A robust downward-looking camera based velocity estimation with height compensation for mobile robots

This paper proposed a new algorithm for optical flow-based monocular vision (MV)/ inertial navigation system (INS) integrated navigation. In this mode, a downward-looking camera is used to get the image sequences, which is used to estimate the velocity of the mobile robot by using optical flow algorithm. INS is employed for the yaw variation. In order to evaluate the performance of the proposed method, a real indoor test has done. The result shows that the proposed method has good performance for velocity estimation. It can be applied to the autonomous navigation of mobile robots when the Global Positioning System (GPS) and code wheel is unavailable.

Recently, compressive sensing theory has opened up a new path for the development of signal processing. According to this theory, a novel single pixel camera system has been introduced to overcome the current limitation and challenges of manufacturing large scale photo sensor arrays. In the system, image can be recovered from original image using random measurements by means of compressive sensing techniques. In the image recovering process, some default parameters are used. It is important to find optimizing to enhance image accuracy and recovering speed. Some images recovering algorithms were attempted to recover images and some parameters of the algorithms need be set appropriately during the recovering process. In order to find the better values of the parameters, in this paper, a multi-objective optimizing method is proposed. Accuracy and rapidity are selected for optimization goal, and a multi-objective fitness function is built. Then some optimization parameters are selected and the ranges of the parameters are decided. And genetic algorithm is used in the optimization process. Finally, the result of optimization is used in a single pixel camera system. And the results of recovering images are better than the default parameter's for a single pixel camera experiment.

Micro Aerial Vehicles (MAVs) have become increasingly popular due to their wide applications in many areas. For autonomous navigation and safe control of MAVs, it is essential to have accurate and reliable velocity and position estimation. However, due to limited computational power and payload, it is still challenging for autonomous operation of MAVs in complex environments. In this paper, we propose a robust and efficient velocity estimation framework for MAVs with a single downward-facing RGB-D camera, which is able to provide metric velocity estimation in three dimensions as well as yaw rate in real-time without the fusion of additional sensors. Unlike traditional algorithms, our method uses kernel cross-correlators (KCCs) to efficiently determine optical flow for motion estimation, which does not rely on costly feature extraction and matching process. Moreover, we utilize depth images to estimate the vertical velocity of MAVs without the assumption of a flat ground. Autonomous flight tests on a quadrotor in complex environments demonstrate the robustness and efficiency of our method.

Transcranial magnetic resonance-guided focused ultrasound (tcMRgFUS) enables the noninvasive treatment of the deep brain. This capacity relies on the ability to focus acoustic energy through the in-tact skull, a feat that requires accurate estimates of the acoustic velocity in individual patient skulls. In current practice, these estimates are generated using a pretreatment computed tomography (CT) scan and then registered to a magnetic resonance (MR) dataset on the day of the treatment. Treatment safety and efficacy can be improved by eliminating the need to register the CT data to the MR images and by improving the accuracy of acoustic velocity measurements. In this study, we examine the capacity of MR to supplement or replace CT as a means of estimating velocity in the skull. We find that MR can predict velocity with less but comparable accuracy to CT. We then use micro-CT imaging to better understand the limitations of Hounsfield unit (HU)-based estimates of velocity, demonstrating that the macrostructure of pores in the skull contributes to the acoustic velocity of the bone. We find evidence that detailed T2 measurements provide information about pore macrostructure similar to the information obtained with micro-CT, offering a potential clinical mechanism for improving patient-specific estimates of acoustic velocity in the human skull.

Kinematic information such as position, velocity, and acceleration is critical to determine the three-dimensional state of the robot in space. In this study, it is aimed to estimate as visual the linear and angular velocity of a mobile robot. Additionally, another aim of this study is to determine the suitability of the concatenation or subtraction layer in the Convolutional Neural Network (CNN) that will make this estimate. For these purposes, first, a simulation environment was created. 9000 pairs of images and necessary velocity information were collected from this simulation environment for training. Similarly, 1000 pairs of images and velocity information were gathered for validation. Four different CNN models were designed and these models were trained and tested using these datasets. As a result of the test, the lowest average error for linear velocity estimation was calculated as 0.93e-3m/s and angular velocity estimation was measured as 4.37e-3rad/s. It was observed that the results were sufficient for linear and angular velocity prediction according to statistical analysis of errors. In addition, it was observed that the subtraction layer can be used instead of the concatenation layer in the CNN architectures for hardware-limited systems. As a result, visual velocity estimation of mobile robots has been achieved with this study and the framework of CNN models has been drawn for this problem.

This paper presents the position calibration of a polygonal array of optical mice for improved mobile robot velocity estimation using optical mice. First, the least squares velocity estimate of an omnidirectional mobile robot is obtained as the simple average of the velocity measurements from optical mice. Second, the sensitivity of the least squares velocity estimation to an imprecisely installed optical mouse array is analyzed. Third, with the aid of other accurate sensors, for example, wheel encoders, a simple but effective calibration for imperfect optical mouse array position is proposed. Finally, the position calibration of a regular triangular optical mouse array is given to demonstrate the effectiveness of the proposed method.

Real-time vision-based navigation is a difficult task largely due to the limited optical properties of single cameras that are usually mounted on robots. Multiple camera systems such as polydioptric sensors provide more efficient and precise solutions for autonomous navigation. They are particularly suitable for motion estimation because they allow one to formulate a linear optimization. These sensors capture the visual information in a more complete form called the plenoptic function that encodes the spatial and temporal light radiance of the scene. The polydioptric sensors are rarely used in robotics because they are usually thought to increase the amount of data produced and require more computational power. This paper shows that these cameras provide more accurate estimation results in mobile robotics navigation if designed properly. It also shows that a plenoptic vision sensor with a resolution ranging from 3 × 3 to 40 × 30 pixels camera, provides higher accuracy than a mono-SLAM running on a 320 × 240 pixels camera. The paper also gives a complete scheme to design usable real-time plenoptic cameras for mobile robotics applications by establishing the link between velocity, resolution and motion estimation accuracy. Finally, experiments on a mobile robot are shown allowing for a comparison between optimal plenoptic visual sensors and single high-resolution cameras. The estimation with the plenoptic sensor is more accurate than a monocular high-definition camera with a processing time 100 times lower.

This paper presents the velocity estimation of a mobile robot using a regular polygonal array of optical flow sensors that are installed at the bottom of a mobile robot. First, the basic principle of the proposed localization scheme is described, which is motivated from the changing optical flow sensor readings according to the mobile robot traveling pattern. Second, the velocity kinematic mapping from a mobile robot to an array of optical flow sensors is derived as an overdetermined linear system. Third, for a given set of optical flow sensor readings, the mobile robot velocity is estimated based on the least squares solution to the derived system. Finally, simulation results are given to demonstrate the validity of the proposed velocity estimation.

The precise localisation of rail vehicles is fundamental to the development and employment of more efficient train control systems in security, logistics and disposition applications. A key element of this is the exact and reliable velocity and distance estimation. The popular satellite navigation (global navigation satellite systems) and visual odometry, based on optical systems, tend to fail in the rough environment of rail application scenarios. This study describes an approach to making a precise distance estimate using an innovative eddy current sensor system. This is achieved through the combination of an augmented cross‐correlation approach to highly accurate velocity estimation. That is subsequently employed to determine the travelled distance based on counting rail clamps in a spatial transformed space. This approach relies on recursive state estimation of the dynamic states and a statistical assessment of the counted rail clamps for the purpose of outlier rejection. The proposed methods are compared and evaluated in a quantitative way on the basis of experimental data. The results prove that the proposed approach can improve the accuracy and reliability of velocity measurement and relative position detection.

Velocity measurement by an incremental encoder is an important issue for advanced motion control applications such as robotics. In this paper, we deal with a kind of MT-type velocity estimation method. Though the conventional MT method is well known and has been well proven in practice, it requires execution of an arithmetic division operation that prevents an efficient implementation on low-cost FPGA-based control platforms. Thus, we propose a divisionless MT-type algorithm, which can provide a similar performance in velocity estimation accuracy as the conventional method, but requiring significantly less FPGA resources, since it implements only simple arithmetic operations such as addition, subtraction, and multiplication, that can be implemented more easily on the processing hardware. Furthermore, the algorithm is fast in execution, thus, it provides the output in only a few clock cycles. Though the proposed algorithm can be described in a recursive form, the stability of the estimation process is not jeopardized, although it is an important issue in this case. Hence, the algorithm is introduced in a form which assures stability in a wide speed range. We show the implementation of the algorithm on the experimental FPGA platform. The experimental results validated the proposed divisionless MT-type algorithm fully for accurate velocity estimation.

Ultrathin metasurfaces have shown the capability to influence all aspects of light propagation. This has made them promising options for replacing conventional bulky imaging optics while adding advantageous optical properties or functionalities. We demonstrate that such metasurfaces can also be applied for single-lens three-dimensional (3-D) imaging based on a specifically engineered point-spread function (PSF). Using Huygens’ metasurfaces with high transmission, we design and realize a phase mask that implements a rotating PSF for 3-D imaging. We experimentally characterize the properties of the realized double-helix PSF, finding that it can uniquely encode object distances within a wide range. Furthermore, we experimentally demonstrate wide-field depth retrieval within a 3-D scene, showing the suitability of metasurfaces to realize optics for 3-D imaging, using just a single camera and lens system.

In a single camera single source gaze tracking system, if we can accurately estimate the 3-D coordinates of the iris center and the center of corneal surface curvature, we can reconstruct 3-D gaze direction. A novel method is proposed in this paper, different from other gaze tracking systems, the user calibration in this paper is the calibration of the iris radius, corneal curvature radius and kappa angle. Through the simulation experiments, the feasibility of the algorithm proposed in this paper is analyzed, and the error of the algorithm is relatively small.

This paper presents the optimal array of optical mice for the accurate velocity estimation of a mobile robot. It is assumed that there can be some restriction on the installation of two or more optical mice at the bottom of a mobile robot. First, the velocity kinematics of a mobile robot with an array of optical mice is derived, which maps the velocity of a mobile robot to the velocities of optical mice. Second, taking into account the consistency in physical units, the uncertainty ellipsoid is obtained to represent the error characteristics of the mobile robot velocity estimation owing to noisy optical mouse measurements. Third, a simple but effective performance index is defined as the inverse of the volume of the uncertainty ellipsoid, which can be used for the optimization of the optimal optical mouse placement. Fourth, simulation results for the optimal placement of three optical mice within a given elliptical region are given.

We previously presented a technique calibrating a Multi-Aperture Array Projector (MAAP) for performing high-speed three-dimensional (3D) surface topology measurements with structured illumination. The MAAP achieves projection rates up to 3 kHz by projecting a sequence of aperiodic sinusoidal fringes through several distinct projecting aperture channels such that up to 330 3D measurements per second could be achieved. This rapid switching between several projecting channels overcomes the projection limit of conventional off-the-shelf single aperture Digital Light Processing (DLP) or Liquid Crystal on Silicon (LCoS) technologies that typically have projection rates on the order of 100 Hz. The proposed MAAP calibration technique re-enabled photogrammetric 3D measurements to be made using a single CCD or CMOS camera where a stereo-camera setup used to be required for triangulation. The focus to reduce a stereo-configuration to a single camera system opens possibilities to decrease cost as well as form factor of such a high-speed metrology system utilizing an MAAP. For such a photogrammetric optical metrology system, intrinsic camera calibration and extrinsic setup calibration are vital to obtain high measurement performance. With the pinhole camera model assumption, intrinsic calibration is typically carried out using the standard Zhang method. However, there is currently no such standard technique that can comprehensively perform the required extrinsic calibration. In this study, the main sources of measurement error arising from extrinsic calibration of the proposed technique are discussed and investigated. Several different extrinsic calibration techniques are proposed and the resulting measurement performance evaluated.

To train a mobile robot to navigate using end-to-end approach which maps sensors data into actions, we can use deep reinforcement learning (DRL) method by providing training environments with proper reward functions. Although some studies have shown the success of DRL in navigation task for mobile robots, the method needs appropriate hyperparameter settings such as the environment's timestep size and the robot's velocity range to produce a good navigation policy. The previous existing DRL framework has proposed the use of odometry sensor to generate dynamic timestep size in the environment to solve the mismatch problem between the timestep size and the robot's velocity. However, the framework lacks a procedure for checking terminal conditions which may occur during action executions resulting inconsistency in the environment and poor navigation policies. In the case of navigation task, terminal conditions may happen when the robot achieves the navigation goal position or collides with obstacles while performing an action in one timestep. To cope with this problem, we propose a state transition checking method in the DRL environment which is specific for navigation task that leverages odometry and laser sensor to ensure that the environment follows Markov Decision Process with dynamic timestep size. We also introduce a velocity increment scheduling to stabilize the mobile robot during training. Our experiment results show that state transition checking along with the velocity increment scheduling are able to make the robot navigate faster with higher success rate compared to other existing DRL frameworks.