ICRA 2018 论文列表

A swarm of randomly moving miniature robots is an effective solution for the exploration of unknown terrains. However, the deployment of a swarm of miniature robots poses two challenges: finding an adequate locomotion strategy for fast exploration and obstacles negotiation; and implementing simple design and control solutions suited for mass manufacturing. Here, we tackle these challenges by developing a new soft robot with a minimalistic design and a simple control strategy that can randomly propel itself above obstacles and roll on the ground upon landing. The robot is equipped with two propellers that are periodically activated to jump, a soft cage that protects the robot from impacts and allows to passively roll on the ground, and a passive self-righting mechanism for repetitive jumps. The minimalistic control and design reduce the complexity of the mechanics and electronics and are instrumental to the production of a large number of robots. In the paper, the key design aspects of the robot are discussed, the locomotion of a single prototype is experimentally characterized, and improvements of the system for future swarm operations are discussed.

Soft actuators are ideal candidates for wearable biomedical devices, their inherent compliance, robustness, lightweight and the possibility to be washable take advantage over rigid actuators. Thus, a soft pneumatic fabric-polymer bending actuator as a base component for a robotic device for lymphedema treatment is reported in this work. The actuator is composed of two mechanical elements, one made of fabric and the other one made of a hyperelastic polymer which is stuck on the fabric element. The fabric element is designed and fabricated with a curved shape longer than the polymer element, that is a hyperelastic beam. To assemble both elements, the fabric element was folded before sticking in order to match the length of the polymer beam. Once the air is pumped into the fabric, it bends towards its original curved shape. Once the air is removed, the hyperelastic beam allows the actuator to recover its initial position. This actuator is capable of exerting compression and lateral force on a human arm mimicking manual lymphatic drainage. A mathematical model is presented which is in good agreement with the experimental data, it could serve to predict the actuator motion. An end-tip free bending displacement of about 2.2 cm and a bending force of about 0.35 N were achieved at 12.5 kPa. A proof-of-concept system for lymphedema treatment is presented as well.

In this work, we present an approach to planning for humanoid mobility. Humanoid mobility is a challenging problem, as the configuration space for a humanoid robot is intractably large, especially if the robot is capable of performing many types of locomotion. For example, a humanoid robot may be able to perform such tasks as bipedal walking, crawling, and climbing. Our approach is to plan for all these tasks within a single search process. This allows the search to reason about all the capabilities of the robot at any point, and to derive the complete solution such that the plan is guaranteed to be feasible. A key observation is that we often can roughly decompose a mobility task into a sequence of smaller tasks, and focus planning efforts to reason over much smaller search spaces. To this end, we leverage the results of a recently developed framework for planning with adaptive dimensionality, and incorporate the capabilities of available controllers directly into the planning process. The resulting planner can also be run in an interleaved fashion alongside execution so that time spent idle is much reduced.

This paper presents a force-sensorless collision detection approach based on virtual contacts. A series of indices based on virtual instantaneous powers are introduced associating to several virtual contacts defined on corresponding links of a manipulator. A possible contact link can be determined by comparing the introduced indices without any force or tactile sensors. Simulation results of 2D collision detection tasks by using a three DOF (Degree of Freedom) planar manipulator and 3D collision detection tasks by using a six DOF spatial manipulator validate the effectiveness of the proposed collision detection approach.

This paper discusses the design method of a self-propelled catheter with thin diameter of 2.6mm, aimed for moving and steering in the bronchi. As the basic configuration, the proposed catheter is composed of an inner tube and McKibben artificial muscles. It can propel by generating travelling waves by a single supply line driven by pneumatics. Since the inner tube equips with orifices of different size, several McKibben chambers can deform with the time delay. We also propose the method to steer the direction at the branch pipe, whose angle can be adjusted by PWM control of pressuring. Besides, we also introduce the design method of how to deal with the pipes with different diameters. The validity of these proposed methods are verified by the prototype through the experiments in the bronchial model.

Indoor operation of unmanned aerial vehicles (UAV s) poses many challenges due to the lack of GPS signal and cramped spaces. The presence of obstacles in an unfamiliar environment requires reliable state estimation and active algorithms to prevent collisions. In this paper, we present a teleoperated quadrotor UAV platform equipped with an onboard miniature computer and a minimal set of sensors for this task. The platform is capable of highly accurate state-estimation, tracking of desired velocity commanded by the user and ensuring collision-free navigation. The robot estimates its linear velocity through a Kalman filter integration of inertial and optical flow (OF) readings with corresponding distance measurements. An RGB-D camera serves the purpose of providing visual feedback to the operator and depth measurements to build a probabilistic, robo-centric obstacle model, allowing the robot to avoid collisions. The platform is thoroughly validated in experiments in an obstacle rich environment.

In modern warehouses, robots are being deployed to perform complex tasks such as fetching a set of objects from various locations in a warehouse to a docking station. This requires a careful task allocation along with route planning such that the total distance traveled (cost) is minimized. The number of tasks that can be performed by a robot on a single route depends on the maximum capacity of the robot and the combined weight of the objects it picks on the route. This task allocation problem is an instance of the Capacity-constrained Vehicle Routing Problem (CVRP), which is known to be NP-hard. Although, there exist a number of heuristics that provide near-optimal solutions to a CVRP instance, they do not scale well with the task size (number of nodes). In this paper, we present a heuristic, called nearest-neighbor based Clustering And Routing (nCAR)(nCAR), which has better execution time compared to the state-of-the-art heuristics. Also, our heuristic reduces cost of the solutions when there are a large number of nodes. We compare the performance of nCARnCAR with the Google OR-Tools and found a speedup of 6 in runtime when task size is 2000. Though OR-Tools provides a low-cost solution for small number of tasks, it's execution time and number of routes is 1.5 times that of nCAR.

In order to build a stable and reliable system for the Amazon Robotics Challenge we went through a detailed study of the performance and system requirements based on the rules and our past experience of the challenge. The challenge was to build a robot that integrates grasping, vision, motion planning, among others, to be able to pick items from a shelf to specific order boxes. This paper presents the development process including component selection, module designs, and deployment. The resulting robot system has dual 6 degrees of freedom industrial arms mounted on fixed bases, which in turn are mounted on a calibrated table. The robot works with a custom-designed top-open extendable shelf. The vision system uses multiple stereo cameras mounted on a fixed calibrated frame. Feature-based comparison and machine-learning based matching are used to identify and determine item pose. The gripper system uses suction cup and the grasping strategy is pick from the top. Error recovery strategies were also implemented to ensure robust performance. During the competition, the robot was able to pick all target items with the shortest amount of time.

This paper presents a novel Nonlinear Model Predictive Control (NMPC) scheme for underwater robotic vehicles operating in a constrained workspace including static obstacles. The purpose of the controller is to guide the vehicle towards specific way points. Various limitations such as: obstacles, workspace boundary, thruster saturation and predefined desired upper bound of the vehicle velocity are captured as state and input constraints and are guaranteed during the control design. The proposed scheme incorporates the full dynamics of the vehicle in which the ocean currents are also involved. Hence, the control inputs calculated by the proposed scheme are formulated in a way that the vehicle will exploit the ocean currents, when these are in favor of the way-point tracking mission which results in reduced energy consumption by the thrusters. The performance of the proposed control strategy is experimentally verified using a 4 Degrees of Freedom (DoF) underwater robotic vehicle inside a constrained test tank with obstacles.

The ability to predict the duration of an activity can enable a robot to plan its behaviors ahead, interact seamlessly with other humans, by coordinating its actions, and allocate effort and resources to tasks that are time-constrained or critical. Despite its usefulness, models that examine the temporal properties of an activity remain relatively unexplored. In the current paper we present, to the best of our knowledge, the first method that can estimate temporal properties of an activity by observation. We evaluate it on three use-cases (i) wiping a table, (ii) chopping vegetables and (iii) cleaning the floor, using ground truth data from real demonstrations, and show that it can make predictions with high accuracy and little training. In addition, we investigate different methods to approximate the progress of each task, and demonstrate how a model can generalize, by reusing part of it in different activities.

In recent years, a number of publications described approaches for model-based manipulation skills and their applicability to a variety of robot tasks-be it assembly, industrial robotics in general, or service robotics. These approaches roughly follow the same pattern: They model robot task description based on the Task Frame Formalism, the Task Function Approach, or iTaSC. They model coordination mechanisms in form of statecharts or Petri nets. And almost all models are accompanied by domain-specific languages (DSLs) that facilitate creating applications based on those models. While one advantage of using models is their reusability across applications, how to explicitly model the reuse itself has not been fully addressed by these publications. Our paper contributes to this field of research by investigating how reuse can be explicitly modeled using prototype-based inheritance. We base our model on iTaSC and provide a simple yet effective DSL for populating the model and creating applications. We demonstrate our approach by creating a comprehensive library of skills, and by showing the use, reuse and incremental refinement of skills for diverse industrial assembly applications.

We propose a model-based control framework for multi-arm manipulation of a rigid object subject to external disturbances. The control framework, based on projected inverse dynamics, decomposes the control law into constrained and unconstrained subspaces. Unconstrained components accomplish the motion task with a desired 6-DOF Cartesian impedance behaviour against external disturbances. Meanwhile, the constrained component enforces contact and friction constraints by optimising for contact forces within the constrained subspace. External disturbances are explicitly compensated for without using force/torque sensors at the contact points. The approach is evaluated on a dual-arm platform manipulating a rigid object while coping with unknown object dynamics and human interaction.

This paper addresses the problem of underwater robotic vehicle navigation relative to moving or stationary sea ice. A brief review of previously-reported under-ice navigation methods is given, as well as a brief motivation for the use of under-ice robotic vehicles with precision navigation capabilities. We then describe our proposed approach, which employs two or more satellite navigation beacons atop the sea ice along with other precision vehicle and ship mounted navigation sensors to estimate vehicle, ice, and ship states by means of an Extended Kalman Filter. Navigation results for a simulated 7.6 km under ice survey are presented for varying satellite beacon separation. Preliminary analysis suggests that for the simulated sensors, vehicle trajectory, and ice velocities, the proposed method can accurately estimate vehicle position up to 1.2 km from the deployment ship given sufficient satellite beacon separation. Decreased beacon separation results in divergence of the vehicle's position estimate at large standoff distances from the ship. We conclude with suggestions for future improvements.

In order to make robot programming more easy and immediate, walk-through programming techniques can be exploited. However, a modification of a portion of the trajectory usually means to execute the path from the beginning. In this paper we propose a passivity-based framework to modify the trajectory online, manually driving the robot throughout the desired correction. The system follows the initial trajectory, encoded with Dynamical Movement Primitives, by setting high gains in the admittance control. When the human operator grabs the end-effector, the robot becomes compliant and the user can easily teach the desired correction, until he/she releases it at the end of the modification. Finally, the correction is optimally joined to the initial trajectory, restarting the path tracking. To avoid unsafe behaviors, the variation of the admittance parameters is performed exploiting energy tanks, in order to preserve the passivity of the interaction.

When designing control strategies for differential-drive mobile robots, one standard tool is the consideration of a point at a fixed distance along a line orthogonal to the wheel axis instead of the full pose of the vehicle. This abstraction supports replacing the non-holonomic, three-state unicycle model with a much simpler two-state single-integrator model (i.e., a velocity-controlled point). Yet this transformation comes at a performance cost, through the robot's precision and maneuverability. This work contains derivations for expressions of these precision and maneuverability costs in terms of the transformation's parameters. Furthermore, these costs show that only selecting the parameter once over the course of an application may cause an undue loss of precision. Model Predictive Control (MPC) represents one such method to ameliorate this condition. However, MPC typically realizes a control signal, rather than a parameter, so this work also proposes a Parametric Model Predictive Control (PMPC) method for parameter and sampling horizon optimization. Experimental results are presented that demonstrate the effects of the parameterization on the deployment of algorithms developed for the single-integrator model on actual differential-drive mobile robots.

Recurrent Neural Networks (RNNs) demonstrated advantages on control precision, system robustness and computational efficiency, and have been widely applied to redundant manipulator control optimization. Existing RNN control schemes locally optimize trajectories and are efficient and reliable on obstacle avoidance. However, for motion planning, they suffer from local minimum and do not have planning completeness. This work explained the cause of the planning incompleteness and addressed the problem with a novel RNN control scheme. The paper presented the proposed method in detail and analyzed the global stability and the planning completeness in theory. The proposed method was compared with other three control schemes on the precision, the robustness and the planning completeness in software simulation and the results shows the proposed method has improved precision and robustness, and planning completeness.

This paper presents a novel magnetic anchored and steered camera robot intended for minimally invasive surgery (MIS), particularly for single port access (SPA) surgery. The design aims to achieve both compactness and a planar pan/tilt workspace (instead of hemispheric) to lower robot footprint in vertical space. Robot comprises two 6mm×6mm diametrically magnetized internal permanent magnets (IPMs) fixed at either ends of a small cylindrical capsule, with camera module and a 45°mirror capped inside capsule. As such, camera view orientation can be steered in 2-DOF across range of 180° tilt and 360° panning, all within a planar workspace close to surface of anchor. Using only two small IPMs for all necessary functions (anchoring, translation along intra-abdominal surface, and steering) reduces bulk and length of robot. The robot is investigated first by finite element methods. Theoretical models for both tilting and panning were then built based on FEM results. The models are evaluated and verified by checking its predictions in benchtop experiments. Ex vivo evaluations was also utilized to prove feasibility of device in environment similar to human anatomy. Overall, the camera robot prototype is compact (4cm length; 7mm diameter), lightweight (3.6g), motor-free, and allow view orientation control (tilting and panning) in a planar workspace. Minimal footprint in vertical space is ideal for many MIS applications, where vertical space is extremely limited.

Thanks to the growing interest in soft robotics, hydropneumatics and inflatable system dynamics are attracting renewed attention from the scientific community. Typical fluidic systems are composed of several chambers and require a complex and bulky network of active components for their control. This paper presents a novel approach to fluidic actuation, which consists in the co-design of both the mechanical parameters of the system and of custom input signals, to enable the elicitation of different behaviors of the system with fewer control components. The principle is presented in theory and simulation and then experimentally validated through the application to a case study, an in-pipe inchworm-like robot. It is shown that it is possible to obtain forward and backward movements by modulating a unique input.

In this paper, we present a novel nonparametric motion flow model that effectively describes a motion trajectory of a human and its application to human robot cooperation. To this end, motion flow similarity measure which considers both spatial and temporal properties of a trajectory is proposed by utilizing the mean and variance functions of a Gaussian process. We also present a human robot cooperation method using the proposed motion flow model. Given a set of interacting trajectories of two workers, the underlying reward function of cooperating behaviors is optimized by using the learned motion description as an input to the reward function where a stochastic trajectory optimization method is used to control a robot. The presented human robot cooperation method is compared with the state-of-the-art algorithm, which utilizes a mixture of interaction primitives (MIP), in terms of the RMS error between generated and target trajectories. While the proposed method shows comparable performance with the MIP when the full observation of human demonstrations is given, it shows superior performance when partial trajectory information is given.