ICRA 2017 论文列表

The twisted string actuator is an actuator that is gaining popularity in various engineering and robotics and applications. However, the fundamental limitation of actuators of this type is the uni-directional action, meaning that the actuator can contract but requires external power to return to its initial state. This paper proposes 2 novel passive extension mechanisms based on buckling effect to solve the uni-directional issue of the twisted string actuator. The proposed mechanisms are mechanically simple and compact and provide a nearly-constant extension force throughout the operation range. The constant force can fully extend the twisted string actuator with minimal loss of force during contraction. The designed mechanisms are evaluated in a series of practical tests, and their performances are compared and discussed.

Parts assembly, in a broad sense, is to make multiple objects to be in specific relative poses in contact with each other. One of the major reasons that make it difficult is uncertainty. Because parts assembly involves physical contact between objects, it requires higher precision than other manipulation tasks like collision avoidance. The key idea of this paper is to use simulation-aided physical reasoning while planning with the goal of finding a robust motion plan for parts assembly. Specifically, in the proposed approach, a) uncertainty between object poses is represented as a distribution of particles, b) the motion planner estimates the transition of particles for unit actions (motion primitives) through physics-based simulation, and c) the performance of the planner is sped up using Multi-Heuristic A* (MHA*) search that utilizes multiple inadmissible heuristics that lead to fast uncertainty reduction. To demonstrate the benefits of our framework, motion planning and physical robot experiments for several parts assembly tasks are provided.

Many standard robotic platforms are equipped with at least a fixed 2D laser range finder and a monocular camera. Although those platforms do not have sensors for 3D depth sensing capability, knowledge of depth is an essential part in many robotics activities. Therefore, recently, there is an increasing interest in depth estimation using monocular images. As this task is inherently ambiguous, the data-driven estimated depth might be unreliable in robotics applications. In this paper, we have attempted to improve the precision of monocular depth estimation by introducing 2D planar observation from the remaining laser range finder without extra cost. Specifically, we construct a dense reference map from the sparse laser range data, redefining the depth estimation task as estimating the distance between the real and the reference depth. To solve the problem, we construct a novel residual of residual neural network, and tightly combine the classification and regression losses for continuous depth estimation. Experimental results suggest that our method achieves considerable promotion compared to the state-of-the-art methods on both NYUD2 and KITTI, validating the effectiveness of our method on leveraging the additional sensory information. We further demonstrate the potential usage of our method in obstacle avoidance where our methodology provides comprehensive depth information compared to the solution using monocular camera or 2D laser range finder alone.

In this paper we present a real-time collision check algorithm based on the parallel computation capabilities of recent graphics card's GPUs. We show an effective application of the proposed algorithm to solve the task-constrained real-time motion planning problem for a redundant manipulator. We propose a proof-of-concept motion planner based on fast collision check of predicted robot motion over a given planning horizon. Obstacles are avoided exploiting the redundancy of the robot. Reactive velocities are computed for some control points placed on the robot and projected in the null space of the task Jacobian. The approach is validated through simulations in V-Rep environments and experiments on the KUKA LWR-IV 7-DoF manipulator.

Current state-of-the-art vehicle safety systems, such as assistive braking or automatic lane following, are still only able to help in relatively simple driving situations. We introduce a Parallel Autonomy shared-control framework that produces safe trajectories based on human inputs even in much more complex driving scenarios, such as those commonly encountered in an urban setting. We minimize the deviation from the human inputs while ensuring safety via a set of collision avoidance constraints. We develop a receding horizon planner formulated as a Non-linear Model Predictive Control (NMPC) including analytic descriptions of road boundaries, and the configurations and future uncertainties of other traffic participants, and directly supplying them to the optimizer without linearization. The NMPC operates over both steering and acceleration simultaneously. Furthermore, the proposed receding horizon planner also applies to fully autonomous vehicles. We validate the proposed approach through simulations in a wide variety of complex driving scenarios such as left-turns across traffic, passing on busy streets, and under dynamic constraints in sharp turns on a race track.

The PaintPot manufacturing process is a new way to create low-cost, low-profile, highly customizable potentiometers for position sensing in robotic applications. It uses widely accessible materials, requires no special expertise, and creates custom potentiometers in a variety of shapes and sizes, including curved surfaces. PaintPots offer accuracy and precision performance comparable with commercial (non-customizable) options through a calibration process that trades small computation for cost. This paper includes detailed PaintPot manufacturing and calibration processes, and experiments that validate the accuracy, precision, and lifetime performance of PaintPots, comparable to commercial sensors. We also provide a case-study application in the SMORES-EP modular robot, and show how the PaintPot process can be used to create resistive surfaces capable of sensing position in 2D on planes and spheres.

Policy search can in principle acquire complex strategies for control of robots and other autonomous systems. When the policy is trained to process raw sensory inputs, such as images and depth maps, it can also acquire a strategy that combines perception and control. However, effectively processing such complex inputs requires an expressive policy class, such as a large neural network. These high-dimensional policies are difficult to train, especially when learning to control safety-critical systems. We propose PLATO, a continuous, reset-free reinforcement learning algorithm that trains complex control policies with supervised learning, using model-predictive control (MPC) to generate the supervision, hence never in need of running a partially trained and potentially unsafe policy. PLATO uses an adaptive training method to modify the behavior of MPC to gradually match the learned policy in order to generate training samples at states that are likely to be visited by the learned policy. PLATO also maintains the MPC cost as an objective to avoid highly undesirable actions that would result from strictly following the learned policy before it has been fully trained. We prove that this type of adaptive MPC expert produces supervision that leads to good long-horizon performance of the resulting policy. We also empirically demonstrate that MPC can still avoid dangerous on-policy actions in unexpected situations during training. Our empirical results on a set of challenging simulated aerial vehicle tasks demonstrate that, compared to prior methods, PLATO learns faster, experiences substantially fewer catastrophic failures (crashes) during training, and often converges to a better policy.

Low textured scenes are well known to be one of the main Achilles heels of geometric computer vision algorithms relying on point correspondences, and in particular for visual SLAM. Yet, there are many environments in which, despite being low textured, one can still reliably estimate line-based geometric primitives, for instance in city and indoor scenes, or in the so-called “Manhattan worlds”, where structured edges are predominant. In this paper we propose a solution to handle these situations. Specifically, we build upon ORB-SLAM, presumably the current state-of-the-art solution both in terms of accuracy as efficiency, and extend its formulation to simultaneously handle both point and line correspondences. We propose a solution that can even work when most of the points are vanished out from the input images, and, interestingly it can be initialized from solely the detection of line correspondences in three consecutive frames. We thoroughly evaluate our approach and the new initialization strategy on the TUM RGB-D benchmark and demonstrate that the use of lines does not only improve the performance of the original ORB-SLAM solution in poorly textured frames, but also systematically improves it in sequence frames combining points and lines, without compromising the efficiency.

State estimation techniques for humanoid robots are typically based on proprioceptive sensing and accumulate drift over time. This drift can be corrected using exteroceptive sensors such as laser scanners via a scene registration procedure. For this procedure the common assumption of high point cloud overlap is violated when the scenario and the robot's point-of-view are not static and the sensor's field-of-view (FOV) is limited. In this paper we focus on the localization of a robot with limited FOV in a semi-structured environment. We analyze the effect of overlap variations on registration performance and demonstrate that where overlap varies, outlier filtering needs to be tuned accordingly. We define a novel parameter which gives a measure of this overlap. In this context, we propose a strategy for robust non-incremental registration. The pre-filtering module selects planar macro-features from the input clouds, discarding clutter. Outlier filtering is automatically tuned at run-time to allow registration to a common reference in conditions of non-uniform overlap. An extensive experimental demonstration is presented which characterizes the performance of the algorithm using two humanoids: the NASA Valkyrie, in a laboratory environment, and the Boston Dynamics Atlas, during the DARPA Robotics Challenge Finals.

In monocopter flight, two important parameters are required for control: angular rates and heading direction. Small monocopters fly at a very high speed (more than 600rpm), which can be out of the typical gyroscope limit. Very high speed gyroscopes do exist, but the price is high and it can only measure a single axis rotation. This paper presents an alternative approach to measure angular rates by using three accelerometers. The readings of the accelerometers are subtracted to calculate the angular rates in all three axes (x, y, and z). This paper also proposes to use the Extended Kalman Filter (EKF) to estimate the heading direction based on the magnetometer reading and the angular rates. The angular rates direction is used as the vertical direction reference. The proposed method has been applied on two setups: DC Motor setup (for quantifying the method's performance) and Monocopter setup. In the DC Motor setup, the motor encoder is used as the ground truth for the heading direction. The result is compared with the usual method of using only the magnetometer to obtain the heading direction of monocopters. The EKF result is more accurate and stable even in the presence of strong magnetic disturbances. In addition, the angle of attack and the coning angle can also be determined by the proposed method.

Tendon-driven continuum manipulators are widely used in minimally invasive surgeries, yet autonomous control has proven difficult given the uncertainty in modeling the robot's interaction with the anatomy. In this paper, we test the hypothesis that the anatomy's effect on the robotic manipulator can be approximated as a rotation of the model-based Jacobian. Two methods are described that estimate the orientation using a magnetic position sensor at the robot's distal tip, enabling task-space control in unknown, constrained environments. One method, Rα, determines the orientation of the robot's model using the measured orientation of the distal tip. In the second approach, RΔx, an unscented Kalman filter transforms the tip's measured change in position into an orientation estimate. These methods are validated on the manipulator in free space at three pre-rotated positions (α = 0°, -45°, -90°) and compared to model-based control and model-less control in an autonomous trajectory-following task through an anatomically accurate silicone phantom of a human lung. The results show that by rotating the model-based Jacobian through either of the two methods presented, feedback control successfully navigates the continuum manipulator farther into the lung phantom than static model-based control and model-less control.

It is unclear what triggers the behavior of active shooters, but their consequences are severe. There is opportunity to implement an automated response system capable of delivery guidance to evacuees' to aide people to safety. Optimizing guidance delivery is challenging because active shooter incidents evolve quickly and are unpredictable. In this paper, we develop a new problem formulation specific for active shooter events by utilizing the structure of each room. To effectively solve this problem, a divide-and-conquer approach is deployed to split evacuees into groups. The egress routes are decomposed and coordinated for each group are optimized using stochastic dynamic programming. Numerical testing and simulation show our solution with a safe room the solution fast relevant to shooting events and effectively.

This paper presents a sequential optimum design approach for estimating the parameters of an atmospheric dispersion process model based on measurement data gathered by a team of cooperating sensor-equipped UAVs. Locally optimal waypoint sequences that account for each UAV's possibly heterogeneous motion dynamics are computed by minimizing a suitable optimality criterion. Following these waypoints, the UAVs cooperatively maximize the information gain of the acquired measurements. A decentralized data-driven online control scheme is proposed that couples parameter estimation, waypoint calculation, and vehicle control and enables the UAVs to adaptively observe the dynamic process and iteratively improve the parameter estimate. Simulations demonstrate the effectiveness of the proposed scheme in reducing the error between the estimated and the true dispersion model parameters compared to non-adaptive sensing strategies. In addition, the effect of using different optimality criteria, different numbers and types of UAVs as well as two options for decentralizing the waypoint calculation are investigated.

When robots perform manipulation tasks, they need to determine their own movement, as well as how to grasp and release an object. Reasoning about the motion of the robot and the object simultaneously leads to a multi-modal planning problem in a high-dimensional configuration space. In this paper we propose an asymptotically optimal manipulation planner. Our approach extends optimal sampling-based roadmap planners to efficiently explore the configuration space of the robot and the object. We prove probabilistic completeness and global, asymptotic optimality. Extensive simulations of a typical pick-and-place scenario show that our approach significantly outperforms a (nonoptimal) state-of-the-art approach. We implemented our planner on a real manipulator and were able to compute high quality solutions in less than a second.

Persistent coverage aims to maintain a certain coverage level over time in an environment where such level deteriorates. This level can be associated to temperature, dust or sensor information. We propose an algorithmic solution in which each robot locally finds the best paths and coverage actions to keep the desired coverage level over the whole environment. Using Fast Marching Methods, optimal paths are computed in terms of coverage quality, while keeping a safety distance to obstacles. Additionally, our solution enables a computationally efficient evaluation of a list of potential trajectories, allowing us to choose the one that mostly improves the coverage along the whole path. The combination of this algorithm with a Dynamic Window navigation makes our approach competitive in terms of flexibility and robustness in changing environments with existing solutions. Finally, we also propose a coverage action controller, locally computed and optimal, that makes the robots maintain the coverage level of the environment significantly close to the objective. Simulations and real experiments validate the whole approach.

This study presents a soft robotic gripper for grasping irregular objects. The optimal design is based on the proposed topology optimization and size optimization methods with the objective to maximize the mechanical advantage (MA, which is defined as the ratio of output force to the input force) of the analyzed compliant mechanism. The optimal design is prototyped using silicon rubber material. Experimental tests including MA test, geometric advantage (GA, which is defined as the ratio of output displacement to the input displacement) test, adaptability test, and grasping test are carried out to investigate the design. A performance index has also been proposed to evaluate the grasping performance of the grippers. The results show the developed gripper is with the highest performance index, which represents the developed gripper is with better adaptability, faster response, higher payload and stability in overall.

Many people with motor disabilities are unable to complete activities of daily living (ADLs) without assistance. This paper describes a complete robotic system developed to provide mobile grasping assistance for ADLs. The system is comprised of a robot arm from a Rethink Robotics Baxter robot mounted to an assistive mobility device, a control system for that arm, and a user interface with a variety of access methods for selecting desired objects. The system uses grasp detection to allow previously unseen objects to be picked up by the system. The grasp detection algorithms also allow for objects to be grasped in cluttered environments. We evaluate our system in a number of experiments on a large variety of objects. Overall, we achieve an object selection success rate of 88% and a grasp detection success rate of 90% in a non-mobile scenario, and success rates of 89% and 72% in a mobile scenario.

We present an algorithm that generates walking motions for quadruped robots without the use of an explicit footstep planner by simultaneously optimizing over both the Center of Mass (CoM) trajectory and the footholds. Feasibility is achieved by imposing stability constraints on the CoM related to the Zero Moment Point and explicitly enforcing kinematic constraints between the footholds and the CoM position. Given a desired goal state, the problem is solved online by a Nonlinear Programming solver to generate the walking motion. Experimental trials show that the algorithm is able to generate walking gaits for multiple steps in milliseconds that can be executed on a real quadruped robot.

This paper deals with the problem of active sensing control for nonlinear differentially flat systems. The objective is to improve the estimation accuracy of an observer by determining the inputs of the system that maximise the amount of information gathered by the outputs over a time horizon. In particular, we use the Observability Gramian (OG) to quantify the richness of the acquired information. First, we define a trajectory for the flat outputs of the system by using B-Spline curves. Then, we exploit an online gradient descent strategy to move the control points of the B-Spline in order to actively maximise the smallest eigenvalue of the OG over the whole planning horizon. While the system travels along its planned (optimized) trajectory, an Extended Kalman Filter (EKF) is used to estimate the system state. In order to keep memory of the past acquired sensory data for online re-planning, the OG is also computed on the past estimated state trajectories. This is then used for an online replanning of the optimal trajectory during the robot motion which is continuously refined by exploiting the state estimation obtained by the EKF. In order to show the effectiveness of our method we consider a simple but significant case of a planar robot with a single range measurement. The simulation results show that, along the optimal path, the EKF converges faster and provides a more accurate estimate than along other possible (non-optimal) paths.