IROS 2016 论文列表

In this paper we present a real-time approach to stitch large-scale aerial images incrementally. A monocular SLAM system is used to estimate camera position and attitude, and meanwhile 3D point cloud map is generated. When GPS information is available, the estimated trajectory is transformed to WGS84 coordinates after time synchronized automatically. Therefore, the output orthoimage retains global coordinates without ground control points. The final image is fused and visualized instantaneously with a proposed adaptive weighted multiband algorithm. To evaluate the effectiveness of the proposed method, we create a publicly available aerial image dataset with sequences of different environments. The experimental results demonstrate that our system is able to achieve high efficiency and quality compared to state-of-the-art methods. In addition, we share the code on the website with detailed introduction and results.

Current traffic rule handling in autonomous driving relies on non-scalable methods such as explicitly hard-coding sets of logical statements to evaluate rules. This is sufficient for early prototypes but is not scalable for larger systems designed to work in arbitrary locations. Such methods can also become problematic when traversing different geographical entities with changing traffic rules. Additionally, they do not adequately convey the uncertainty of the traffic rule evaluation stemming from uncertainty in sensor measurements. We propose an exchangeable traffic rules module which takes a knowledge base of sentences in first-order logic as input. These sentences consist of a limited number of high-level queries which are independent of local jurisdiction (e.g. can I turn right?). The knowledge base is then compiled into a potentially large logic graph. However, only a relatively small subset of the knowledge base is relevant for specific road geometries (e.g. some of the rules applicable to an intersection is not relevant for a T-junction). Since detailed road maps are available for autonomous driving, this information can be used to resolve these subsets in the knowledge base which only require map knowledge. Therefore, map information can be used to convert a single, large logic graph into a set of smaller, map-optimized logic graphs pruned for specific road geometries. The optimized graphs are then converted into Bayesian networks to facilitate probabilistic inference. Experiments were conducted using a traffic and scenario simulation framework. The results demonstrate a significant improvement in performance when using map-optimized logic graphs over a traditional first-order logic knowledge base.

Environment models are the primary matter to autonomous decisions for mobile robots, and also to cooperation within teams of robots that operate in the same environment. The decisions to take within a robot or a robot team relate to motions, perceptions and communications: various types of environment models are therefore required to evaluate and plan these actions. While the literature abounds with approaches to environment modeling using data perceived by the robots, very few work tackle the problem of managing such models within a team of robots. Managing environment models implies first defining the proper data structures and associated mechanisms that allow both their efficient update and use by the decisional processes that require them, and second ensuring the models consistency as the robots evolve. This article presents the definition of a framework dedicated to the managing of environment models within a robot team. It establishes the principles that govern the framework design, and illustrates them throughout some examples.

Surgeons and robots typically use different approaches for bone milling. Surgeons adjust their speed and tool incidence angle constantly, which enables them to efficiently mill porous bone. Surgeons also adjust milling parameters such as speed and depth of cut throughout the procedure based on proximity to sensitive structures like nerves and blood vessels. In this paper we use image-based bone density estimates and segmentations of vital anatomy to make a robot mill more like a surgeon and less like an industrial computer numeric controlled (CNC) milling machine. We produce patient-specific plans optimizing velocity and incidence angles for spherical cutting burrs. These plans are particularly useful in bones of variable density and porosity like the human temporal bone. They result in fast milling in non-critical areas, reducing overall procedure time, and lower forces near vital anatomy. We experimentally demonstrate the algorithm on temporal bone phantoms and show that it reduces mean forces near vital anatomy by 63% and peak forces by 50% in comparison to a CNC-type path, without adding time to the procedure.

In this work, we grant robot agents the capacity to sense and react to their human supervisor's changing trust state, as a means to maintain the efficiency of their collaboration. We propose the novel formulation of Trust-Aware Conservative Control (TACtiC), in which the agent alters its behaviors momentarily whenever the human loses trust. This trust-seeking robot framework builds upon an online trust inference engine and also incorporates an interactive behavior adaptation technique. We present end-to-end instantiations of trust-seeking robots for distinct task domains of aerial terrain coverage and interactive autonomous driving. Empirical assessments comprise a large-scale controlled interaction study and its extension into field evaluations with an autonomous car. These assessments substantiate the efficiency gains that trust-seeking agents bring to asymmetric human-robot teams.

Mobile microrobots are typically fabricated in a multi-step microfabrication process and then transported into an enclosed workspace for operation. This paper presents a new, 3D printing inspired method for in-situ fabrication of mobile magnetic microrobots with complex topology from a polymer filament on demand directly inside an enclosed operational environment. Through the use of a tip magnet on the filament, the target shape is formed by magnetic guidance from external electromagnetic coils which wirelessly project fields into the workspace as the filament is fed through a hot needle which is inserted into the workspace. A bending model and a shape planner are developed for predicting and controlling the fabrication process. Magnetically-active millimeter-scale robotic devices of different shapes and sizes are fabricated using polylactic acid (PLA) filament with diameter as small as 50 µm. As a demonstration of the in-situ formation of a functional microrobotic device, a force-sensing microrobot with integrated sensing spring is fabricated inside an enclosed space, and then is used to measure the manipulation force during a pushing experiment by optical deformation measurement. We thus show the utility of the fabrication method for creating complex microrobot shapes remotely in enclosed environments for advanced microrobotic applications, with the potential for scaled down applications in healthcare and microfluidics.

The paper deals with comparison study of serial and quasi-serial industrial robots used for machining operations. It proposes a new methodology for robot ranking, which is based on estimation of the end-effector resistance to cutting forces for several machining tasks, which are optimally located within the robot workspace. To cover wide range of applications, a set of isotropic, quasi-isotropic and extended benchmark tasks are considered. It is shown that regardless of the benchmark problem, seral manipulators are preferable for small and medium tasks while quasi-serial ones' better suit large-dimensional tasks. The proposed technique was applied for the comparison analysis of 10 industrial robots of both serial and quasi-serial architectures with similar working radius and payload of about 200 kg.

In this paper, we present MO-SLAM, a novel visual SLAM system that is capable of detecting duplicate objects in the scene during run-time without requiring an offline training stage to pre-populate a database of objects. Instead, we propose a novel method to detect landmarks that belong to duplicate objects. Further, we show how landmarks belonging to duplicate objects can be converted to first-order entities which generate additional constraints for optimizing the map. We evaluate the performance of MO-SLAM with extensive experiments on both synthetic and real data, where the experimental results verify the capabilities of MO-SLAM in detecting duplicate objects and using these constraints to improve the accuracy of the map.

In this paper, we present a novel global descriptor M2DP for 3D point clouds, and apply it to the problem of loop closure detection. In M2DP, we project a 3D point cloud to multiple 2D planes and generate a density signature for points for each of the planes. We then use the left and right singular vectors of these signatures as the descriptor of the 3D point cloud. Our experimental results show that the proposed algorithm outperforms state-of-the-art global 3D descriptors in both accuracy and efficiency.

Real-time obstacle detection is a key component of autonomous vehicles. In this context, low obstacles are particularly challenging, as they are often discarded by traditional algorithms. Curb detection methods that can potentially be suitable for the problem usually target roads with clearly defined curbs and sidewalks. We propose a real-time algorithm for the detection of low obstacles (including, but not restricted to curbs), merging 2-D and 3-D information from stereo imaging. A set of candidate object lines is extracted based on their combined 2-D and 3-D features, tracked over time and clustered according to a novel similarity metric. Finally, a 3rd order polynomial spline is fitted to each cluster to represent the obstacle. The proposed system can deal with noisy and incomplete point clouds and keeps the model assumptions to a minimum. To evaluate the algorithm, a new stereo dataset is provided and made available online. We present experiments in different scenarios and lighting conditions, illustrating the applicability of the method.

New vision sensors, such as the Dynamic and Active-pixel Vision sensor (DAVIS), incorporate a conventional camera and an event-based sensor in the same pixel array. These sensors have great potential for robotics because they allow us to combine the benefits of conventional cameras with those of event-based sensors: low latency, high temporal resolution, and high dynamic range. However, new algorithms are required to exploit the sensor characteristics and cope with its unconventional output, which consists of a stream of asynchronous brightness changes (called “events”) and synchronous grayscale frames. In this paper, we present a low-latency visual odometry algorithm for the DAVIS sensor using event-based feature tracks. Features are first detected in the grayscale frames and then tracked asynchronously using the stream of events. The features are then fed to an event-based visual odometry algorithm that tightly interleaves robust pose optimization and probabilistic mapping. We show that our method successfully tracks the 6-DOF motion of the sensor in natural scenes. This is the first work on event-based visual odometry with the DAVIS sensor using feature tracks.

Motor or visual impairments may prevent a user from steering a wheelchair effectively in indoor environments. In such cases, joystick jerks arising from uncontrolled motions may lead to collisions with obstacles. We here propose a perceptive shared control system that progressively corrects the trajectory as a user manually drives the wheelchair, by means of a sensor-based shared control law capable of smoothly avoiding obstacles. This control law is based on a low complex optimization framework validated through simulations and extensive clinical trials. The provided model uses distance information. Therefore, for low-cost considerations, we use ultrasonic sensors to measure the distances around the wheelchair. The solution therefore provides an efficient assistive tool that does not alter the quality of experience perceived by the user, while ensuring his security in hazardous situations.

Detecting small obstacles on the road ahead is a critical part of the driving task which has to be mastered by fully autonomous cars. In this paper, we present a method based on stereo vision to reliably detect such obstacles from a moving vehicle.

Proper place recognition on an environment that can change over time is fundamental for long-term SLAM. In such scenarios the observations obtained in the same region can drastically differ due to changes caused by semi-static objects, such as doors, furniture, etc. In this work, we extend a strategy that represents environment regions using words, based on spatial density information extracted from laser readings. This time, in order to deal with changes in the environment, our method not only builds words representing the real observations made by the robot, but also alternative multi-level words to account for possible changes in a place's observations generated by non-static objects. Place recognition is made by searching matches of sequences of N consecutive words (both real or alternatives). Experiments performed in real and simulated scenarios are shown, and demonstrate the advantages associated to the use of multi-level words.

Despite significant progress in GPS-denied autonomous flight, long-distance traversals (> 100 km) in the absence of GPS remain elusive. This paper demonstrates a method capable of accurately estimating the aircraft state over a 218 km flight with a final position error of 27 m, 0.012% of the distance traveled. Our technique efficiently captures the full state dynamics of the air vehicle with semi-intermittent global corrections using LIDAR measurements matched against an a priori Digital Elevation Model (DEM). Using an error-state Kalman filter with IMU bias estimation, we are able to maintain a high-certainty state estimate, reducing the computation time to search over a global elevation map. A sub region of the DEM is scanned with the latest LIDAR projection providing a correlation map of landscape symmetry. The optimal position is extracted from the correlation map to produce a position correction that is applied to the state estimate in the filter. This method provides a GPS-denied state estimate for long range drift-free navigation. We demonstrate this method on two flight data sets from a full-sized helicopter, showing significantly longer flight distances over the current state of the art.

We present a long-range visual signal detection system that is suitable for an unmanned aerial vehicle to find an optical signal released at a desired landing site for the purposes of cargo delivery or rescue situations where radio signals or other communication systems are not available or the wind conditions at the landing site need to be signaled. The challenge here is to have a signal and detection system that works from long range (> 1000m) amongst ground clutter during various seasonal conditions on passive imagery. We use a smoke-grenade as a ground signal, which has the advantageous properties of being easy to carry by ground crews because of its light weight and small size, but when released has a long visual signaling range. We employ a camera system on the UAV with a visual texture feature extraction approach in a machine learning framework to classify image patches as `signal' or `background'. We study conventional approaches and develop a visual feature descriptor that can better differentiate the appearance of the visual signal under varying conditions and, when used to train a random-forest classifier, outperforms commonly used feature descriptors. The system was rigorously and quantitatively evaluated on data collected from a camera mounted on a helicopter and flown towards a plume of signal smoke over a variety of seasons, ground conditions, weather conditions, and environments. Our system was capable of detecting the smoke cloud with both precision and recall rates greater than 0.95 from ranges between 1000m and 1500m. Further, we develop a method to estimate wind orientation and approximate wind strength by assessing the shape of the smoke signal. We present a preliminary evaluation of the wind estimation in conditions with different wind intensities and orientations relative to the approach direction.

We propose a log-space solution for robotic path planning with harmonic functions that solves the long-standing numerical precision problem. We prove that this algorithm: (1) performs the correct computations in log-space, (2) returns the true equivalent path using the log-space mapping, and (3) has a strong error bound given its convergence criterion. We evaluate the algorithm on 7 problem domains. A Graphics Processing Unit (GPU) implementation is also shown to greatly improve performance. We also provide an open source library entitled epic with extensive ROS support and demonstrate this method on a real humanoid robot: the uBot-6. Experiments demonstrate that the log-space solution rapidly produces smooth obstacle-avoiding trajectories, and supports planning in exponentially larger real-world robotic applications.

Animals provide inspiration for developing soft machines in bionics, robotics research as well as potential applications. This paper presents an integrated development of locomotive soft robot platforms: starfish-inspired robots with multi-limb bodies actuated by shape memory alloys (SMAs). The designs of these robots were based on the biological specifications of the locomotion and water-vascular systems of regenerating starfish. Multi-limb robot prototypes were fabricated with soft materials and 3D printing technology, satisfying modest movement requirements with SMA spring actuators. Experimental testing of multi-limb robots were conducted in accordance with biological locomotion principles. The robots vividly displayed typical starfish motion models. Animals provide inspiration for developing soft machines in bionics, robotics research as well as potential applications.

In order for robots to interact safely and intelligently with their environment they must be able to reliably estimate and localize external contacts. This paper introduces CPF, the Contact Particle Filter, which is a general algorithm for detecting and localizing external contacts on rigid body robots without the need for external sensing. CPF finds external contact points that best explain the observed external joint torque, and returns sensible estimates even when the external torque measurement is corrupted with noise. We demonstrate the capability of the CPF to track multiple external contacts on a simulated Atlas robot, and compare our work to existing approaches.