-linear scanning pattern in the form of Lissajous trajectories that produces a dense scan of inner pipe surfaces. Evaluating the surface integrals of these 3D point clouds acquired at pipe turns allow for turn direction discrimination on the horizontal plane which is not realizable using IMUs. A non-SLAM method is used for robust localization: the actual distance traveled by the robot inside a pipe network is measured explicitly by measuring tether displacement; this measurand is then used as the arc length parameter of a parametric curve. This parameterization represents the robot trajectory such that the 3D robot positions along this trajectory are defined in terms of distances traveled (arc length of parametric curve). Pose transitions measured using an on-board IMU combined with the arc-length parametric trajectory let us determine the full 6D robot pose inside featureless pipes. Instantaneous Lidar scans and the 6D poses are used to create a high density 3D map of the pipe network to help identify pipe joint deformation at cm level.

Generating meaningful spatial models of physical environments is a crucial ability for autonomous navigation of mobile robots. This paper considers the problem of building continuous occupancy maps from sparse and noisy sensor data. To this end, we propose a new method named random mapping maps that advances the popular methods in two aspects. Firstly, it can represent environment models in a memory-saving and time-saving manner by randomly mapping a low-dimensional feature space to a high-dimensional one where a linear model is learnt. Secondly, it can rapidly obtain accurate inferences of the occupancy states of the spatial locations. This technique is based on the random mapping that projects the measurement data into a random feature space in which a discriminative model is learnt by the available data. It can asymptotically represent the complexity of the real world as the mapping dimension increases. Evaluations of the proposed method were conducted on various environments to verify its availability to environment modeling. Its performances in terms of time and memory consumptions were evaluated quantitatively. Finally, as a practical application, experiments about path planning were conducted based on the gradients of the proposed representation of environment model.

Motion optimization for legible robot intent has largely ignored the robot’s dynamics, citing burdensome complexity that prevents online deployment. Even where the original task (to be communicated) could be solved on the dynamical system, the legibility problem (to communicate that task’s intent) could not. This work simplifies the legibility objective to have equivalent computational complexity as the original objective to be communicated. This enables any optimal control algorithm that can solve the original task to also solve the legible version of that task.Along the way, we expand the definition of "intent" to include any parameter of the optimal control problem, thereby opening the door to extend communications beyond merely desired end-points to running preferences or even, in the future, hard capabilities or safety constraints. We demonstrate how this method can replicate the properties introduced in previous communicative motion state-of-the-art (like legibility, exaggeration, and anticipation) as well as apply to non-holonomic dynamical systems.

Humans effortlessly solve pushing tasks in everyday life but unlocking these capabilities remains a challenge in robotics because physics models of these tasks are often inaccurate or unattainable. State-of-the-art data-driven approaches learn to compensate for these inaccuracies or replace the approximated physics models altogether. Nevertheless, approaches like Deep Q-Networks (DQNs) suffer from local optima in large state-action spaces. Furthermore, they rely on well-chosen deep learning architectures and learning paradigms. In this paper, we propose to frame the learning of pushing policies (where to push and how) by DQNs as an image-to-image translation problem and exploit an Hourglass-based architecture. We present an architecture combining a predictor of which pushes lead to changes in the environment with a state-action value predictor dedicated to the pushing task. Moreover, we investigate positional information encoding to learn position-dependent policy behaviors. We demonstrate in simulation experiments with a UR5 robot arm that our overall architecture helps the DQN learn faster and achieve higher performance in a pushing task involving objects with unknown dynamics.

The paper presents a shared control architecture for robotic systems commanded through a motor imagery based Brain-Computer Interface (BCI). The overall system is aimed at assisting people to perform teleoperated manipulation tasks, and it is structured so as to leave different levels of autonomy to the user depending on the actual stage of the task execution. The low-level part of the shared control architecture is also in charge of taking into account safety and operational tasks, such as to avoid collisions or to manage robot joint limits. The overall architecture has been realized by integrating control and perception software modules developed within the ROS environment, with the OpenVibe framework used to operate the BCI device. The effectiveness of the proposed architecture has been validated through experiments where a healthy user, wearing a Unicorn g.tec BCI, performs an assisted task through motor imagery sessions, with a 7 Degrees of Freedoms Kinova Jaco2 robotic arm.

Evaluation of social robot navigation inherently requires human input due to its qualitative nature. Motivated by the need to scale human evaluation, we propose a general method for deploying interactive, rich-client robotic simulations on the web. Prior approaches implement specific web- compatible simulators or provide tools to build a simulator for a specific study. Instead, our approach builds on standard Linux tools to share a graphical desktop with remote users. We leverage these tools to deploy simulators on the web that would typically be constrained to desktop computing environments. As an example implementation of our approach, we introduce the SEAN Experimental Platform (SEAN-EP). With SEAN- EP, remote users can virtually interact with a mobile robot in the Social Environment for Autonomous Navigation, without installing any software on their computer or needing specialized hardware. We validated that SEAN-EP could quickly scale the collection of human feedback and its usability through an online survey. In addition, we compared human feedback from participants that interacted with a robot using SEAN- EP with feedback obtained through a more traditional video survey. Our results suggest that human perceptions of robots may differ based on whether they interact with the robots in simulation or observe them in videos. Also, they suggest that people perceive the surveys with interactive simulations as less mentally demanding than video surveys.

Over the last decade, adaptive tendon driven devices have gained an increased interest from the research community for their lightweight, compact, and affordable design features attributed to the utilisation of underactuation, differential mechanisms, and structural compliance. Although adaptive tendon driven devices are capable of efficiently executing stable grasps under significant object pose uncertainties with simplistic control algorithms, they lack the controllability over individual fingers in comparison to traditional fully actuated designs. In this paper, we focus on the development of a selectively lockable differential mechanism that is powered through a small and low torque servo to provide increased autonomy to highly underactuated and adaptive prosthetic hands, without compromising the weight, cost, and compactness of the device. The proposed prosthetic hand is experimentally validated through four tests: i) grasping posture and gesture execution experiments, ii) grasping experiments with everyday life objects, iii) force exertion experiments, and iv) Electromyography (EMG) based control of the prosthetic hand.