This paper proposes a computationally efficient approach to detecting objects natively in 3D point clouds using convolutional neural networks (CNNs). In particular, this is achieved by leveraging a feature-centric voting scheme to implement novel convolutional layers which explicitly exploit the sparsity encountered in the input. To this end, we examine the trade-off between accuracy and speed for different architectures and additionally propose to use an L1 penalty on the filter activations to further encourage sparsity in the intermediate representations. To the best of our knowledge, this is the first work to propose sparse convolutional layers and L1 regularisation for efficient large-scale processing of 3D data. We demonstrate the efficacy of our approach on the KITTI object detection benchmark and show that Vote3Deep models with as few as three layers outperform the previous state of the art in both laser and laser-vision based approaches by margins of up to 40% while remaining highly competitive in terms of processing time.

Abstract-In this paper, we revisit batch state estimation through the lens of Gaussian process (GP) regression. We consider continuous-discrete estimation problems wherein a trajectory is viewed as a one-dimensional GP, with time as the independent variable. Our continuous-time prior can be defined by any linear, time-varying stochastic differential equation driven by white noise; this allows the possibility of smoothing our trajectory estimates using a variety of vehicle dynamics models (e.g., 'constant-velocity'). We show that this class of prior results in an inverse kernel matrix (i.e., covariance matrix between all pairs of measurement times) that is exactly sparse (block-tridiagonal) and that this can be exploited to carry out GP regression (and interpolation) very efficiently. Though the prior is continuous, we consider measurements to occur at discrete times. When the measurement model is also linear, this GP approach is equivalent to classical, discrete-time smoothing (at the measurement times). When the measurement model is nonlinear, we iterate over the whole trajectory (as is common in vision and robotics) to maximize accuracy. We test the approach experimentally on a simultaneous trajectory estimation and mapping problem using a mobile robot dataset.

In this paper, we revisit batch state estimation through the lens of Gaussian process (GP) regression. We consider continuous-discrete estimation problems wherein a trajectory is viewed as a one-dimensional GP, with time as the independent variable. Our continuous-time prior can be defined by any nonlinear, time-varying stochastic differential equation driven by white noise; this allows the possibility of smoothing our trajectory estimates using a variety of vehicle dynamics models (e.g., 'constant-velocity'). We show that this class of prior results in an inverse kernel matrix (i.e., covariance matrix between all pairs of measurement times) that is exactly sparse (block-tridiagonal) and that this can be exploited to carry out GP regression (and interpolation) very efficiently. When the prior is based on a linear, time-varying stochastic differential equation and the measurement model is also linear, this GP approach is equivalent to classical, discrete-time smoothing (at the measurement times); when a nonlinearity is present, we iterate over the whole trajectory to maximize accuracy. We test the approach experimentally on a simultaneous trajectory estimation and mapping problem using a mobile robot dataset.

In this paper, we present Gaussian Process Gauss–Newton (GPGN), an algorithm for non-parametric, continuous-time, nonlinear, batch state estimation. This work adapts the methods of Gaussian process (GP) regression to address the problem of batch simultaneous localization and mapping (SLAM) by using the Gauss–Newton optimization method. In particular, we formulate the estimation problem with a continuous-time state model, along with the more conventional discrete-time measurements. Two derivations are presented in this paper, reflecting both the weight-space and function-space approaches from the GP regression literature. Validation is conducted through simulations and a hardware experiment, which utilizes the well-understood problem of two-dimensional SLAM as an illustrative example. The performance is compared with the traditional discrete-time batch Gauss–Newton approach, and we also show that GPGN can be employed to estimate motion with only range/bearing measurements of landmarks (i.e. no odometry), even when there are not enough measurements to constrain the pose at a given timestep.

Roboticists often formulate estimation problems in discrete time for the practical reason of keeping the state size tractable; however, the discrete-time approach does not scale well for use with high-rate sensors, such as inertial measurement units, rolling-shutter cameras, or sweeping laser imaging sensors. The difficulty lies in the fact that a pose variable is typically included for every time at which a measurement is acquired, rendering the dimension of the state impractically large for large numbers of measurements. This issue is exacerbated for the simultaneous localization and mapping problem, which further augments the state to include landmark variables. To address this tractability issue, we propose to move the full Maximum-a-Posteriori estimation problem into continuous time and use temporal basis functions to keep the state size manageable. We present a full probabilistic derivation of the continuous-time estimation problem, derive an estimator based on the assumption that the densities and processes involved are Gaussian and show how the coefficients of a relatively small number of basis functions can form the state to be estimated, making the solution efficient. Our derivation is presented in steps of increasingly specific assumptions, opening the door to the development of other novel continuoustime estimation algorithms through the application of different assumptions at any point. We use the simultaneous localization and mapping problem as our motivation throughout the paper, although the approach is not specific to this application. Results from two experiments are provided to validate the approach: (i) self-calibration involving a camera and a high-rate inertial measurement unit, and (ii) perspective localization with a rolling-shutter camera.

This paper explores the idea of reducing a robot's energy consumption while following a trajectory by turning off the main localisation subsystem and switching to a lowerpowered, less accurate odometry source at appropriate times. This applies to scenarios where the robot is permitted to deviate from the original trajectory, which allows for energy savings. Sensor scheduling is formulated as a probabilistic belief planning problem. Two algorithms are presented which generate feasible perception schedules: the first is based upon a simple heuristic; the second leverages dynamic programming to obtain optimal plans. Both simulations and real-world experiments on a planetary rover prototype demonstrate over 50% savings in perception-related energy, which translates into a 12% reduction in total energy consumption. I. INTRODUCTIONRobots require energy to operate. Yet they only have access to limited energy storage during missions. As we extend the reach of autonomous systems to operate in remote locations, over long distances and for long periods of time, energy considerations are becoming increasingly important. To date, these considerations are often brought to bear in schemes where trajectories or speed profiles are optimised to minimise the energy required for actuation (see, for example, [1], [2], [3]). Here we take a different, yet complementary, approach in considering the energy expenditure for sensing (and, implicitly, computation) associated with navigation. In particular, our goal is to activate the perception system only as required to maintain the vehicle within a given margin around a predetermined path. As the main navigation sensors are switched off and the robot reverts to a lower-powered, less accurate odometry source for parts of the trajectory, any associated computation will also be reduced, leading to further savings in energy.Naively, such perception schedules could be constructed by switching sensors on and off randomly or according to, for example, a fixed frequency. This does, however, suffer the drawback that no heed is paid to drift in the robot's position with respect to the original trajectory: it may not be desirable to deviate by more than an allowed margin from the predetermined route. This arises, for example, in a planetary exploration scenario when conducting long traverses over featureless terrain. Other possible considerations include traversability, obstacles, and the robustness of the localisation system to deviations from the original path. Such naive approaches would also need to be tuned to individual trajectories as savings would depend significantly on trajectory shape. In this work we present two approaches which explicitly account for drift and trajectory shape (though the

This paper describes a collection of 272 three-dimensional laser scans gathered at two unique planetary analogue rover test facilities in Canada, which offer emulated planetary terrain at manageable scales for algorithmic development. This dataset is subdivided into four individual subsets, each gathered using panning laser rangefinders on different mobile rover platforms. This data should be of interest to field robotics researchers developing rover navigation algorithms suitable for use in three-dimensional, unstructured, natural terrain. All of the data are presented in human-readable text files, and are accompanied by Matlab parsing scripts to facilitate use thereof. This paper provides an overview of the available data.

In this paper, we present a robust framework suitable for conducting three‐dimensional simultaneous localization and mapping (3D SLAM) in a planetary work site environment. Operation in a planetary environment imposes sensing restrictions, as well as challenges due to the rugged terrain. Utilizing a laser rangefinder mounted on a rover platform, we have demonstrated an approach that is able to create globally consistent maps of natural, unstructured 3D terrain. The framework presented in this paper utilizes a sparse‐feature‐based approach and conducts data association using a combination of feature constellations and dense data. Because of feature scarcity, odometry measurements are also incorporated to provide additional information in feature‐poor regions. To maintain global consistency, these measurements are resolved using a batch alignment algorithm, which is reinforced with heterogeneous outlier rejection to improve its robustness to outliers in either measurement type (i.e., laser or odometry). Finally, a map is created from the alignment estimates and the dense data. Extensive validation of the framework is provided using data gathered at two different planetary analogue facilities, which consist of 50 and 102 3D scans, respectively. At these sites, root‐mean‐squared mapping errors of 4.3 and 8.9 cm were achieved. Relative metrics are utilized for localization accuracy and map quality, which facilitate detailed analysis of the performance, including failure modes and possible future improvements. © 2012 Wiley Periodicals, Inc.