Mean Position Error Research Articles

Impact of PBL and convection parameterization schemes for prediction of severe land-falling Bay of Bengal cyclones using WRF-ARW model

This study evaluates the performance of the Advanced Research Weather Research and Forecasting (WRF-ARW) model for prediction of land-falling Bay of Bengal (BoB) tropical cyclones (TCs). Model integration was performed using two-way interactive double nested domains at 27 and 9 km resolutions. The present study comprises two major components. Firstly, the study explores the impact of five different planetary boundary layer (PBL) and six cumulus convection (CC) schemes on seven land-falling BoB TCs. A total of 85 numerical simulations were studied in detail, and the results signify that the model simulated better both the track and intensity by using a combination of Yonsei University (YSU) PBL and the old simplified Arakawa-Schubert CC scheme. Secondly, the study also investigated the model performance based on the best possible combinations of model physics on the real-time forecasts of four BoB cyclones (Phailin, Helen, Lehar, and Madi) that made landfall during 2013 based on another 15 numerical simulations. The predicted mean track error during 2013 was about 71 km, 114 km, 133 km, 148 km, and 130 km respectively from day-1 to day-5. The Root Mean Square Error (RMSE) for Minimum Central Pressure (MCP) was about 6 hPa and the same noticed for Maximum Surface Wind (MSW) was about 4.5 m s−1 noticed during the entire simulation period. In addition the study also reveals that the predicted track errors during 2013 cyclones improved respectively by 43%, 44%, and 52% from day-1 to day-3 as compared to cyclones simulated during the period 2006–2011. The improvements noticed can be attributed due to relatively better quality data that was specified for the initial mean position error (about 48 km) during 2013. Overall the study signifies that the track and intensity forecast for 2013 cyclones using the specified combinations listed in the first part of this study performed relatively better than the other NWP (Numerical Weather Prediction) models, and thereby finds application in real-time forecast.

Development of a prototype robot and fast path-planning algorithm for static laser weeding

To demonstrate the feasibility and improve the implementation of laser weeding, a prototype robot was built and equipped with machine vision and gimbal mounted laser pointers. The robot consisted of a mobile platform modified from a small commercial quad bike, a camera to detect the crop and weeds and two steerable gimbals controlling the laser pointers. Visible-one laser pointers were used to simulate the powerful laser trajectories. A colour segmentation algorithm was utilised to extract plants from the soil background; size estimation was used to differentiate crop from weeds; and an erosion and dilation algorithm was developed to separate objects that were touching. Conversely, another algorithm, which utilised shape descriptors, was able to distinguish plant species in non-touching status regardless of area difference. Next, in order to reduce route length and run time, a new path-planning algorithm for static weeding was proposed and tested. It was demonstrated to be more efficient especially when addressing a higher density of weeds. A model was then established to determine the optimal segmentation size, based on the route length for treatment. It was found that the segmentation algorithm has the potential to be widely used in fast path-planning for the travelling-salesman problem. Finally, performance tests in the indoor environments showed that the weeding mean positional error was 1.97 mm, with a 0.88 mm standard deviation. Another test indicated that with a laser traversal speed of 30 mm/s and a dwell time of 0.64 s per weed, it had a hit rate of 97%.

Fiducial Markers are Necessary for Accurate Delivery of Liver SBRT

We standardly use fiducial markers for target localization when administering liver stereotactic body radiation therapy (SBRT). However, there are few data demonstrating the impact of matching to fiducials on accuracy of liver SBRT delivery. Given the risks of percutaneous placement of fiducials, the aim of our study was to evaluate the benefit of fiducial-based SBRT and to potentially define a subset of patients that can be treated without fiducial markers. Nineteen metastatic (n = 15) and primary (n = 4) liver tumors were treated with SBRT. Doses ranged from 30-54 Gy in 3-5 fractions, and 75 individual fractions were delivered in total. Target localization at the time of treatment was performed using fiducial markers on cone beam computed tomography (CBCT). To assess the difference between fiducial-based and non-fiducial based set-up, each CBCT was then retrospectively realigned to best match the liver edge. The fiducial marker position was considered the gold standard. Shifts from the fiducial to liver edge alignment were recorded and an absolute Cartesian shift was calculated. To predict the dosimetric impact of these position errors, an artificial neural network was used to model the dose in the vicinity of the tumor. When aligning to the liver edge, the mean absolute Cartesian shift from fiducial alignment was 4.5 mm (0.73 – 13.3 mm). The lateral, longitudinal, and vertical shifts, were randomly distributed with mean values of 0.19 mm (-5.9 – 8.2 mm), -0.23 mm (-10.5 – 8.8 mm), and 0.43 mm (-9.0 – 8.4 mm), respectively. Thirty-three percent of fractions were found to have an absolute shift greater than 5 mm [our standard planning target volume (PTV) expansion] and 63% of patients had at least one absolute shift greater than 5 mm. In the neural network dose model, a 5 mm shift was associated with a dose reduction of 18.5% + 2.5%. The maximum shift seen was 13.3 mm which was associated with a projected target dose reduction of 45.0% + 5.6%. There were no tumor factors (including tumor size, distance from liver edge to the tumor, and distance from the tumor to the liver dome) that had a statistically significant correlation with worse alignment when matching to the liver edge without fiducials. Our findings demonstrate that treatment set-up without fiducial markers in liver SBRT leads to a mean positional error of 4.5 mm as compared to fiducial-based set-up. In one third of treatments, this variation was greater than our standard PTV expansion and would result in a substantial decrease in delivered dose. We were unable to define a subset of patients where fiducial markers may not be necessary. While these data suggest that with current technology, fiducial markers are necessary for the accurate delivery of liver SBRT; further work is needed to improve on-board imaging and localization techniques to allow for fiducial-less liver SBRT.

Intrafractional Tracking Accuracy of a Transperineal Ultrasound Image Guidance System for Prostate Radiotherapy.

Purpose:The aim of this study is to evaluate the tracking accuracy of a commercial ultrasound system under relevant treatment conditions and demonstrate its clinical utility for detecting significant treatment deviations arising from inadvertent intrafractional target motion.Methods:A multimodality male pelvic phantom was used to simulate prostate image-guided radiotherapy with the system under evaluation. Target motion was simulated by placing the phantom on a motion platform. The tracking accuracy of the ultrasound system was evaluated using an independent optical tracking system under the conditions of beam-on, beam-off, poor image quality with an acoustic shadow introduced, and different phantom motion cycles. The time delay between the ultrasound-detected and actual phantom motion was investigated. A clinical case example of prostate treatment is presented as a demonstration of the utility of the system in practice.Results:Time delay between the motion phantom and ultrasound tracking system is 223 ± 45.2 milliseconds including video and optical tracking system frame rates. The tracking accuracy and precision were better with a longer period. The precision of ultrasound tracking performance in the axial (superior–inferior) direction was better than that in the lateral (left–right) direction (root mean square errors are 0.18 and 0.25 mm, respectively). The accuracy of ultrasound tracking performance in the lateral direction was better than that in the axial direction (the mean position errors are 0.23 and 0.45 mm, respectively). Interference by radiation and image quality do not affect tracking ability significantly. Further, utilizing the tracking system as part of a clinical study for prostate treatment further verified the accuracy and clinical appropriateness.Conclusions:It is feasible to use transperineal ultrasound daily to monitor prostate motion during treatment. Our results verify the accuracy and precision of an ultrasound system under typical external beam treatment conditions and further demonstrate that the tracking system was able to identify important prostate shifts in a clinical case.

Design of a cable-driven hyper-redundant robot with experimental validation

This article presents a test bed for comprehensive study of a cable-driven hyper-redundant robot in terms of mechanical design, kinematics analysis, and experimental verification. To design the hyper-redundant robot, the multiple section structure is used. Each section consists of two rotational joints, a link mechanism, and three cables. In this sense, two degrees of freedom are achieved. For kinematics analysis between the actuator space and joint space, each section of the development is treated as three spherical–prismatic–spherical chains and a universal joint chain (3-SPS-U), which results in a four-chain parallel mechanism model. In order to obtain the forward kinematics from the joint space to task space directly and easily, the coordinate frames are established by the geometrical rules rather than the traditional Denavit–Hartenburg (D-H) rules. To solve the problem of inverse kinematics analysis, we utilize the product of exponentials approach. Finally, a prototype of 24-degrees of freedom hyper-redundant robot with 12 sections and 36 cables is fabricated and an experiment platform is built for real-time control of the robot. Different experiments in terms of trajectories tracking test, positioning accuracy test, and payload test are conducted for the validation of both mechanical design and model development. Experiment results demonstrate that the presented hyper-redundant robot has fine position accuracy, flexibility with mean position error less than 2%, and good load capacity.

Kinematical and dynamical modeling of a multipurpose upper limbs rehabilitation robot

Abstract Knowing accurate model of a system is always beneficial to design a robust and safe control while allowing reduction of sensors-related cost as the system outputs are predictable using the model. In this context, this paper addresses the kinematical and dynamical model identification of the multipurpose rehabilitation robot, Universal Haptic Pantograph (UHP), and present experimental validations of the identified models. The UHP is a Pantograph based innovative robot actuated by two SEAs (Series Elastic Actuator), aiming at training impaired upper limbs after a stroke. This novel robot, thanks to its lockable/unlockable joints, can change its mechanical structure so that it enables stroke patient to perform different training exercises of the shoulder, elbow and wrist. This work focuses on the ARM mode, which is a training mode used to rehabilitate elbow and shoulder. The kinematical model of UHP is identified based on the loop vector equations, while the dynamical model is derived based on the Lagrangian formulation. To demonstrate the accuracy of the models, several experimental tests were performed. The results reveal that the mean position error between estimated values with the model and actual measured values stays in 3 mm (less than 2% of the maximum motion range). Moreover, the error between estimated and measured interaction force is smaller than 10% of maximum force range. So, the developed models can be adopted to estimate motion and force of UHP as well as control it without the need of additional sensors such as a force sensor, resulting in the reduction of total robot cost.

Novel curved surface preparation technique for knee resurfacing

Conventional tools are incapable of preparing the curved articular surface geometry required during cartilage repair procedures. A novel curved surface preparation technique was proposed and tested to provide an accurate low-cost solution. Three shapes of samples, with flat, 30 mm radius and 60 mm radius surfaces, were manufactured from foam bone substitute for testing. Registering guides and cutting guides were designed and 3-D printed to fit onto the foam samples. A rotational cutting tool with an adapter was used to prepare the surfaces following the guidance slots in the cutting guides. The accuracies of the positions and shapes of the prepared cavities were measured using a digital calliper, and the surface depth accuracy was measured using a 3-D scanner. The mean shape and position errors were both approximately ± 0.5 mm and the mean surface depth error ranged from 0 to 0.3 mm, range − 0.3 to + 0.45 mm 95% CI. This study showed that the technique was able to prepare a curved surface accurately; with some modification it can be used to prepare the knee surface for cartilage repair.

Improvement of Ultrasound-Based Localization System Using Sine Wave Detector and CAN Network

This paper presents an improved indoor localization system based on radio frequency (RF) and ultrasonic signals, which we named the SNSH system. This system is composed of a transmitter mounted in a mobile target and a series of receiver nodes that are managed by a coordinator. By measuring the Time Delay of Arrival (TDoA) of RF and ultrasonic signals from the transmitter, the distance from the target to each receiver node is calculated and sent to the coordinator through the CAN network, then all the information is gathered in a PC to estimate the 3D position of the target. A sine wave detector and dynamic threshold filter are applied to provide excellent accuracy in measuring the range from the TDoA results before multilateration algorithms are realized to optimize the accuracy of coordinate determination. Specifically, Linear Least Square (LLS) and Nonlinear Least Square (NLS) techniques are implemented to contrast their performances in target coordinate estimation. RF signal encoding/decoding time, time delay in CAN network and math calculation time are carefully considered to ensure optimal system performance and prepare for field application. Experiments show that the sine wave detector algorithm has greatly improved the accuracy of range measurement, with a mean error of 2.2 mm and maximum error of 6.7 mm for distances below 5 m. In addition, 3D position accuracy is greatly enhanced by multilateration methods, with the mean error in position remaining under 15 mm. Furthermore, there are 90% confidence error values of 23 mm for LLS and 20 mm for NLS. The update in the overall system has been verified in real system operations, with a maximum rate of 25 ms, which is a better result than many other existing studies.

Needle release mechanism enabling multiple insertions with an ultrasound-guided prostate brachytherapy robot.

We present a robotic system for transrectal ultrasound-guided prostate brachytherapy that employs a quick release mechanism to enable multiple needles to be inserted into the prostate prior to plan optimization. The mechanism consists of two actuated fingers that act as needle guides, thereby allowing insertion of both parallel and angled needles. Path planning, including reordering of needles within a batch, is required to avoid collisions with previously inserted needles. We perform two phantom experiments using clinical implant plans. The extra time required for the robotic motions, including finger actuation, is less than three minutes for the entire procedure. Mean position error is measured to be less than 0.5 mm, presumably due to the design of the needle guides, which have a toroidal shape to enable needle angulation.

Robot calibration using a portable photogrammetry system

Abstract This work investigates the potential use of a commercially-available portable photogrammetry system (the MaxSHOT 3D) in industrial robot calibration. To demonstrate the effectiveness of this system, we take the approach of comparing the device with a laser tracker (the FARO laser tracker) by calibrating an industrial robot, with each device in turn, then comparing the obtained robot position accuracy after calibration. As the use of a portable photogrammetry system in robot calibration is uncommon, this paper presents how to proceed. It will cover the theory of robot calibration: the robot's forward and inverse kinematics, the elasto-geometrical model of the robot, the generation and ultimate selection of robot configurations to be measured, and the parameter identification. Furthermore, an experimental comparison of the laser tracker and the MaxSHOT 3D is described. The obtained results show that the FARO laser tracker ION performs slightly better: The absolute positional accuracy obtained with the laser tracker is 0.365 mm and 0.147 mm for the maximum and the mean position errors, respectively. Nevertheless, the results obtained by using the MaxSHOT 3D are almost as good as those obtained by using the laser tracker: 0.469 mm and 0.197 mm for the maximum and the mean position errors, respectively. Performances in distance accuracy, after calibration (i.e. maximum errors), are respectively 0.329 mm and 0.352 mm, for the laser tracker and the MaxSHOT 3D. However, as the validation measurements were acquired with the laser tracker, bias favors this device. Thus, we may conclude that the calibration performances of the two measurement devices are very similar.

A Novel Protocol for Planning and Navigation in Craniofacial Surgery: A Preclinical Surgical Study

In modern craniofacial surgery, the accuracy of osteotomies plays a central role in surgical technique. To reach a higher level of accuracy, many centers use virtual presurgical planning. In the past decade, some surgeons also have applied navigational approaches to craniofacial procedures. In this work, a novel protocol for surgical planning and intraoperative navigation is described and validated in a preclinical setting to determine its accuracy in guiding osteotomies. This study was based on planning a set of osteotomies using 3-dimensional models of computed tomographic images of human skulls. The planned osteotomies were reproduced on real skulls using an optical infrared navigation system. Positions of the performed osteotomies and planned osteotomies were compared. Results were described as the mean positional error and as a Lin concordance coefficient. The Bland-Altman interval of agreement also was defined to assess a range that could include 95% of possible errors. The mean error was 0.044mm (95% confidence interval [CI], -0.128 to+0.216), the Lin concordance interval was 0.999 (95% CI, 0.999-1.000), and the Bland-Altman limit of agreement ranged from -1.500 to+1.589mm. These results show a submillimetric mean error and a very narrow interval of agreement, providing preclinical validation of this new protocol and suggesting that it could be applied in a clinical setting.

Model Predictive Control for Underwater Robots in Ocean Waves

Underwater robots beneath ocean waves can benefit from feedforward control to reduce position error. This letter proposes a method using model predictive control (MPC) to predict and counteract future disturbances from an ocean wave field. The MPC state estimator employs a linear wave theory (LWT) solver to approximate the component fluid dynamics under a wave field. Wave data from deployed ocean buoys are used to construct the simulated wave field. The MPC state estimator is used to optimize a set of control actions by gradient descent along a prediction horizon. The optimized control input minimizes a global cost function, the squared distance from the target state. The robot then carries out the optimized trajectory with an emphasis on real-time execution. Several prediction horizons are compared, with a horizon of 0.8 s selected as having a good balance of low error and fast computation. The controller with the chosen prediction horizon is simulated and found to show a 74% reduction in position error over traditional feedback control. Additional simulations are run where the MPC takes in noisy measurements of the wave field parameters. The MPC algorithm is shown to be resistant to sensor noise, showing a mean position error 44% lower than the noise-free feedback control case.

Simulation of Tropical Cyclones over Bay of Bengal with NCMRWF Regional Unified Model

This study delineates the relative performance of the 12-km resolution NCMRWF regional Unified Model (NCUM-R) over the operational global NCUM (NCUM-G) model. Forecasts of four Bay of Bengal (BoB) landfalling tropical cyclones (TCs) using several different initial conditions (ICs) are used to compare the performance of two models. The position and intensity errors of the TCs are estimated with respect to the India Meteorological Department (IMD) and Joint Typhoon Warning Center (JTWC) best-track datasets and an inter-comparison study is also carried out between IMD and JTWC. The overall results suggest that the NCUM-R simulates the position and intensity of TCs more accurately compared to the NCUM-G. A majority of the TC tracks in the NCUM-G diverge more from the IMD track when compared to NCUM-R simulated tracks. It is also clearly noticed that both the models are more skillful in track prediction when initialized at intensity stages greater than “cyclone” category. However, the mean position errors at different forecast hours and landfall errors of TCs are reduced by approximately 31 and 47% in the NCUM-R simulations compared to NCUM-G simulations, respectively. The mean gain in skill of the NCUM-R in cross track (CT) and along track (AT) error is around 29 and 24% over NCUM-G, respectively. The intensity errors are less in the NCUM-R simulations. The mean rainfall skill scores are considerably improved in the NCUM-R simulations in day-1 and day-2 as compared to the NCUM-G simulations. It is noticed that the mean position errors of the TCs are approximately 8% lower when compared against the JTWC tracks than the IMD tracks. However, the intensity errors are higher against the JTWC than that of IMD most likely due to the averaging period of the wind speed.

Robot-assistant for MRI-guided liver ablation: A pilot study.

Percutaneous ablation under MRI-guidance allows treating otherwise inoperable liver tumors locally using a catheter probe. However, manually placing the probe is an error-prone and time consuming task that requires a considerable amount of training. The aim of this paper was to present a pneumatically actuated robotic instrument that can assist clinicians in MRI-guided percutaneous intervention of the liver and to assess its functionality in a clinical setting. The robot positions a needle-guide inside the MRI scanner bore and assists manual needle insertions outside the bore. The robot supports double oblique insertions that are particularly challenging for less experienced clinicians. Additionally, the system employs only standard imaging sequences and can therefore be used on different MRI scanners without requiring prior integration. The repeatability and the accuracy of the robot were evaluated with an optical tracking system. The functionality of the robot was assessed in an initial pilot study on two patients that underwent MRI-guided laser ablation of the liver. The robot positioned the needle-guide in a repeatable manner with a mean error of 0.35 mm and a standard deviation of 0.32 mm. The mean position error corresponding to the needle tip, measured for an equivalent needle length of 195 mm over 25 fixed points, was 2.5 mm with a standard deviation of 1.2 mm. The pilot study confirmed that the robot does not interfere with the equipment used for MRI-guided laser ablation and does not visibly affect the MR images. The robot setup integrated seamlessly within the established clinical workflow. The robot-assisted procedure was successfully completed on two patients, one of which required a complex double oblique insertion. For both patients, the insertion depth and the tumor size were within the range reported for previous MRI-guided percutaneous interventions. A third patient initially enrolled in the pilot study and was considerably heavier than the others, preventing the use of the robot and requiring several freehand insertion attempts. The robot repeatability and accuracy are appropriate for liver tumors normally treated with MRI-guided ablation. The results of the pilot study endorse the clinical use of the robot in its current form: the robot is fully functional and MRI-compatible in a clinical setting and is suitable for double-oblique needle insertions.

A Multi‐Autonomous Underwater Vehicle System for Autonomous Tracking of Marine Life

This paper presents a multi-autonomous underwater vehicle system capable of cooperatively and autonomously tracking and following marine targets (i.e., fish) tagged with an acoustic transmitter. The AUVs have been equipped with stereo-hydrophones that receive signals broadcasted by the acoustic transmitter tags to enable real-time calculation of bearing-to-tag and distance-to-tag measurements. These measurements are shared between AUVs via acoustic modem and fused within each AUV's particle filter for estimating the target's position. The AUVs use a leader/follower multi-AUV control system to enable the AUVs to drive toward the estimated target state by following collision-free paths. Once within the local area of the target, the AUVs circumnavigate the target state until it moves to another area. The system builds on previous work by incorporating a new SmartTag package that can be attached to an individual's dorsal fin. The SmartTag houses a full inertial measurement unit (INU), video logger, acoustic transmitter, and timed release mechanism. After real-time AUV tracking experiments, the SmartTag is recovered. Logged IMU data are fused with logged AUV-obtained acoustic tag measurements within a particle filter to improve state estimation accuracy. This improvement is validated through a series of multi-AUV shark and boat tracking experiments conducted at Santa Catalina Island, California. When compared with previous work that did not use the SmartTag package, results demonstrated a decrease in mean position estimation error of 25–75%, tag orientation estimation errors dropped from 80° to 30° , the sensitivity of mean position error with respect to distance to the tag was less by a factor of 50, and the sensitivity of mean position error with respect to acoustic signal reception frequency to the tag was 25 times less. These statistics demonstrate a large improvement in the system's robustness when the SmartTag package is used.

HW/SW Codesign and FPGA Acceleration of Visual Odometry Algorithms for Rover Navigation on Mars

Future Mars exploration missions rely heavily on high-mobility autonomous rovers equipped with sophisticated scientific instruments and possessing advanced navigational capabilities. Increasing their navigation velocity and localization accuracy is essential for enabling these rovers to explore large areas on Mars. Contemporary Mars rovers move slowly, partially due to the long execution time of complex computer vision algorithms running on their slow space-grade CPUs. This paper exploits the advent of high-performance space-grade field-programmable gate arrays (FPGAs) to accelerate the navigation of future rovers. Specifically, it focuses on visual odometry (VO) and performs HW/SW codesign to achieve one order of magnitude faster execution and improved accuracy. Conforming to the specifications of the European Space Agency, we build a proof-of-concept system on an HW/SW platform with processing power resembling that to be available onboard future rovers. We develop a codesign methodology adapted to the rover's specifications, design parallel architectures, and customize several feature extraction, matching, and motion estimation algorithms. We implement and evaluate five distinct HW/SW pipelines on a Virtex6 FPGA and a 150 MIPS CPU. We provide a detailed analysis of their cost-time-accuracy tradeoffs and quantify the benefits of employing FPGAs for implementing VO. Our solution achieves a speedup factor of 16× over a CPU-only implementation, handling a stereo image pair in less than 1 s, with a 1.25% mean positional error after a 100 m traverse and an FPGA cost of 54 K LUTs and 1.46-MB RAM.

3D interactive system based on vision computing of direct‐flective cameras

AbstractA bare‐finger 3D interactive technology for portable devices was developed. Using directive‐flective cameras to reform the field of viewing, a blind working range close to the camera is eliminated. Moreover, the algorithm of vision computing, different from skin color detection, is presented to determine the positions of fingertips. The interactive range is workable from 1.5 to 50 cm above the entire surface of the display. The mean position error of less than 1 cm is achieved. This accuracy realizes a camera‐based 3D interactive system allowing for near‐distance functionality. Therefore, floating 3D images can be touched and interacted with, potentially creating more application and intuitive user‐machine interface.

ROBUST PARALLEL MOTION ESTIMATION AND MAPPING WITH STEREO CAMERAS IN UNDERGROUND INFRASTRUCTURE

Abstract. Presently, we developed a novel robust motion estimation method for localization and mapping in underground infrastructure using a pre-calibrated rigid stereo camera rig. Localization and mapping in underground infrastructure is important to safety. Yet it’s also nontrivial since most underground infrastructures have poor lighting condition and featureless structure. Overcoming these difficulties, we discovered that parallel system is more efficient than the EKF-based SLAM approach since parallel system divides motion estimation and 3D mapping tasks into separate threads, eliminating data-association problem which is quite an issue in SLAM. Moreover, the motion estimation thread takes the advantage of state-of-art robust visual odometry algorithm which is highly functional under low illumination and provides accurate pose information. We designed and built an unmanned vehicle and used the vehicle to collect a dataset in an underground garage. The parallel system was evaluated by the actual dataset. Motion estimation results indicated a relative position error of 0.3%, and 3D mapping results showed a mean position error of 13cm. Off-line process reduced position error to 2cm. Performance evaluation by actual dataset showed that our system is capable of robust motion estimation and accurate 3D mapping in poor illumination and featureless underground environment.

ROBUST PARALLEL MOTION ESTIMATION AND MAPPING WITH STEREO CAMERAS IN UNDERGROUND INFRASTRUCTURE

Presently, we developed a novel robust motion estimation method for localization and mapping in underground infrastructure using a pre-calibrated rigid stereo camera rig. Localization and mapping in underground infrastructure is important to safety. Yet it’s also nontrivial since most underground infrastructures have poor lighting condition and featureless structure. Overcoming these difficulties, we discovered that parallel system is more efficient than the EKF-based SLAM approach since parallel system divides motion estimation and 3D mapping tasks into separate threads, eliminating data-association problem which is quite an issue in SLAM. Moreover, the motion estimation thread takes the advantage of state-of-art robust visual odometry algorithm which is highly functional under low illumination and provides accurate pose information. We designed and built an unmanned vehicle and used the vehicle to collect a dataset in an underground garage. The parallel system was evaluated by the actual dataset. Motion estimation results indicated a relative position error of 0.3%, and 3D mapping results showed a mean position error of 13cm. Off-line process reduced position error to 2cm. Performance evaluation by actual dataset showed that our system is capable of robust motion estimation and accurate 3D mapping in poor illumination and featureless underground environment.

Position referenced force augmentation in teleoperated hydraulic manipulators operating under delayed and lossy networks: A pilot study

Position error between motions of the master and slave end-effectors is inevitable as it originates from hard-to-avoid imperfections in controller design and model uncertainty. Moreover, when a slave manipulator is controlled through a delayed and lossy communication channel, the error between the desired motion originating from the master device and the actual movement of the slave manipulator end-effector is further exacerbated. This paper introduces a force feedback scheme to alleviate this problem by simply guiding the operator to slow down the haptic device motion and, in turn, allows the slave manipulator to follow the desired trajectory closely. Using this scheme, the master haptic device generates a force, which is proportional to the position error at the slave end-effector, and opposite to the operator’s intended motion at the master site. Indeed, this force is a signal or cue to the operator for reducing the hand speed when position error, due to delayed and lossy network, appears at the slave site. Effectiveness of the proposed scheme is validated by performing experiments on a hydraulic telemanipulator setup developed for performing live-line maintenance. Experiments are conducted when the system operates under both dedicated and wireless networks. Results show that the scheme performs well in reducing the position error between the haptic device and the slave end-effector. Specifically, by utilizing the proposed force, the mean position error, for the case presented here, reduces by at least 92% as compared to the condition without the proposed force augmentation scheme. The scheme is easy to implement, as the only required on-line measurement is the angular displacement of the slave manipulator joints.