ICRA 2019 论文列表

It is well known that sensors using strain gauges have a potential dependency on temperature. This creates temperature drift in the measurements of six axis force torque sensors (F/T). The temperature drift can be considerable if an experiment is long or the environmental conditions are different from when the calibration of the sensor was performed. Other in situ methods disregard the effect of temperature on the sensor measurements. Experiments performed using the humanoid robot platform iCub show that the effect of temperature is relevant. The model based in situ calibration of six axis force torque sensors method is extended to perform temperature compensation.

Efficient and reliable dynamic modelling and analysis is critical to the shape control and estimate of a backbone continuum manipulator. This paper presents a novel dynamic modelling method to investigate the deformation modal properties of a vertical stretch-retractable continuum manipulator (VSRCM) under the tip horizontal load. Based on the equivalence of the elastic bending potential energy, this method simplifies the backbone continuum manipulator to a single beam, an equivalent-guided beam (EGB). One end of the EGB is fixed, and the other end is guided. The modal dynamics model of the EGB is then established by employing the vibration theory of a continuous beam, and the modal properties of the continuum manipulator are analyzed. Static and Dynamic experiments of the VSRCM are performed to validate the dynamic modeling method through the comparison and analysis of the free dynamic response and frequency properties of the continuum manipulator. To analyze the free dynamic response, the large deflection static model with tip horizontal load is also established based on the elliptic integral approach. Experimental results and comparison demonstrate that the static model is efficient and precise to predict the large deformation, and the modal dynamic models well reflect the actual vibration modal characteristics of the VSRCM.

Flying modular robots have the potential to rapidly form temporary structures. In the literature, docking actions rely on external systems and indoor infrastructures for relative pose estimation. In contrast to related work, we provide local estimation during the self-assembly process to avoid dependency on external systems. In this paper, we introduce ModQuad-Vi, a flying modular robot that is aimed to operate in outdoor environments. We propose a new robot design and vision-based docking method. Our design is based on a quadrotor platform with onboard computation and visual perception. Our control method is able to accurately align modules for docking actions. Additionally, we present the dynamics and a geometric controller for the aerial modular system. Experiments validate the vision-based docking method with successful results.

Robotic painting is well-established in controlled factory environments, but there is now potential for mobile robots to do functional painting tasks around the everyday world. An obvious first target for such robots is painting a uniform single color. A step further is the painting of textured images. Texture involves a varying appearance, and requires that paint is delivered accurately onto the physical surface to produce the desired effect. Robotic painting of texture is relevant for architecture and in themed environments.A key challenge for robotic painting of texture is to take a desired image as input, and to generate the paint commands to as closely as possible create the desired appearance, according to the robotic capabilities. This paper describes a deep learning approach to take an input ink map of a desired texture, and infer robotic paint commands to produce that texture.We analyze the trade-offs between quality of reconstructed appearance and ease of execution. Our method is general for different kinds of robotic paint delivery systems, but the emphasis here is on spray painting. More generally, the framework can be viewed as an approach for solving a specific class of inverse imaging problems.

We present an augmented reality human-swarm interface that combines two modalities of interaction: environment-oriented and robot-oriented. The environment-oriented modality allows the user to modify the environment (either virtual or physical) to indicate a goal to attain for the robot swarm. The robot-oriented modality makes it possible to select individual robots to reassign them to other tasks to increase performance or remedy failures. Previous research has concluded that environment-oriented interaction might prove more difficult to grasp for untrained users. In this paper, we report a user study which indicates that, at least in collective transport, environment-oriented interaction is more effective than purely robot-oriented interaction, and that the two combined achieve remarkable efficacy.

Pedestrian detection has attracted enormous research attention in the field of Intelligent Transportation System (ITS) due to that pedestrians are the most vulnerable traffic participants. So far, almost all pedestrian detection solutions are based on the conventional frame-based camera. However, they cannot perform very well in scenarios with bad light condition and high-speed motion. In this work, a Dynamic and Active Pixel Sensor (DAVIS), whose two channels concurrently output conventional gray-scale frames and asynchronous low-latency temporal contrast events of light intensity, was first used to detect pedestrians in a traffic monitoring scenario. Data from two camera channels were fed into Convolutional Neural Networks (CNNs) including three YOLOv3 models and three YOLO-tiny models to gather bounding boxes of pedestrians with respective confidence map. Furthermore, a confidence map fusion method combining the CNN-based detection results from both DAVIS channels was proposed to obtain higher accuracy. The experiments were conducted on a custom dataset collected on TUM campus. Benefiting from the high speed, low latency and wide dynamic range of the event channel, our method achieved higher frame rate and lower latency than those only using a conventional camera. Additionally, it reached higher average precision by using the fusion approach.

We work with geoscientists studying erosion and desertification to improve the spatial and temporal resolution of their data collection over long transects in difficult realworld environments such as deserts [1]. The Minitaur [2] robot, which can run quickly over uneven terrain and use a single leg to measure relevant ground properties such as stiffness [3], is an attractive scout robot candidate for inclusion in a heterogeneous team in collaboration with a heavily geared, sensor-laden RHex [4]. However, Minitaur is challenged by long-distance locomotion on sand dunes. Previous simulation results [5] suggested that the energetic cost of transport can be mitigated by programming a virtual damping force to slow the intrusion of a Minitaur foot into simulated granular media following a bulk-behavior force law [6]. In this paper, we present a ground emulator that can be used to test such locomotion hypotheses with a physical single-legged hopper jumping on emulated ground programmed to exhibit any compliance and damping characteristics of interest. The new emulator allows us to corroborate the conclusions of our previous simulation with physical hopping experiments. Programming the substrate emulator to exhibit the mechanics of a simplified bulk-behavior model of granular media characterized by linear stiffness and quadratic damping, we achieve a consistent energy savings of 20% in comparison with a nominal controller, with savings of up to 50% under specific conditions.

The following topics are dealt with: mobile robots; learning (artificial intelligence); robot vision; path planning; motion control; medical robotics; optimisation; object detection; position control; collision avoidance.

We present a medical-needle-sized robotic tube with multi-degrees of freedom (M-DOF) rotational hinge joints at the instrument tip, fabricated by two-axis laser micro-machining. Due to the presence of an ample working channel and direct tip controllability, this tube is a potential candidate for the precise delivery of radioactive seeds, probes and micro-forceps to regions of interest within the human body. In this paper, the advantages of the proposed hinged instrument are studied in contrast with that of flexure joints, which include-fine angle control (posability), and a shorter joint length that enables compact articulation. In addition, the joint strength both in the axial and lateral directions was experimentally investigated to demonstrate its feasibility as a robust delivery platform. Further, the intuitive nature of hinge rotation permits the use of a simple kinematic model for accurate tip motion control, under fewer simplifying assumptions than flexure joints which are impeded by material non-linearity and geometric discontinuities. An instrumented prototype was used to test this model by delivering a laser beam along a prescribed path (synonymous to simple ablation tasks). The observed RMS position error for the projected beam was ~0.364 mm.

Mini Cheetah is a small and inexpensive, yet powerful and mechanically robust quadruped robot, intended to enable rapid development of control systems for legged robots. The robot uses custom backdriveable modular actuators, which enable high-bandwidth force control, high force density, and robustness to impacts. Standing around 0.3 m tall and weighing 9 kg, Mini Cheetah can easily be handled by a single operator. We have demonstrated dynamic trot, trot-run, bounding, and pronking gaits on the robot to speeds of up to 2.45 meters per second using Convex Model-Predictive Control (cMPC). In addition to locomotion, we have used the robot to execute 360° backflips, with trajectories generated using offline nonlinear optimization.

A robot system is designed as a set of embodied agents. An embodied agent is decomposed into cooperating subsystems. In our previous work activities of subsystems were defined by hierarchical finite state machines. With their states activities were associated. In that approach communication between subsystems was treated as an implementation issue. This paper represents activities of a robot system using hierarchical Petri nets with conditions. Such net is created by specifying consecutive layers: multi-agent robot system layer, agent layer, subsystem layer, behaviour layer and communication layer. This decomposition not only organizes in a systematic manner the development of a robot system, but also introduces a comprehensive description of concurrently acting subsystems. Based on those theoretical considerations, a tool was created for producing hierarchical Petri nets defining the model of a robotic system and enabling automatic generation of the robot controller code, resulting in a significant acceleration of the implementation phase. The capabilities of the tool are presented by the development of a robot controller performing a rudimentary task.

Data-driven approach for grasping shows significant advance recently. But these approaches usually require much training data. To increase the efficiency of grasping data collection, this paper presents a novel grasp training system including the whole pipeline from data collection to model inference. The system can collect effective grasp sample with a corrective strategy assisted by antipodal grasp rule, and we design an affordance interpreter network to predict pixelwise grasp affordance map. We define graspability, ungraspability and background as grasp affordances. The key advantage of our system is that the pixel-level affordance interpreter network trained with only a small number of grasp samples under antipodal rule can achieve significant performance on totally unseen objects and backgrounds. The training sample is only collected in simulation. Extensive qualitative and quantitative experiments demonstrate the accuracy and robustness of our proposed approach. In the real-world grasp experiments, we achieve a grasp success rate of 93% on a set of household items and 91% on a set of adversarial items with only about 6,300 simulated samples. We also achieve 87% accuracy in clutter scenario. Although the model is trained using only RGB image, when changing the background textures, it also performs well and can achieve even 94% accuracy on the set of adversarial objects, which outperforms current state-of-the-art methods.

In this paper, we focus on generating complex robotic trajectories by merging sequential motion primitives. A robotic trajectory is a time series of positions and orientations ending at a desired target. Hence, we first discuss the generation of converging pose trajectories via dynamical systems, providing a rigorous stability analysis. Then, we present approaches to merge motion primitives which represent both the position and the orientation part of the motion. Developed approaches preserve the shape of each learned movement and allow for continuous transitions among succeeding motion primitives. Presented methodologies are theoretically described and experimentally evaluated, showing that it is possible to generate a smooth pose trajectory out of multiple motion primitives.

In supervised machine learning, an agent is typically trained once and then deployed. While this works well for static settings, robots often operate in changing environments and must quickly learn new things from data streams. In this paradigm, known as streaming learning, a learner is trained online, in a single pass, from a data stream that cannot be assumed to be independent and identically distributed (iid). Streaming learning will cause conventional deep neural networks (DNNs) to fail for two reasons: 1) they need multiple passes through the entire dataset; and 2) non-iid data will cause catastrophic forgetting. An old fix to both of these issues is rehearsal. To learn a new example, rehearsal mixes it with previous examples, and then this mixture is used to update the DNN. Full rehearsal is slow and memory intensive because it stores all previously observed examples, and its effectiveness for preventing catastrophic forgetting has not been studied in modern DNNs. Here, we describe the ExStream algorithm for memory efficient rehearsal and compare it to alternatives. We find that full rehearsal can eliminate catastrophic forgetting in a variety of streaming learning settings, with ExStream performing well using far less memory and computation.

When operating in unstructured environments such as warehouses, homes, and retail centers, robots are frequently required to interactively search for and retrieve specific objects from cluttered bins, shelves, or tables. Mechanical Search describes the class of tasks where the goal is to locate and extract a known target object. In this paper, we formalize Mechanical Search and study a version where distractor objects are heaped over the target object in a bin. The robot uses an RGBD perception system and control policies to iteratively select, parameterize, and perform one of 3 actions - push, suction, grasp - until the target object is extracted, or either a time limit is exceeded, or no high confidence push or grasp is available. We present a study of 5 algorithmic policies for mechanical search, with 15,000 simulated trials and 300 physical trials for heaps ranging from 10 to 20 objects. Results suggest that success can be achieved in this long-horizon task with algorithmic policies in over 95% of instances and that the number of actions required scales approximately linearly with the size of the heap. Code and supplementary material can be found at http://ai.stanford.edu/mech-search.

In this study, a human-chair model was developed as the basis for a wearable chair design. A prototype chair, HUST-EC, was fabricated and evaluated. Employing the optimization under an inner point penalty function, an optimized simulation of the operating mode with the lowest chair height was implemented. The solid models were established by using the finite element analysis program embedded in Solidworks, which revealed that the support from the designed chair was steady to the user. An electromyography (EMG) test platform has been developed, consisting of four EMG sensors, a MATLAB-based acquisition software, and a loaded vest. Four healthy subjects participated in the evaluation experiment, in which EMGs were collected from the muscle groups of rectus femoris, biceps femoris, vastus medialis, and vastus lateralis under different loads and chair angles. The experimental data demonstrate that (1) the HUST-EC can greatly reduce muscle activation at a variety of loads and bending angles; (2) under the same load, the muscle activation decreases slightly with an increased bending angle; and (3) at the same bending angle, muscle activation increases slightly with an increased load. The results show that the designed chair can effectively reduce the physical burden in workers and may improve work efficiency.

In this study, a single-step additive manufacturing approach to fabricate an array of bio-inspired soft whisker-like sensors is presented. Each whisker is a damped oscillator system that is tuned to a different resonant frequency whose response is characterized through experimentation in an air medium. A 3 × 1 polymer based whisker-array is used to distinguish individual frequency components in mixed-frequency signals, achieving a form of “mechanical Fourier transform”.

Touch sensing is widely acknowledged to be important for dexterous robotic manipulation, but exploiting tactile sensing for continuous, non-prehensile manipulation is challenging. General purpose control techniques that are able to effectively leverage tactile sensing as well as accurate physics models of contacts and forces remain largely elusive, and it is unclear how to even specify a desired behavior in terms of tactile percepts. In this paper, we take a step towards addressing these issues by combining high-resolution tactile sensing with data-driven modeling using deep neural network dynamics models. We propose deep tactile MPC, a framework for learning to perform tactile servoing from raw tactile sensor inputs, without manual supervision. We show that this method enables a robot equipped with a GelSight-style tactile sensor to manipulate a ball, analog stick, and 20-sided die, learning from unsupervised autonomous interaction and then using the learned tactile predictive model to reposition each object to user-specified configurations, indicated by a goal tactile reading. Videos, visualizations and the code are available here: https://sites.google.com/view/deeptactilempc.

A manipulation technique based on optothermally generated surface bubbles is proposed in this paper. The manipulation and assembly of microstructures are completed by using bubbles. In addition, the hydrogel microstructures are also used as microrobots driven by the bubble to operate and pattern the microspheres. Considering that many materials and lasers with different wavelength have been used for generating bubbles by optothermal effects, absorptivity and transmissivity are used as indicators of selections. Besides, the size of the bubble can be controlled by the frequency and time of the laser. This technique is supposed to be applied for manipulation of cells, microparticles and microstructures.