ICRA 2020 论文列表

Mapping and localization is a critical module of autonomous driving, and significant achievements have been reached in this field. Beyond Global Navigation Satellite System (GNSS), research in point cloud registration, visual feature matching, and inertia navigation has greatly enhanced the accuracy and robustness of mapping and localization in different scenarios. However, highly urbanized scenes are still challenging: LIDAR- and camera-based methods perform poorly with numerous dynamic objects; the GNSS-based solutions experience signal loss and multi-path problems; the inertia measurement units (IMU) suffer from drifting. Unfortunately, current public datasets either do not adequately address this urban challenge or do not provide enough sensor information related to map-ping and localization. Here we present UrbanLoco: a mapping/localization dataset collected in highly-urbanized environments with a full sensor-suite. The dataset includes 13 trajectories collected in San Francisco and Hong Kong, covering a total length of over 40 kilometers. Our dataset includes a wide variety of urban terrains: urban canyons, bridges, tunnels, sharp turns, etc. More importantly, our dataset includes information from LIDAR, cameras, IMU, and GNSS receivers. Now the dataset is publicly available through the link in the footnote 1.

Hand-crafting generalised decision-making rules for real-world urban autonomous driving is hard. Alternatively, learning behaviour from easy-to-collect human driving demonstrations is appealing. Prior work has studied imitation learning (IL) for autonomous driving with a number of limitations. Examples include only performing lane-following rather than following a user-defined route, only using a single camera view or heavily cropped frames lacking state observability, only lateral (steering) control, but not longitudinal (speed) control and a lack of interaction with traffic. Importantly, the majority of such systems have been primarily evaluated in simulation - a simple domain, which lacks real-world complexities. Motivated by these challenges, we focus on learning representations of semantics, geometry and motion with computer vision for IL from human driving demonstrations. As our main contribution, we present an end-to-end conditional imitation learning approach, combining both lateral and longitudinal control on a real vehicle for following urban routes with simple traffic. We address inherent dataset bias by data balancing, training our final policy on approximately 30 hours of demonstrations gathered over six months. We evaluate our method on an autonomous vehicle by driving 35km of novel routes in European urban streets.

This study has developed a fault-tolerant controller that is able to recover a quadrotor from arbitrary initial orientations and angular velocities, despite the complete failure of a rotor. This cascaded control method includes a position/altitude controller, an almost-global convergence attitude controller, and a control allocation method based on quadratic programming. As a major novelty, a constraint of undesirable angular velocity is derived and fused into the control allocator, which significantly improves the recovery performance. For validation, we have conducted a set of Monte-Carlo simulation to test the reliability of the proposed method of recovering the quadrotor from arbitrary initial attitude/rate conditions. In addition, real-life flight tests have been performed. The results demonstrate that the post-failure quadrotor can recover after being casually tossed into the air.

This paper presents scalable designs and fabrication, actuation, and manipulation techniques for soft millirobots under uniform magnetic field control. The millirobots were fabricated through an economic and robust moulding technique using polydimethylsiloxane (PDMS), acrylonitrile butadiene styrene (ABS) filaments, and 3D printed polylactic acid (PLA) rings. The soft millirobots were simple hollow rod-like structures with different configurations of embedded permanent magnets inside of their soft-body or at their ends. The soft-robots were actuated using six different motion modes including: pivot walking, rolling, tumbling, side-tapping, wiggling, and wavy-motion under an external uniform magnetic field control system. The velocities of the millirobots under different motion modes were analyzed under varying magnetic flux densities (B). Moreover, deformation of the soft-robotic body in response to the magnetic field strength was measured and a deflection curve showing bending angle (φ) was produced. Soft millirobots were navigated through a maze using a combination of the available motion modes. Different arrangements of the embedded permanent magnets enabled individual soft millirobots to respond heterogeneously under the same magnetic field inputs towards performing assembly and disassembly operation as modular subunits. Overall, this soft millirobot platform shows enormous potential for minimally invasive in vivo applications.

Performing Reinforcement Learning in sparse rewards settings, with very little prior knowledge, is a challenging problem since there is no signal to properly guide the learning process. In such situations, a good search strategy is fundamental. At the same time, not having to adapt the algorithm to every single problem is very desirable. Here we introduce TAXONS, a Task Agnostic eXploration of Outcome spaces through Novelty and Surprise algorithm. Based on a population-based divergent-search approach, it learns a set of diverse policies directly from high-dimensional observations, without any task-specific information. TAXONS builds a repertoire of policies while training an autoencoder on the high-dimensional observation of the final state of the system to build a low-dimensional outcome space. The learned outcome space, combined with the reconstruction error, is used to drive the search for new policies. Results show that TAXONS can find a diverse set of controllers, covering a good part of the ground-truth outcome space, while having no information about such space.

Visual place recognition in changing environments is the problem of finding matchings between two sets of observations, a query set and a reference set, despite severe appearance changes. Recently, image comparison using CNNbased descriptors showed very promising results. However, the experiments in the literature typically assume a single distinctive condition within each set (e.g., reference images are captured at daytime and the query sequence is at night). In this paper, we will demonstrate that as soon as the conditions change within one set (e.g., reference is daytime and now the query is a traversal daytime-dusk-night-dawn), different places under the same condition can suddenly look more similar than same places under different conditions. As a consequence, state-of-the-art approaches like CNN-based descriptors fail. This paper discusses this practically very important problem of in-sequence condition changes and defines a hierarchy of problem setups from (1) no in-sequence changes, (2) discrete in-sequence changes, to (3) continuous in-sequence changes. We will experimentally evaluate the effect of in-sequence condition changes on two state-of-the-art CNN-descriptors and investigate unsupervised methods to improve their performance. This includes an evaluation of the importance of statistical normalization (standardization) of descriptors, which is often omitted in existing approaches but can considerably improve results for problems up to discrete in-sequence changes. To address the practical most relevant setup of continuous changes, we investigate the application of unsupervised learning methods using two PCA-based approaches from the literature and propose a novel clustering-based extension of the statistical normalization. We experimentally demonstrate that these approaches can significantly improve place recognition performance in case of continuous in-sequence condition changes. Matlab implementations of the presented approaches are available online:... Show More

Learning-based ego-motion estimation approaches have recently drawn strong interest from researchers, mostly focusing on visual perception. A few learning-based approaches using Light Detection and Ranging (LiDAR) have been re-ported; however, they heavily rely on a supervised learning manner. Despite the meaningful performance of these approaches, supervised training requires ground-truth pose labels, which is the bottleneck for real-world applications. Differing from these approaches, we focus on unsupervised learning for LiDAR odometry (LO) without trainable labels. Achieving trainable LO in an unsupervised manner, we introduce the uncertainty-aware loss with geometric confidence, thereby al-lowing the reliability of the proposed pipeline. Evaluation on the KITTI, Complex Urban, and Oxford RobotCar datasets demonstrate the prominent performance of the proposed method compared to conventional model-based methods. The proposed method shows a comparable result against SuMa (in KITTI), LeGO-LOAM (in Complex Urban), and Stereo-VO (in Oxford RobotCar). The video and extra-information of the paper are described in https://sites.google.com/view/deeplo.

We propose a novel anomaly detection framework for a fleet of hybrid aerial vehicles executing high-speed package pickup and delivery missions. The detection is based on machine learning models of normal flight profiles, trained on millions of flight log measurements of control inputs and sensor readings. We develop a new scalable algorithm for robust regression which can simultaneously fit predictive flight dynamics models while identifying and discarding abnormal flight missions from the training set. The resulting unsupervised estimator has a very high breakdown point and can withstand massive contamination of training data to uncover what normal flight patterns look like, without requiring any form of prior knowledge of aircraft aerodynamics or manual labeling of anomalies upfront. Across many different anomaly types, spanning simple 3sigma statistical thresholds to turbulence and other equipment anomalies, our models achieve high detection rates across the board. Our method consistently outperforms alternative robust detection methods on synthetic benchmark problems. To the best of our knowledge, dynamics modeling of hybrid delivery drones for anomaly detection at the scale of 100 million measurements from 5000 real flight missions in variable flight conditions is unprecedented.

Currently for balance recovery, humans outperform humanoid robots which use hand-designed controllers in terms of the diverse actions. This study aims to close this gap by finding core control principles that are shared across ankle, hip, toe and stepping strategies by formulating experiments to test human balance recoveries and define criteria to quantify the strategy in use. To reveal fundamental principles of balance strategies, our study shows that a minimum jerk controller can accurately replicate comparable human behaviour at the Centre of Mass level. Therefore, we formulate a general Model-Predictive Control (MPC) framework to produce recovery motions in any system, including legged machines, where the framework parameters are tuned for time-optimal performance in robotic systems.

Many approaches for camera and LiDAR calibration are presented in literature but none of them estimates all intrinsic and extrinsic parameters simultaneously and therefore optimally in a probabilistic sense.In this work, we present a method to simultaneously estimate intrinsic and extrinsic parameters of cameras and LiDARs in a unified problem. We derive a probabilistic formulation that enables flawless integration of different measurement types without hand-tuned weights. An arbitrary number of cameras and LiDARs can be calibrated simultaneously. Measurements are not required to be time-synchronized. The method is designed to work with any camera model.In evaluation, we show that additional LiDAR measurements significantly improve intrinsic camera calibration. Further, we show on real data that our method achieves state-of-the-art calibration precision with high reliability.

We present a deep residual network-based generative model for single image super-resolution (SISR) of underwater imagery for use by autonomous underwater robots. We also provide an adversarial training pipeline for learning SISR from paired data. In order to supervise the training, we formulate an objective function that evaluates the perceptual quality of an image based on its global content, color, and local style information. Additionally, we present USR-248, a large-scale dataset of three sets of underwater images of `high' (640×480) and `low' (80 × 60, 160 × 120, and 320×240) resolution. USR-248 contains paired instances for supervised training of 2×, 4×, or 8× SISR models. Furthermore, we validate the effectiveness of our proposed model through qualitative and quantitative experiments and compare the results with several state-of-the-art models' performances. We also analyze its practical feasibility for applications such as scene understanding and attention modeling in noisy visual conditions.

Despite their abundance in robotics and nature, underactuated systems remain a challenge for control engineering. Trajectory optimization provides a generally applicable solution, however its efficiency strongly depends on the skill of the engineer to frame the problem in an optimizer-friendly way. This paper proposes a procedure that automates such problem reformulation for a class of tasks in which the desired trajectory is specified by a sequence of waypoints. The approach is based on introducing auxiliary optimization variables that represent waypoint activations. To validate the proposed method, a letter drawing task is set up where shapes traced by the tip of a rotary inverted pendulum are visualized using long exposure photography.

Gecko-inspired adhesives have several desirable characteristics in robotic grasping: controllable activation and deactivation of adhesion, ability to grasp and release with minimal disturbance, and grasping without the need of form closure. Previously proposed grippers with this technology either require a complex activation mechanism or multiple activation steps. In this paper, we present an underactuated gecko-inspired adhesive gripper that can grasp a wide range of curved surfaces using a single actuator through a simple tendon-driven mechanism that attaches and adheres in one step. We derive a theoretical model of the adhesive contact area and resulting gripper grasp force, which is verified experimentally. The actual performance of the proposed mechanism is demonstrated by successfully grasping several surfaces with different curvature diameters.

This paper presents a self-supervised framework for learning to detect robust keypoints for odometry estimation and metric localisation in radar. By embedding a differentiable point-based motion estimator inside our architecture, we learn keypoint locations, scores and descriptors from localisation error alone. This approach avoids imposing any assumption on what makes a robust keypoint and crucially allows them to be optimised for our application. Furthermore the architecture is sensor agnostic and can be applied to most modalities. We run experiments on 280km of real world driving from the Oxford Radar RobotCar Dataset and improve on the state-of-the-art in point-based radar odometry, reducing errors by up to 45% whilst running an order of magnitude faster, simultaneously solving metric loop closures. Combining these outputs, we provide a framework capable of full mapping and localisation with radar in urban environments.

We propose an uncertainty-based sensor fusion framework for visual-inertial odometry, which is the task of estimating relative motion using images and measurements from inertial measurement units. Visual-inertial odometry enables robust and scale-aware estimation of motion by incorporating sensor states, such as metric scale, velocity, and the direction of gravity, into the estimation. However, the observability of the states depends on sensor motion. For example, if the sensor moves in a constant velocity, scale and velocity cannot be observed from inertial measurements. Under these degenerate motions, existing methods may produce inaccurate results because they incorporate erroneous states estimated from non-informative inertial measurements. Our proposed framework is able to avoid this situation by adaptively switching estimation modes, which represents the states that should be incorporated, based on their uncertainties. These uncertainties can be obtained at a small computational cost by reusing the Jacobian matrices computed in bundle adjustment. Our approach consistently outperformed conventional sensor fusion in datasets with different motion characteristics, namely, the KITTI odometry dataset recorded by a ground vehicle and the EuRoC MAV dataset captured from a micro aerial vehicle.

Deep neural network controllers for autonomous driving have recently benefited from significant performance improvements, and have begun deployment in the real world. Prior to their widespread adoption, safety guarantees are needed on the controller behaviour that properly take account of the uncertainty within the model as well as sensor noise. Bayesian neural networks, which assume a prior over the weights, have been shown capable of producing such uncertainty measures, but properties surrounding their safety have not yet been quantified for use in autonomous driving scenarios. In this paper, we develop a framework based on a state-of-the-art simulator for evaluating end-to-end Bayesian controllers. In addition to computing pointwise uncertainty measures that can be computed in real time and with statistical guarantees, we also provide a method for estimating the probability that, given a scenario, the controller keeps the car safe within a finite horizon. We experimentally evaluate the quality of uncertainty computation by three Bayesian inference methods in different scenarios and show how the uncertainty measures can be combined and calibrated for use in collision avoidance. Our results suggest that uncertainty estimates can greatly aid decision making in autonomous driving.

Successful robot path planning is challenging in the presence of visual occlusions and moving targets. Classical methods to solve this problem have used visioning and perception algorithms in addition to partially observable markov decision processes to aid in path planning for pursuit-evasion and robot tracking. We present a predictive path planning process that measures and utilizes the uncertainty present during robot motion planning. We develop a variant of subjective logic in combination with the Markov decision process (MDP) and provide a measure for belief, disbelief, and uncertainty in relation to feasible trajectories being generated. We then model the MDP to identify the best path planning method from a list of possible choices. Our results show a high percentage accuracy based on the closest acquired proximity between a target and a tracking robot and a simplified pursuer trajectory in comparison with related work.

Indoor positioning technology is essential for indoor mobile robots and drones. However, there has never been a general-purpose technology or infrastructure that enables indoor positioning with an accuracy of less than 10 cm. We have developed an attitude measurement method using multiple dynamic moires with a lenticular lens and developed an ultra- high-accuracy visual marker with an attitude estimation error of less than 0.1°. We also developed a calculation method that minimizes the marker position error by reminimizing reprojection error using its good attitude accuracy. We proved that accurate local positioning with a position error of about 1 cm in a marker coordinate system is possible even when a marker is shot from a distance of 10 m. In addition, a demonstration test was performed in a public space, and it was shown that high-accuracy global positioning with a position error of about 10 cm is possible.

The fusion of multiple sensor modalities, especially through deep learning architectures, has been an active area of study. However, an under-explored aspect of such work is whether the methods can be robust to degradation across their input modalities, especially when they must generalize to degradation not seen during training. In this work, we propose an uncertainty-aware fusion scheme to effectively fuse inputs that might suffer from a range of known and unknown degradation. Specifically, we analyze a number of uncertainty measures, each of which captures a different aspect of uncertainty, and we propose a novel way to fuse degraded inputs by scaling modality-specific output softmax probabilities. We additionally propose a novel data-dependent spatial temperature scaling method to complement these existing uncertainty measures. Finally, we integrate the uncertainty-scaled output from each modality using a probabilistic noisy-or fusion method. In a photo-realistic simulation environment (AirSim), we show that our method achieves significantly better results on a semantic segmentation task, as compared to state-of-art fusion architectures, on a range of degradation (e.g. fog, snow, frost, and various other types of noise), some of which are unknown during training.