ICRA 2016 Papers

The dynamics of many large-scale robotic formation systems, including structured systems as well as some random scale-free networks of agents, can be accurately described using fractional-order differential equations. A fractional-order differential equation can contain derivative terms with non-integer order, e.g., the one-half derivative. This paper demonstrates that the fractional order of the dynamics of a system may be a potentially powerful new way to monitor the operational status of such systems. When the order of the system changes, it can indicate an important change in the status of the system. Integer-order models will never exhibit a change in order because the order is dictated by a natural first principle and the structure of the system. For this reason, traditional health monitoring tools essentially focus on identifying parameter variations in a mathematical description of the system, but not changes in order. When fractional-order models are considered, the infinite number of possible real-valued orders between any two integer orders may capture essential changes in the system's dynamics. This paper provides an example of such changes, provides a theoretical justification for the approach, and explores possible limitations to the approach.

Cloud Robotics techniques based on Learning from Demonstrations suggest promising alternatives to manual programming of robots and autonomous vehicles. One challenge is that demonstrated trajectories may vary dramatically: it can be very difficult, if not impossible, for a system to learn control policies unless the trajectories are clustered into meaningful consistent subsets. Metric clustering methods, based on a distance measure, require quadratic time to compute a pairwise distance matrix and do not naturally distinguish topologically distinct trajectories. This paper presents an algorithm for topological clustering based on relative persistent homology, which, for a fixed underlying simplicial representation and discretization of trajectories, requires only linear time in the number of trajectories. The algorithm incorporates global constraints formalized in terms of the topology of sublevel or superlevel sets of a function and can be extended to incorporate probabilistic motion models. In experiments with real automobile and ship GPS trajectories as well as pedestrian trajectories extracted from video, the algorithm clusters trajectories into meaningful consistent subsets and, as we show in an experiment with ship trajectories, results in a faster and more efficient clustering than a metric clustering by Fréchet distance.

We present a method for estimating the contact event between sensor-free active subtracks, named flippers, of an articulated tracked vehicle (ATV) and the terrain surface. The main idea is to consider both the moving base link and unexpected collisions dynamics as disturbances of the flipper dynamics. On this basis we extend the generalized momenta fault detection and isolation (FDI) method to compute the residual dynamics of the flippers, without resorting to additional sensory information. Under the hypothesis that the residual signal presents disturbance patterns that can be discriminated by those generated by unexpected collisions of the flippers with the ground, we apply a classification method to recover the contact event. The wavelet packet transform is used to decompose the signal and to generate a feature space for the residual, from the different subbands. Finally, sparse SVM, based on feature selection discriminates the contact signal.

Tactile manipulation is the ability to control objects in real-time using the sense of touch. Here we examine tactile manipulation from the perspective of active touch with a biomimetic tactile sensor, which combines tactile perception with control of sensor location. Experiments are performed with the tactile fingertip mounted as end effector to a robot arm, to manipulate (roll) a cylinder in contact with the fingertip. Performance is validated with offline (cross-validation) and online (real-time operation) assessments. Location perception is finer than the sensor resolution, leading to superresolved tactile manipulation along a complex trajectory. However, the original methods were non-robust to large unknown disturbances of object location, necessitating modification of the perceptual process to diminish prior beliefs relative to past posterior beliefs. In consequence robust and accurate tactile manipulation was attained. In general, it appears there is a trade-off between the responsiveness to unknown change and manipulation accuracy, which must be set appropriately for each task.

We present a new Active Learning approach for classifying objects from streams of 3D point cloud data. The major problems here are the non-uniform occurrence of class instances and the unbalanced numbers of samples per class. We show that standard online learning methods based on decision trees perform comparably bad for such data streams, which are however particularly relevant for mobile robots that need to learn semantics persistently. To address this, we use Mondrian forests (MF), a recent online learning algorithm that is independent on the data order. We present an extension of that algorithm and show that MF are less overconfident than standard Random Forests. In experiments on the KITTI benchmark, we show that this leads to a substantially improved classification performance for data streams, rendering our approach very attractive for lifelong robot learning applications.

We have proposed a multimodal sensing with optical device and its application to robotic grasp control. The proposed device provides both tactile and proximity information with high spatial resolution on the same scale of coordinates, that is necessary while grasping small and complicated-shaped objects using a robotic hand. In our proposed device, tactile information is obtained as infrared image through a light conductive plate, which allows the capture of a tactile image of high spatial resolution using a camera. Proximity information is detected based on stereo matching of a pair of images obtained by two cameras. By using a compound-eye camera, we constructed a compact device so that the sensor can be mounted on the robotic hand. Through experiments with 6 DoF robotic arm and robotic hand equipped with the proposed sensing device, we show that the robotic motion for searching, approaching and grasping can be controlled based on sensor information obtained from only the proposed device. In addition, our combined sensing approach enables an adaptive grasping motion of the robotic hand.

In this paper we present an algorithm to compute risk averse policies in Markov Decision Processes (MDP) when the total cost criterion is used together with the average value at risk (AVaR) metric. Risk averse policies are needed when large deviations from the expected behavior may have detrimental effects, and conventional MDP algorithms usually ignore this aspect. We provide conditions for the structure of the underlying MDP ensuring that approximations for the exact problem can be derived and solved efficiently. Our findings are novel inasmuch as average value at risk has not previously been considered in association with the total cost criterion. Our method is demonstrated in a rapid deployment scenario, whereby a robot is tasked with the objective of reaching a target location within a temporal deadline where increased speed is associated with increased probability of failure. We demonstrate that the proposed algorithm not only produces a risk averse policy reducing the probability of exceeding the expected temporal deadline, but also provides the statistical distribution of costs, thus offering a valuable analysis tool.

We address the question of how resource-aware concepts can be utilized in motion planning algorithms. Resource-awareness facilitate better resource allocation on global system level, e.g. when a humanoid robot needs to distribute and schedule a wide variety of concurrent algorithms. We present a motion planning approach that employs self-monitoring concepts in order to identify the difficulty of the planning problem. Resources are requested dynamically and adapted based on problem difficulty and current planning progress. We show how dynamic adaptation of resource allocation on algorithmic level can reduce the system workload as compared to static resource allocation while meeting Quality of Service (QoS) measures such as average workload or efficiency. We evaluate our approach both in several synthetic setups with varying difficulty and with the humanoid robot ARMAR-4.

We present several modifications to the previously proposed MSPP algorithm that can speed-up its execution considerably. The MSPP algorithm leverages a multi-scale representation of the environment in n dimensions encoded in tree structure constructed by recursive dyadic partitioning of the search space. We first present a new method to compute the graph neighbors in order to reduce the complexity of each iteration, from O(|V|2) to O(|V| log |V|). We then show how to delay expensive intermediate computations until we know that new information will be required, hence saving time by not operating on information that is never used during the search. Finally, we present a way to remove the very expensive need to calculate a full multi-scale map with the use of sampling and derive an upper bound on the probability of failure as a function of the number of samples.

We present a novel algorithm for real-time collision-free navigation between elliptical agents. Each robot or agent is represented using a tight-fitting 2D ellipse in the plane. We extend the reciprocal velocity obstacle formulation by using conservative linear approximations of ellipses and derive sufficient conditions for collision-free motion based on low-dimensional linear programming. We use precomputed Minkowski Sum approximations for real-time and conservative collision avoidance in large multi-agent environments. Finally, we present efficient techniques to update the orientation to compute collision-free trajectories. Our algorithm can handle thousands of elliptical agents in real-time on a single core and provides significant speedups over prior algorithms for elliptical agents. We compare the runtime performance and behavior with circular agents on different benchmarks.

We present a decentralized algorithm for group-based coherent and reciprocal multi-agent navigation. In addition to generating collision-free trajectories for each agent, our approach is able to simulate macroscopic group movements and proxemic behaviors that result in coherent navigation. Our approach is general, makes no assumptions about the size or shape of the group, and can generate smooth trajectories for the agents. Furthermore, it can dynamically adapt to obstacles or the behavior of other agents. The additional overhead of generating proxemic group behaviors is relatively small and our approach can simulate hundreds of agents in real-time. We highlight its benefits on different benchmarks.

We present a probabilistic model for joint representation of several sensory modalities and action parameters in a robotic grasping scenario. Our non-linear probabilistic latent variable model encodes relationships between grasp-related parameters, learns the importance of features, and expresses confidence in estimates. The model learns associations between stable and unstable grasps that it experiences during an exploration phase. We demonstrate the applicability of the model for estimating grasp stability, correcting grasps, identifying objects based on tactile imprints and predicting tactile imprints from object-relative gripper poses. We performed experiments on a real platform with both known and novel objects, i.e., objects the robot trained with, and previously unseen objects. Grasp correction had a 75% success rate on known objects, and 73% on new objects. We compared our model to a traditional regression model that succeeded in correcting grasps in only 38% of cases.

Robots are increasingly being used in situations such as search and rescue that require robust navigation capabilities, potentially in areas with little or no GPS or other high-quality localization information. As more robots are used in these scenarios, it becomes viable to collaborate between heterogeneous types of robots to leverage their individual strengths while minimizing their weaknesses. More specifically, in a scenario involving unmanned ground and aerial vehicles (UGV, UAV), the ground robot can contribute its high payload capacity to provide computational resources and high accuracy sensors while the aerial robot can bring its high mobility and capability to traverse obstacles to the team. However, in order for the team to benefit from these capabilities, it must be capable of generating a plan for both robots that allows them to collaboratively localize when necessary. Our approach to this problem is to combine a recently developed state lattice planner using controller-based motion primitives (SLC) with planning using adaptive dimensionality (PAD). The SLC planner allows for robust navigation using a wide variety of sensors including in areas with no or limited high-quality localization information while the PAD planner allows us to expand beyond a single robot and generate plans for a team of robots operating in a high dimensional space. We present our results to this combined approach for a UGV/UAV team operating indoors in areas with limited visual features.

This paper presents a novel method for using volumetric information in the C-space for generating collision-free paths for robotic manipulators in the presence of obstacles. The method is based on the RRT paradigm, i.e., incrementally building trees in C-space in order to connect the initial configuration to the final one. Bubbles of free C-space are used within a carefully modified extend procedure to add lengthy edges to the tree. In particular, we propose an expanded version of the bubble, which enables even more efficient exploration. The method is validated and compared to other planning algorithms within a simulation study that includes several types of manipulators and obstacle scenarios.

In this paper, we propose an integrated optimal planning and control technique for optical manipulation of a group of cells beyond the field of view. The proposed technique consists of a planning layer where a model predictive control is utilized to obtain a virtual input for manipulation of cells, and a control layer where the desired inputs for laser beams and motorized stage are obtained by means of the control allocation technique. Unlike existing methods on simultaneous manipulation of laser and motorized stage which the desired inputs for lasers and stage are given separately, here, the planning and control layers are integrated together to obtain optimal inputs for both laser and motorized stage. Finally, experimental results are presented to illustrate the performance of the proposed method.

We describe a method for learning dexterous manipulation skills with a pneumatically-actuated tendon-driven 24-DoF hand. The method combines iteratively refitted time-varying linear models with trajectory optimization, and can be seen as an instance of model-based reinforcement learning or as adaptive optimal control. Its appeal lies in the ability to handle challenging problems with surprisingly little data. We show that we can achieve sample-efficient learning of tasks that involve intermittent contact dynamics and under-actuation. Furthermore, we can control the hand directly at the level of the pneumatic valves, without the use of a prior model that describes the relationship between valve commands and joint torques. We compare results from learning in simulation and on the physical system. Even though the learned policies are local, they are able to control the system in the face of substantial variability in initial state.

The observation of robotic assembly tasks is required as feedback for decisions and adaption of the task execution on the current situation. A sequential Monte Carlo observation algorithm is proposed, which uses a fast and accurate collision detection algorithm as a reference model for the contacts between complex shaped parts. The main contribution of the paper is the extension of the classic random motion model in the propagation step with sampling methods known from the domain of probabilistic roadmap planning in order to increase the sample density in narrow passages of the configuration space. As a result, the observation performance can be improved and a risk of sample impoverishment reduced. Experimental validation is provided for a peg-in-hole task executed by a lightweight-robot arm equipped with joint torque sensors.

Nanoforce sensors using passive magnetic springs associated to a macroscopic seismic mass are known to be a possible alternative to force sensors based on elastic microstructures like Atomic Force Microscopes if the nanoforces that have to be measured are characterized by a bandwidth limited to a few Hertz. The estimation of the unknown force applied to the seismic mass is based on the deconvolution of the noisy measurement of the mass displacement which has an under-damped dynamic. Despite their high performances in terms of linearity, resolution and measurement range, such force sensors are extremely sensitive to low frequency environmental mechanical disturbances. They are also sensitive to the temperature evolution of the ambient air. The evaluation, modeling and compensation of such environmental disturbances have to be specifically studied in the context of magnetic springs associated to a macroscopic seismic mass. This article presents an estimation and a passive compensation strategy of the low frequency and non-stationary mechanical disturbances that is based on a differential principle. This approach is applied to a nanoforce sensor based on diamagnetic levitation developed in the last decade and gives an uncertainty below the nanonewton level.

Movement primitives are concise movement representations that can be learned from human demonstrations, support generalization to novel situations and modulate the speed of execution of movements. The speed modulation mechanisms proposed so far are limited though, allowing only for uniform speed modulation or coupling changes in speed to local measurements of forces, torques or other quantities. Those approaches are not enough when dealing with general velocity constraints. We present a movement primitive formulation that can be used to non-uniformly adapt the speed of execution of a movement in order to satisfy a given constraint, while maintaining similarity in shape to the original trajectory. We present results using a 4-DoF robot arm in a minigolf setup.