GPS World

Category: Transportation

  • Lane-level positioning with low-cost map-aided GNSS/MEMS IMU integration

    Lane-level positioning with low-cost map-aided GNSS/MEMS IMU integration

    Lane errors in a three-lane road, giving lane determination (yellow triangle). (Photo: Pavel Vinnik/Shutterstock.com)
    Lane errors in a three-lane road, giving lane determination (yellow triangle). (Photo: Pavel Vinnik/Shutterstock.com)

    A lane-keeping system uses a sensor-fusion engine integrating GPS and an IMU with a two-stage map-matching algorithm. The system does not require explicit lane-level geo-referencing, saving massive storage required for lane-level spatial reference information, and reduces the computational complexity of the map-matching algorithm.

    By Mohamed M. Atia, Carleton University and Allaa Hilal, Intelligent Mechatronics Systems

    Lane determination is an important feature of advanced automotive navigation and guidance systems. It can be used in advanced driving assistance systems (ADAS), lane-departure warnings, and self-driving cars to perform lane-level, turn-by-turn guidance and control. It is also valuable information for telematics applications such as usage-based insurance. Lane-estimation systems have been dominated by vision and infrared sensors. Light detection and ranging (lidar) has also been used as a lane-determination technique. Those systems depend on visually recognizable features and landmarks that may not be available in some areas due to weather conditions or unstructured environments.

    In addition, visual data processing may need specialized accelerators and parallel computing platforms to satisfy real-time constraints. To explore other alternatives, several research projects have started to investigate the feasibility of using low-cost global positioning and navigation technologies such as GPS, micro-electromechanical systems (MEMS) inertial measurement units (IMU) and geographical information systems (GIS) as an alternate lane-determination technology. However, most current systems have two main drawbacks: they use high-end RTK GPS, which suffers from coverage issues, and they use explicit lane geo-referencing, which leads to increased storage and processing.

    Here we investigate the feasibility of using standard GPS fused with low-cost MEMS-IMU and a road network that includes lane information but not explicitly storing geo-referenced lane-level links.

    The accuracy of Standard Positioning Service (SPS) GPS is within 3.351 meters (m) with a 95 percent confidence level. Figure 1 shows the results of standard single-point positioning test for a stationary receiver.

    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 1. Standard GPS 2D position accuracy in a stationary test. (Figure: Mohamed M. Atia and Allaa Hilal)

    The standard lane width in North America is approximately 3.6 m, requiring an unbiased precise positioning solution of much less than 1.8 m. If a safety margin of 50% is considered, unbiased precise positioning of less than 0.9 m is needed. Therefore, a standard SPS GPS technology may not be precise enough to accurately determine the vehicle’s lane. Advanced precise positioning technology like differential GPS (DGPS) can be used with high-resolution lane-level maps to achieve the lane determination.

    However, these techniques may require additional cost/infrastructures and extra processing. To target a lower cost lane-determination system, this work suggests the fusion of measurements from a standard GPS, MEMS IMU and road-level network.

    The work includes a sensor-fusion engine that is developed to integrate GPS and IMU using a loosely coupled extended Kalman filter (EKF). Then, a two-stage map-matching algorithm using a Hidden-Markov-Model (HMM) and a least-squares (LS) regression is developed.

    The system does not require explicit lane-level geo-referencing; consequently, it saves massive storage required to save explicit lane-level spatial reference information, and it reduces the computational complexity of the HMM algorithm by reducing the number of road segments the HMM needs to decode. The overall system is illustrated in Figure 2.

    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 2. Illustration of the proposed system. (Figure: Mohamed M. Atia and Allaa Hilal)

    PROBLEM DEFINITION

    A geometric illustration of the problem is shown in Figure 3. The road-network map is represented as a set of connected segments. Each road segment is defined by a straight line segment with a start position and end position. Curved roads are approximated by a sufficiently large number of straight line segments. Based on this notation and geometric illustration, the estimation problem that this article is addressing is the determination of the lane on which the vehicle is moving.

    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 3. Illustration of the lane determination problem. (Figure: Mohamed M. Atia and Allaa Hilal)

    Map-Matching with Hidden-Markov Model. The simplest map-matching method, point-to-curve-matching, is performed by searching for the nearest road segments within a threshold from the current vehicle’s position. The distance is calculated between the vehicle’s position and its projection on the map segment. However, this approach is sensitive to state estimation errors, and it fails at intersections, joins, branches or dense parallel roads. For example, Figure 4 shows a situation where biased GNSS position measurements exist, and the wrong map segment is selected because of the pure dependence on the distance metric only (for instance, D1 is less than D2).

    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 4. Wrong map-segment selection in intersection. (Figure: Mohamed M. Atia and Allaa Hilal)

    To avoid these errors and to improve map-matching accuracy, the matching criteria must include several constraints such as map topology (connectivity), vehicle dynamics, road geometry and legal direction of motions. In this work, to consider these constraints, we keep a recent portion of the vehicle motion history and use it in the matching criteria. This strategy is known as curve-to-curve matching.

    To process a noisy stream of data, the HMM algorithm is used. A Markov model is a stochastic model that describes a sequence of states. The transition from one state to another can be modeled by a conditional transition probability.

    If the states are not directly observable (hidden) but can be indirectly observed through a sequence of outputs, the process is called a Hidden Markov Process. The HMM in this case is characterized by the transition probability and an emission probability that represents the probability that a given state generates a certain observable.

    Both transition probability and emission probability constitute the Bayesian network of HMM. A fundamental problem of HMM is that, given a sequence of outputs, what is the best sequence of states that explains the observed outputs? This problem is solved by selecting the sequence of states that maximize the HMM probability.

    This estimation process, called decoding, is solved using the Viterbi algorithm. In the proposed system, the hidden states represent map links, and the observable outputs are the vehicle poses. To develop a robust map-matching framework, the vehicle pose history, roads geometry, and map topology constraints must be considered. Therefore, the emission and transition probabilities of an HMM are formulated such that they reflect all of these constraints. The Bayesian network of the HMM for our system is shown in Figure 5. The vehicle states (poses) is obtained from the INS/GNSS filter described shortly.

    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 5. Hidden Markov model for vehicle’s state map-matching. (Figure: Mohamed M. Atia and Allaa Hilal)

    In the proposed work, the length of the processed buffer of the vehicle’s state is determined based on the traveled distance. The aim is to accumulate a reasonable geometric knowledge about the trajectory segment that enables the HMM to accumulate enough geometric and topological constraints to be able to select the correct sequence of road segments in difficult intersections, joins and exit/entry roads.

    EKF GNSS/INS SYSTEM

    The navigation problem can be modeled as a dynamic system of states vector x(t) as follows:

    (Figure: Mohamed M. Atia and Allaa Hilal) (1)
    (Figure: Mohamed M. Atia and Allaa Hilal) (2)

    (Figures: Mohamed M. Atia and Allaa Hilal)

    where f(.) is a nonlinear dynamic model, w(t) is a stochastic system noise vector, u(t) is a control signal vector that triggers the transition from current state to a future state, y(t) is external measurements vector (observables), h(.) is a nonlinear measurement model and v(t) is a stochastic measurement noise vector. Using first-order Taylor series approximation, (1) and (2) can be linearized as follows:

    (Figure: Mohamed M. Atia and Allaa Hilal) (3)
    (Figure: Mohamed M. Atia and Allaa Hilal) (4)

    (Figure: Mohamed M. Atia and Allaa Hilal) (5)

    (Figure: Mohamed M. Atia and Allaa Hilal) (6)

    (Figures: Mohamed M. Atia and Allaa Hilal)

    A Kalman filter calculates an optimal estimation of provided that w(t) and v(t) are zero-mean Gaussian noise vectors with covariance matrices defined by:

    (Figure: Mohamed M. Atia and Allaa Hilal) (7)

    (Figure: Mohamed M. Atia and Allaa Hilal) (8)

    and δx is the error vector with zero-mean and a covariance matrix P defined by:

    (Figure: Mohamed M. Atia and Allaa Hilal) (9)

    Using zero-hold discretization where derivative is approximated by:

    (Figure: Mohamed M. Atia and Allaa Hilal) (10)

    where T is the sampling time, equations involving HMM probability can be written in discrete form as follows:

    (Figure: Mohamed M. Atia and Allaa Hilal)(11)

    (Figure: Mohamed M. Atia and Allaa Hilal)(12)

    The optimal estimation of the error vector, δxk, given measurements, yk, is calculated using two steps: prediction,

    (Figure: Mohamed M. Atia and Allaa Hilal) (13)

     (Figure: Mohamed M. Atia and Allaa Hilal) (14)

    and update,

    (Figure: Mohamed M. Atia and Allaa Hilal)(15)

    (Figure: Mohamed M. Atia and Allaa Hilal)(16)

    (Figure: Mohamed M. Atia and Allaa Hilal)(17)

    (Figures: Mohamed M. Atia and Allaa Hilal)

    In INS/GNSS systems, the dynamic system state transition (x(t)) is triggered by IMU sensors (accelerometer and gyroscopes) while GNSS measurements are used as observables (y(t)). The observables update in our case is GNSS position and velocity. Therefore, the measurement error model is defined as follows:

    (Figure: Mohamed M. Atia and Allaa Hilal)(18)

    where H is defined as follows:

    (Figure: Mohamed M. Atia and Allaa Hilal)(19)

    Lane Estimation. When the road segments have been accurately selected based on the filtered vehicle’s pose, the projection of the vehicle’s positions on segment lanes can be easily calculated knowing the lane widths and number of lanes. The sum of squared errors for each lane is then calculated by:

    (Figure: Mohamed M. Atia and Allaa Hilal)(20)

    where N is number of epochs, and pv is the projection of vehicle’s position on lane. The lane associated with the minimum error is selected as the designated lane.

    (Figures: Mohamed M. Atia and Allaa Hilal)

    Lane-Change Detection. If a lane change occurred within the processed buffer of data, the least-squares regression will not converge to the correct lane. Therefore, the buffer needs to be partitioned at the lane-switch locations. Thus, a lane-change detection module is developed. In this work, a lane-change detection method is designed based on capturing the patterns of the vehicle’s orientation and raw gyroscope measurements. The heading and raw gyroscope measurements during lane changes are shown in Figure 6 and Figure 7.

    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 6. Vehicle’s heading during lane change to left. (Figure: Mohamed M. Atia and Allaa Hilal)
    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 7. Vehicle’s gyroscope measurements during lane change to left. (Figure: Mohamed M. Atia and Allaa Hilal)

    The general pattern that the lane-change module detects is a peak or a valley in azimuth accompanied by a peak/valley or valley/peak sequence in the gyroscope measurements. To detect peaks and valleys, the standard deviation of a moving window of data is calculated and compared to a peak/valley threshold. If both gyro and azimuth peak/valley sequence are consistent and matched with the pattern described above, a lane change is declared.

    Two algorithm phases of processing are then applied:

    Acquisition Phase. GNSS and IMU measurements are fused in the main EKF, and HMM map-matching is performed and a lane is estimated. The innovation sequence of the main EKF, which is the difference between the predicted state and GNSS updates, is calculated over a buffer of data. If the innovation sequence is within a small threshold and no lane change has been detected, the acquisition phase is concluded and the tracking phase begins.

    Tracking Phase. Two EKF filters are initiated. One EKF accepts position updates from the projection of the vehicle’s position on the selected lane, and the other EKF accepts GNSS position updates only. A discrepancy measure is evaluated between the two EKF instances for a short window of time. If this discrepancy measure is higher than a threshold, a temporary GNSS deviation is assumed and the system keeps reporting the current lane as the designated lane. If GNSS measurements started to be centered again on the new lane, a lane change is confirmed and the output of the first EKF instance will be the correct state. Otherwise, this lane change is declared as false and the second EKF output is the correct output. The overall block diagram of the proposed system is shown in Figure 8.

    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 8. Overall block diagram of the proposed system. (Figure: Mohamed M. Atia and Allaa Hilal)

    TESTS AND RESULTS

    The proposed system has been tested on a computer connected to a GNSS receiver and an automotive MEMS-grade IMU, and road-network map data. A GPS-enabled camera was installed to capture video of the experiment, to be used as a ground truth to verify the results of our algorithms. Sensor specifications are given in Table 1 and Table 2. The effect of level arm (distance between IMU and GNSS antenna) was not considered in this implementation.

    (Table: Mohamed M. Atia and Allaa Hilal)
    TABLE 1. GNSS receiver accuracy. (Table: Mohamed M. Atia and Allaa Hilal)
    (Table: Mohamed M. Atia and Allaa Hilal)
    TABLE 2. IMU specifications. (Table: Mohamed M. Atia and Allaa Hilal)

    Three testing trajectories were collected during July 2015 through Highway 400 from Wilson Avenue in the south to Davis Drive in the north. Approximately 65 kilometers of trip data was collected. The data included some urban areas but was mostly open sky. It also included challenging road intersections and road joining/branching points. The experimental setup was designed such that the system automatically started when the vehicle’s engine was turned on. A Linux OS was installed on the gigabyte computer box, and a data acquisition firmware was configured to automatically begin when the computer starts. Measurements from the GNSS receiver at 1 Hz and the IMU at 50 Hz were synchronized on the computer. The main algorithm including GNSS/INS fusion and map-matching was developed in native ANSI C language for efficient processing. Original raw IMU data was set to 50 Hz down-sampled to 5 Hz. Within this interval, the real-time system could fetch map information from a cached database file, perform basic prediction steps and implement the forward calculation of a Viterbi algorithm (including calculation of emission and transition probabilities) that is needed for the HMM map-matching step.

    Lane-Determination Results. The lane estimation results were logged and time-tagged. Using the video recording, the ground truth lane-level solution was visually inspected and manually recorded in a file. Since both the video camera and the proposed INS/GNSS/maps systems log data tagged by GPS time, synchronization between ground truth and the estimated lane were possible. The estimated lanes were visually inspected record by record and results were saved in an Excel sheet. The results were written into a time-tagged file where each row can be easily visually inspected by looking at the portion of images corresponding to the same time-tag. The time-tag used was the UTC-time contained in the NMEA GNSS raw measurements. The overall accuracy of the proposed system in lane determination is shown in Table 3.

    (Table: Mohamed M. Atia and Allaa Hilal)
    TABLE 3. Lane-estimation accuracy. (Table: Mohamed M. Atia and Allaa Hilal)

    Figure 9 and Figure 10 show example snapshots of the visual inspection software tool developed to evaluate the accuracy of the system. As can be seen in the figures, an image of the road that indicates the correct lane is displayed in the upper graph, while the estimated lane information is displayed along with road information including lane errors in the lower graph. Figure 10 shows that the system can identify the correct lane when the number of lanes is increased.

    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 9. Lane errors in a three-lane road. (Figure: Mohamed M. Atia and Allaa Hilal)
    (Figure: Mohamed M. Atia and Allaa Hilal)
    FIGURE 10. Lane errors in a four-lane road. (Figure: Mohamed M. Atia and Allaa Hilal)

    CONCLUSION

    This work described a low-cost lane-level positioning system using a conventional GNSS receiver, MEMS IMUand commercially available road-level network without the need for explicit spatial storage of lanes. The research used a conventional GNSS receiver and MEMS IMU with a computationally efficient two-stage HMM-based map-matching algorithm that avoids the explicit use of lanes as hidden states, which significantly reduces the size of the HMM network and consequently enhance its real-time performance. The proposed system provides an alternative lane determination method without the need for computationally expensive vision/lidar methods that may fail in dark, foggy or dynamically changing environments. The work showed extensive experiments under different road sections, showing an average lane-determination accuracy of 97.14%.

    ACKNOWLEDGMENTS

    This work was first presented at ION International Technical Meeting, January 2018.

    MANUFACTURERS

    The system comprises an Intel Celeron N2807 1.58-GHz Mini PC connected to a u-blox EVK-7P kit GNSS receiver and an automotive MEMS-grade IMU 3D space sensor IMU from YOST Labs, and road-network map data from HERE. A GPS-enabled HP f310 car camcorder captured video.

    MOHAMED M. ATIA received a Ph.D. in electrical and computer engineering from Queen’s University at Kingston. He is assistant professor and founder/director of the Embedded Multi-sensor Systems research laboratory in Carleton University, Ontario, Canada.

    ALLAA HILAL received a Ph.D. degree in electrical and computer Engineering from the University of Waterloo. She is director of the innovation and emerging technology department at Intelligent Mechatronic Systems, a connected-car company based in Waterloo, Canada.

  • Warwick U to test location system for intelligent vehicles

    Intelligent vehicles and smart devices could gain more accurate location awareness by fusing GNSS and Wi-Fi signals. A test for this is the focus of an Innovate UK project led by Spirent Communications and involving the Warwick Manufacturing Group (WMG) at the University of Warwick.

    The £694k Enhanced Assured Location Simulator Leveraging Wi-Fi and GNSS Sensor Fusion (ELWAG) project will seek to develop and test the pioneering hybrid Wi-Fi and GNSS location system in a cost-effective, repeatable and safe environment so that manufacturers can verify its performance.

    International Manufacturing Centre at WMG. (Photo: WMG)

    Researchers at WMG, led by Matthew Higgins, will play a significant role in the project. They will take physical layer measurements of both Wi-Fi and GNSS signals in autonomous vehicle scenarios in and around the University of Warwick campus and the local urban road network.

    The measurements will then assist in Spirent’s development of an RF propagation model that will overlay RF effects on its Wi-Fi Access Point simulator.

    WMG researchers will then perform RF validation and verification activities around the developed model, to provide a level of assurance on its performance.

    “The safety and functional assurance of future autonomous vehicles will be one of the many critical paths to large consumer adoption,” said Higgins, who is an associate professor in the intelligent vehicles group at WMG. “Through this project, we will contribute towards providing innovative solutions to the challenges of using sensor fusion in this testing context.”

    “This is a highly technical project, which will require a holistic understanding of the signal propagation characteristics between satellites, infrastructure and vehicles. The results will impact future autonomous testing methodologies,” said Erik Kampert, senior research fellow at WMG.

    The ELWAG project will run for 18 months, and also involves Chronos Technology.

    Project background. Many devices currently rely on a singular location technology (typically GPS), which is one type of the wider eco system of GNSS. These systems, whilst becoming more capable, still suffer at times from the user’s environment — typically in urban areas where buildings and other cityscape features interfere with the signal.

    The urban environment is, however, where most users need to know their location to the highest level of accuracy, due to increasing population or device density. Wi-Fi signals exist almost universally within dense urban areas, so there is a possibility of fusing these signals with the GNSS signals to identify one’s location very accurately.

    “Currently Wi-Fi access point plus GNSS simulation can only be achieved in an ad hoc manner and does not allow for the testing of moving vehicles, multipath effects, insertion of data errors, spoofing and above all controlled, repeatable testing,” said Mark Holbrow, director of engineering and product development at Spirent’s positioning business unit.

    “In the autonomous vehicle sector location accuracy can vary by up to 5 meters, which is unacceptable from a safety perspective. Bringing that accuracy down to 30 centimeters through sensor fusion will have substantial implications for autonomous navigation.”

    Self-aware smart devices. The need for smart devices to have a highly accurate self-awareness of their own location, and the location of other smart devices around is becoming increasingly important.

    In applications such as autonomous vehicles and transport systems, accurate location awareness is an obvious operational requirement for their safe operation in and around other vehicles, pedestrians or infrastructure.

    In the personal devices space, smartwatches, phones and health monitoring and exercise aids are all striving to be able to make a judgment of the user’s state based upon location.

    In the emergency and security services space, knowing the location of people and objects is also increasingly important as to target response capabilities effectively.

  • NovAtel pioneers autonomous solutions with positioning engine, corrections services, integrity research

    NovAtel pioneers autonomous solutions with positioning engine, corrections services, integrity research

    NovAtel has demonstrated high-accuracy positioning performance using automotive-grade GNSS chipsets Teseo APP and Teseo V from STMicroelectronics. Combining automotive-grade multi-frequency GNSS chipsets with positioning algorithms and correction services from NovAtel improves the achievable positioning accuracy available to automotive users and provides a solution suitable for autonomous operation.

    According to the company, these chipsets provide multi-frequency GNSS data for precise point positioning (PPP) and real-time kinematic (RTK) to enable accurate positioning capabilities. Teseo APP features built-in integrity checking for use in safety-critical systems, whereas Teseo V is used for non-safety-critical precise positioning applications.

    The collaboration between the two companies is designed to reach car manufacturers and Tier 1 suppliers for future production models.

    Driven Today. “STMicro is one of many chipset manufacturers coming to market with dual-frequency chipsets targeting the automotive sector,” said Jonathan Auld, VP Engineering and Safety Critical Systems for NovAtel. “We are taking advantage of their expertise in automotive measurement engines for high-volume, cost-effective reliable positioning. NovAtel brings high-precision algorithm expertise and integration with global corrections supplied by Hexagon Correction Services to this initiative.”

    NovAtel’s positioning engine combines the GNSS measurements from these chipsets with inertial measurement unit (IMU) data and Hexagon Correction Services to deliver centimeter-level PPP positioning solutions in real time.

    “Working closely with STMicroelectronics allowed us to innovate and drastically reduce time to market of our assured positioning solution tailored specifically for safe positioning of autonomous vehicles,” added Auld.

    Comparison of GNSS Performance possible in automotive today (red), L1 automotive with corrections (green) and L1/L2 automotive with corrections (blue).

    Driverless Tomorrow. “Precise absolute positioning is just one piece of the overall autonomous vehicle puzzle and must be done with safety and integrity concepts in mind.” Auld pointed to the partnership announced in 2016 between NovAtel, the Illinois Institute of Technology, and Stanford University to conduct leading-edge research to determine how GNSS technology can deliver a positioning solution that meets both the safety and accuracy requirements of autonomous automotive vehicles.

    Previous research by academia and industry into GNSS integrity produced the successful WAAS program for aviation. The new work underway will extend the scope to include the autonomous ground vehicle use case. The research includes updated and expanded concepts for high-integrity carrier-phase algorithms as well as expanded threat models and safety monitors.

    At the Automotive Tech.AD in Berlin, Auld added: “Today the primary use case for positioning in navigation is single-frequency GNSS, with up to 2 constellations, using narrowband RF and antennas, obtaining accuracy at the 1–2 meter level. This is primarily done with pseudorange-based positioning techniques, with some carrier-phase assistance. There are no functional safety standards, and so safety data is provided on the output solution.”

    Autonomous Requirements. By contrast, he continued, autonomous operation will require lane-level and better accuracy: 3D centimeter to decimeter absolute positioning. This means multi-frequency, multi-constellation receivers and antennas to improve overall accuracy and increase available measurements. It will also require increased availability through sensor fusion with IMUs and other sensors. All of this must be brought together through a functionally safe development process targeted at ISO26262 Automotive Safety Integrity Level (ASIL) B.

    Moving from meter to centimeter level position requires additional processing to handle all the added signals coming in; residual monitoring and observation exclusion, and carrier phase, “the key to centimeter-level positioning,” as opposed to code phase. The vehicle’s localization system must include enhanced positioning algorithms for multipath mitigation, a fast converging corrections network, enhanced Kalman Filters, and sophisticated sensor fusion.

    Flexible Integration. NovAtel’s positioning engine architecture enables a flexible integration with different GNSS receiver chipsets, augmentation sensors and processor environments, providing automotive manufacturers with additional flexibility when it comes to sourcing of components and subsystems of advanced driver assistance systems (ADAS) and autonomous driving solutions.

    The positioning engine is being developed to ASIL-B standards and will include a proprietary GNSS integrity solution to ensure safe positioning within defined protection limits tailored to the customer’s application requirements.

  • PNT Roundup: Uber turns on shadow matching

    PNT Roundup: Uber turns on shadow matching

    The technological underpinning for stock markets’ techno-darlings doesn’t always work perfectly. That problem produces lost revenue and lost value. So Uber, for one, has done something about it, partly based on research developed by Paul Groves at University College London and featured in the February 2012 cover story of GPS World.

    Smartphones finding each other in the urban landscape constitute Uber’s business basis. When driver and rider can’t find each other, because they’re on opposite sides of the street or even opposite sides of the block, a ride can’t happen. In the GPS world, we call this multipath, reflected signals, shadowing or simply urban canyon. In Uber parlance it is “wasted supply.”

    To eradicate it, Uber acquired Shadow Maps in 2016 and has integrated the company’s technology into the Uber app. Beta testing now goes on in 15 cities; early results indicate that positioning accuracy has improved twofold.

    The Shadow Maps process, derived from Groves’ shadow-matching concept, directs the Uber algorithm to examine a 3D rendering of the cityscape and perform a probabilistic estimate of user location based on — simultaneously — which satellites are in direct line-of-sight and which aren’t, in conjunction with predicted satellite location, or almanac.

    The process uses ray tracing, color-coding satellite signals by strength to predict likely locations. Each probability calculation takes 20–100 milliseconds, and can run every four seconds for riders and more frequently for drivers, according to Uber engineers and former Shadow Maps principals Andrew Irish and Danny Iland.

    “You just want to have a better, tighter estimate to account for how much faster cars move,” Irish said.

    Prior Work. Paul Groves has researched this area for nearly a decade at the Space Geodesy and Navigation Laboratory, University College London, where he is an associate professor. Lei Wang won ION’s Parkinson Award for his Ph.D. thesis on shadow matching and now works at Apple. Marek Ziebart is a professor and vice-dean, research, UCL.

    “There are many different approaches to 3D-mapping-aided GNSS and several different research groups around the world working on them,” said Groves. “At UCL, we have been integrating shadow matching with 3D-mapping-aided GNSS ranging algorithms. We now have a real-time demo system running on an Android smartphone, albeit limited to Central London. By making full use of the new Android ‘raw measurements’ capability, we get around a factor of 5 accuracy improvement over conventional single-epoch GNSS in dense urban areas.”

    “It’s great to see people actually making use of our research rather than it just languishing in research papers. The more widely that shadow matching and other 3D-mapping-aided GNSS techniques are used, the better.”
    In February 2012, Groves and his co-authors presciently wrote:

    “A practical shadow-matching algorithm must be implementable in real time on a mobile device. Three models may be considered.

    • A network-based solution, whereby GNSS measurements are transmitted to a server, which stores the building boundary data, computes a solution and then sends it to the user.
    • A handset-based solution, where the shadow-matching algorithm is run on the handset, which also stores the building boundary data.
    • A hybrid model, whereby the shadow-matching algorithm runs on the handset, but the building boundary data is streamed from a server as and when required.

    “Using stored or streamed building boundaries, fewer than 50 comparison and addition operations are required to calculate an overall shadow-matching score for one candidate position with two GNSS constellations. Therefore, shadow matching may be performed in real time on a mobile device with several hundred candidate positions, where necessary.”

    The magazine article was based on a presentation at the European Navigation Conference 2011 in London. The authors will present their latest research, reflecting significant progress over the last seven years, at ION GNSS+ 2018 in Miami, Sept. 24-28.

  • Inertial Sense debuts rugged micro GNSS-INS module

    Inertial Sense has announced the availability of a micro-sized rugged version of its combined GNSS-INS module, which has an onboard GNSS receiver as well as a fully fused inertial navigation solution.

    Designed to fill autonomous vehicle and sensing needs, the module is also available in AHRS/IMU versions.

    At 10 grams and with 1 x 1-inch footprint, the solution provides accuracy of 0.1-degree roll/pitch and 0.3-degree dynamic heading. It is also ITAR-free module.

    The modules represent 15 years of inertial navigation and motion measurement experience, according to the company.

    “When I set out on this journey to provide an accurate and low-cost navigation solution, I wanted to produce a product that engineers could purchase off the shelf, hassle free,” said company founder Walt Johnson. “In my past as a UAV engineer, I was always looking for ways to save myself time and money. It’s all about convenience. There is no need to spend time choosing IMU sensors and writing the algorithms to fuse navigation data. We provide it all for you.”

  • Sharper Shape introduces multi-sensor payload for manned helicopters

    Sharper Shape, a provider of unmanned aerial utility inspection solutions, has released the Heliscope 2.0, an onboard payload system that expands the company’s aerial sensing portfolio into the manned helicopter industry.

    According to the company, the Heliscope 2.0 integrates multiple sensor systems into a single, lightweight helicopter payload, capable of simultaneously collecting a range of data types required for utility maintenance and vegetation management inspections.

    Deployment of the Heliscope 2.0 enables optimized inspection and maintenance schedules, offering potential cost savings in those operational activities by as much as 50 percent.

    The Heliscope 2.0 also stands out with its flexible mounting configurations and ability to adapt for mounting on many different helicopter types.

    For example, the system can be mounted on most Bell Jet/Long Ranger helicopters using its FAA-approved nose mount, or attached to numerous other typical helicopter models using its unique Glider aerodynamic sled.

    The U.S. Federal Aviation Administration (FAA) permits mounting the Heliscope 2.0 to helicopters by using the cargo hook found on many helicopter models; this user-friendly method is approved by FAA under a classification for gliders.

    “While drones are a very flexible and safe method for performing utility inspections, there are situations where manned helicopters are the preferred vehicle to host sensors during certain utility inspections,” said Mikko Saarisalo, Sharper Shape’s vice president of drones and project lead for the Heliscope 2.0 project. “The new Heliscope 2.0 provides a solution for those situations where we need to operate over greater distances or in harsher environments than the drones can easily accommodate. This system takes our data harvesting efficiency and productivity up to a level unprecedented in the industry.”

    CORE includes algorithms to automatically analyze lidar point clouds and quickly generate utility vegetation management reports. Further, its unique automatic issue detection (AID) machine vision software uses artificial intelligence (AI) to eliminate the daunting task of performing frame-by-frame image data inspection, allowing personnel to focus on other aspects of inspection compliance.

    CORE applications work equally well with either Sharper Shape’s proven unmanned aerial inspection services, or with the new Heliscope 2.0 manned aircraft solution.

    “The fact that the Heliscope 2.0 integrates fully with our CORE software suite is a huge benefit,” said Sharper Shape CEO Ilkka Hiidenheimo. “We can collect all the key inspection assets and measurements in one high-speed pass, and then easily pass these files to our CORE suite for automatic processing. Sharper Shape is the only company on the market that offers this range of options for collecting aerial data and for processing this data automatically into a wide range of digital report formats.”

    The Heliscope 2.0 system is now available for immediate contract services in the U.S., South America and Europe.

  • Cohda V2X-Locate system beats GPS black spots

    Australian company Cohda Wireless has released a vehicle positioning system to eliminate GPS black spots in “urban canyons” between high-rise buildings.

    Using Cohda’s expertise in developing collision avoidance systems for mines, the vehicle-based system, V2X-Locate, can identify vehicle position to sub-meter accuracy in environments that degrade GPS accuracy, such as tunnels, underground carparks and between high-rise buildings.

    As well as enhancing current connected vehicles, V2X-Locate delivers a critical component for connected autonomous vehicles (CAV), which will require uninterrupted positioning data to safely navigate on roads, the company said.

    Image: Cohda Wireless
    Image: Cohda Wireless

    Cohda has designed V2X-Locate to enable equipped vehicles to identify their location using existing Smart City V2X (vehicle-to-everything) roadside infrastructure from any standards-based manufacturer.

    Cohda Wireless Chief Technology Officer Paul Alexander said V2X-Locate was a globally unique product. “We solve the problem caused by GPS and satellite-based positioning systems not working in all use-cases,” he said.

    “If you’re in a major downtown area with tall buildings, or in a tunnel or in an underground parking lot, a GPS system can fail, preventing it from delivering accurate results,” Alexander said. “As well as being inconvenient for current drivers, this is not an option as we enter the era of driverless cars. The V2X-Locate breakthrough is to position the vehicle with sub-meter accuracy by using the existing communications signals produced by V2X Smart City infrastructure deployments. The result is that V2X-Locate can eliminate positioning black spots in city centers where they are most likely to occur.”

    Cohda Wireless demonstrated V2X-Locate in a 2017 trial in New York City, where it repeatedly demonstrated sub-meter accuracy while driving along Sixth Avenue, which has the tallest buildings in the Big Apple. Comparably tested GPS-based systems were as much as tens of meters off-course, at times showing cars driving through buildings.

    Alexander said Cohda Wireless had designed V2X-Locate by using its experience developing collision avoidance technology for underground mines. “The hardest place to do positioning is one kilometer underground with a cubic kilometer of copper above your head,” he said.

    “That’s where V2X-Locate was born,” Alexander said. “Cohda has worked in that area for several years, providing accurate positioning for vehicles where no GPS connectivity is available. We’ve now successfully migrated that technology from mine sites of the outback to the urban canyons of New York City.”

    V2X_Locate uses the NXP SAF5400 single-chip modem for V2X. (Photo: NXP)

    Both Cohda’s standard V2X onboard units and roadside units use the NXP RoadLINK chipset, which supports V2X-Locate’s highly accurate performance by delivering multipath channel tracking.

    After a pre-release international roadshow in October last year, Cohda Wireless received strong interest in V2X-Locate from both Smart Cities and Tier 1 automotive manufacturers. To meet that demand, Cohda Wireless has released a V2X-Locate Evaluation Kit, which contains the system and four roadside unit devices, which equip prospective customers to put V2X-Locate through its paces.

  • GSA, Joint Research Centre test automotive eCall with Spirent

    Spirent Communications plc is working with the European Commission’s Joint Research Centre (JRC) to help implement the eCall system, which is required in new cars sold in Europe starting in April.

    Experts from the JRC have been working with Spirent GNSS test equipment during the European GNSS Agency (GSA) eCall test campaign. The campaign aims to pre-test eCall in-vehicle modules and evaluate their compatibility with the positioning services provided by Galileo and the European Geostationary Navigation Overlay Service (EGNOS) in accordance with the test procedures established by the regulation.

    As the eCall initiative goes live this month, the GSA launched a test initiative to support eCall device manufacturers in their preparation for type approval. In safety-critical situations, eCall must be as accurate as possible, so defining and conducting proper test procedures is imperative.

    Spirent is cooperating with the JRC to develop its own eCall test solution. “Working with JRC enabled us to develop better tests to verify that eCall devices are working properly,” said Steve Hickling, product director for Spirent’s positioning business.

    When a collision occurs, an eCall-equipped car automatically calls the nearest emergency centre. Even if no passenger is able to speak – such as because of injuries — a “minimum set of data” is sent, which includes the exact location of the crash site. eCall is expected to significantly reduce emergency service response times, leading to lives saved and injuries reduced.

    The JRC used a Spirent GSS9000 simulator to assess eCall devices’s capability to support the reception and processing of the Galileo and EGNOS signals. Using feedback from the JRC, Spirent has developed an eCall Test Suite for its automation solution, PT TestBench.

    Tested with various eCall devices, the eCall Test Suite is available for eCall device manufacturers and include, among others, positioning accuracy, time to first fix, GNSS receiver sensitivity and reacquisition performance.

    For more information on Spirent’s GNSS testing solutions, visit the website or download the company’s white paper Detecting and Protecting Against GPS Cyberthreats.

  • Airgain fleet management antenna features GNSS + 6 Wi-Fi ports

    Airgain Inc., a provider of advanced antenna technologies used to enable high-performance wireless networking, has released its Ultramax MIMO 9-in-1 antenna, which can receive multiple GNSS signals.

    Designed for public safety fleet management, it provides high rejection GNSS technology with coverage for multiple satellite systems including GPS, GLONASS, Galileo and BeiDou.

    The new Ultramax MIMO 9-in-1 antenna will help improve public safety and fleet solutions with enhanced Wi-Fi capability, the company said. It includes 6 x 6 MIMO Wi-Fi, dual LTE and multi-GNSS technology antennas in a single enclosure.

    MIMO — multiple input multiple output — is used within LTE to provide better signal performance and higher data rates. With integrated 6×6 Wi-Fi antennas, the antenna provides support for full high-definition (HD) streaming video and other high bandwidth applications.

    The antenna is the first in an Airgain series designed to support state-of-the-art communications technology in fleet routers, including the Cradlepoint IBR1700.

    The Ultramax MIMO 9-in-1 antenna is equipped with nine ports, supporting tri-band Wi-Fi, LTE/MIMO (including Band 14 for FirstNet) and GNSS. With a single compact footprint, the antenna avoids multiple mounting and cable entry points.

    “Technology advances in routers, including enhanced Wi-Fi and expanded MIMO LTE, require enabling antenna technology to deliver an optimized end user experience,” said Reed Pangborn, vice president of channel sales for North America. “We designed a new antenna to support the fleet management applications required in today’s evolving mobile environment. The Ultramax MIMO 9-in-1 antenna demonstrates our commitment to providing leading antenna solutions for our mobility customers covering a wide range of vehicles, including police, fire, ambulance and other fleet assets.”

    The Ultramax MIMO 9-in-1 antenna complements Cradlepoint’s IBR1700 and supports all six of its Wi-Fi ports.

    Airgain will unveil the new antenna at the Cradlepoint Global Partner Summit in Scottsdale, Arizona, April 11-12. The Ultramax MIMO 9-in-1 antenna will be available to order starting in June.

  • Becker helicopter transponder receives ADS-B certification

    Becker Avionics has received certification for the its BXT65XX Mode S transponder, designed specifically for the rigorous flying environment characteristic of helicopter operations.

    Paired with a FreeFlight Systems’ 1203C SBAS/GNSS sensor, the remote-mounted solutions provide helicopter operators a complete and cost-effective way to equip with ADS-B Out to meet the Jan. 1, 2020, mandate.

    The system is a part of the company’s robust BXT65XX line of ADS-B Mode S transponders. Manufactured with a standard ARINC 429/743 output, this transponder easily integrates with the FreeFlight Systems Model 1203C SBAS/GNSS sensor for complete ADS-B Out compliance, the company said. It can also be installed either as dual installation for primary transponder interrogations or as single install for a dedicated ADS-B transmission.

    The transponder can be installed on aircraft not equipped with a traffic collision avoidance system. Its enhanced privacy settings can disable both ADS-B and Mode S transmissions.

    “We are pleased to announce this new milestone in our transponder product line,” said Forrest Colliver, Becker Avionics’ managing director. “This new system showcases how we tailor our compact, robust, and durable avionics to our clients’ requirements in order to provide the best solution for where and how they fly.”

     

  • Launchpad: RTK receivers, autonomous driving modules

    Launchpad: RTK receivers, autonomous driving modules

    A roundup of recent products in the GNSS and inertial positioning industry from the April 2018 issue of GPS World magazine.

    OEM

    GNSS RTK Board

    For OEMs, system integrators

    The BX306Z GNSS real-time kinematic (RTK) board has powerful flexibility and compatibility to meet the needs of original equipment manufacturers (OEMs) and system integrators. The BX306Z is a cost-efficient board for positioning and raw measurement output. The board is a compact, multi-GNSS (GPS L1/L2, GLONASS G1/G2, BeiDou B1/B2) RTK module with centimeter-level accurate positioning capability. It is able to integrate with autopilots and inertial navigation units. Log and command is compatible with major GNSS boards.

    Tersus GNSS, www.tersus-gnss.com

    The Taoglas Terrablast antenna line is designed for UAVs and transportation. (Photo: Taoglas)

    Rugged antennas

    For automotive, drone markets

    Terrablast polymer-based patch antennas are 30 percent lighter than their ceramic counterparts and extremely resistant to fracture upon impact. They are designed for the automotive and unmanned aerial vehicle (UAV) markets, where impacts are possible but antenna performance cannot be compromised. The 35-mm GPS/GLONASS/BeiDou patch antenna has high efficiency of more than 70 percent across all bands, improving time to first fix. All Terrablast antennas undergo rigorous temperature, vibration and impact tests, exceed ISO 16750 standards, and are manufactured in Taoglas’ purpose-built facilities in Taiwan and the United States.

    Taoglas, www.taoglas.com

    GPS/GAGAN receiver

    Module designed for Indian market

    The S1216F8-GI2 is a NavIC + GPS/GAGAN receiver module for emerging intelligent transport systems (ITS) applications requiring NavIC/GPS capability in India. It integrates an L1/L5 RF front-end and baseband processor capable of receiving up to 14 L5 NavIC signals and up to 20 L1 GPS/GAGAN signals simultaneously. With six NavIC signals and three GAGAN signals, it offers 18–23 usable signals, providing improved accuracy in urban canyons. The S1216F8-GI2 is form-factor and pin-out compatible with 12 x 16-millimeter modules, enabling drop-in replacement. NavIC sub-frame data outputs broadcast warning messages for weather alerts and natural disasters. The S1216F8-GI2 is manufactured with ISO/TS 16949 automotive certification.

    SkyTraq Technology, www.skytraq.com.tw

    Automotive module

    To meet stringent requirements in harsh environments

    The automotive-grade MAX‑M8Q‑01A GNSS module measures 9.7 x 10.1 x 2.5 millimeters and has an operating temperature range from –40° C to 105° C. It is designed to meet the stringent requirements of the automotive market, providing superior positioning accuracy even in challenging environments such as urban canyons. Its temperature range ensures reliable performance in harsh environments, such as when mounted in a car‑roof antenna.

    u-blox, www.u-blox.com

    Multi-band receiver

    Provides safety compliance for autonomous driving

    The Teseo APP receiver enables safer autonomous driving. The multi-frequency GNSS receiver chipset is suitable for safety-critical automotive applications and high-accuracy positioning at the decimeter and centimeter levels for precise point positioning (PPP) and RTK applications. By tracking satellites of all GNSS constellations simultaneously on at least two of the frequencies used by each system, ST’s automotive-quality Teseo APP (automotive precise positioning) receiver provides high-quality raw GNSS data for PPP and RTK algorithms, which allows accurate positioning and rapid convergence time worldwide. The receiver monitors the integrity of the satellite data to alert the system if accuracy is degraded for any reason. This permits Tier-1 manufacturers to certify safety-critical systems in accordance with ISO 26262.

    STMicroelectronics, www.st.com

    SURVEY & MAPPING

    Post-processing software

    Released following intensive beta testing

    Qinertia post-processing kinematic software has been designed to help surveyors get the most of their surveys. After the mission, Qinertia gives access to offline real-time kinematic (RTK) up-to-date corrections from more than 7,000 base stations in 164 countries. By creating a virtual base station near a project, the software delivers the highest level of accuracy without having to set up a base station. An advanced tight coupling algorithm delivers high accuracy and maximizes RTK availability. Trajectory and orientation are greatly improved by processing inertial data and raw GNSS observables in forward and backward directions, especially in challenging environments. With Qinertia, surveyors can quickly identify and solve issues such as mechanical installations or sensor alignment.

    SBG Systems, www.sbg-systems.com

    Survey receiver

    Upgraded receiver offers built-in tilt compensation

    The T300 Plus GNSS receiver is designed for demanding surveying tasks, with full-constellation tracking capability, tilt compensation, 4G/Wi-Fi connection, 8-GB internal memory and an easy survey workflow with Android-based Survey Master Software. It is designed to make collecting accurate data easy and fast, whether done by a beginner or experienced professional surveyor. As an upgrade of the T300, SinoGNSS T300 Plus combines a GNSS board, Bluetooth and adjustable TX&RX UHF, Wi-Fi and 4G modem into one rugged device. Its built-in 4G modem ensures the T300 Plus works with all kinds of continuously operating reference stations (CORS) worldwide. A built-in tilt sensor supports maximum 30° pole tilt and keeps the compensation accuracy within 3 centimeters; the user can check the electronic bubble on the controller for fast surveys in the field.

    ComNav Technology, www.comnavtech.com

    TRANSPORTATION

    Marine receiver

    Atlas-capable GNSS receiver for precision 3D applications

    The Vector V1000 GNSS receiver is designed for precision marine applications, such as hydrographic and bathymetric surveys, dredging, oil platform positioning, buoys and other applications that demand the highest level 3D positioning accuracies. It provides high-accuracy heading, position, pitch, roll and heave data. The V1000 supports multi-frequency GPS, GLONASS, BeiDou, Galileo, QZSS and IRNSS (with future firmware upgrade and activation) for simultaneous satellite tracking. The receiver is powered by Hemisphere’s Athena real-time kinematic (RTK) engine and is Atlas L-band capable. Based on Hemisphere’s Eclipse Vector technology, the V1000 uses the most accurate differential corrections including RTK and Atlas L-band. It has an integrated display that can be conveniently installed near the operator. The V1000 has heading accuracy of better than 0.01 degree when using a 10-meter antenna separation.

    Hemisphere GNSS, hemispheregnss.com

    Asset connectivity

    Machine-to-machine (M2M) and internet of things (IoT) device

    The SmartOne Solar M2M/IoT device is solar-powered and offers Bluetooth Low Energy connectivity while addressing the growing global demand for reliable and affordable remote monitoring and automated data collection of assets located both within and beyond terrestrial networks. The SmartOne expands the market for remote connectivity to include assets that are otherwise difficult or expensive to reach for power replacement, and lowers the operating cost of monitoring assets being served by legacy SmartOne products. SmartOne Solar’s rechargeable batteries can deliver more than eight years of serviceable life. Without exposure to the sun, a fully charged unit can operate for many months while reporting twice daily. The product’s Bluetooth connectivity allows wireless device configuration and firmware upgrades in the field.

    Globalstar, www.globalstar.com

    UAV

    PPK drone

    Designed for large-scale surveying and mapping projects

    The WingtraOne post-processed kinematic (PPK) drone is the result of collaboration with Pix4D and Septentrio. It is able to deliver orthomosaic maps and 3D models with an absolute accuracy down to 1 centimeter (cm), offering broad coverage and high resolution with ultra-precise accuracy. The WingtraOne can cover 130 hectares (320 acres), equivalent to 240 football fields, in a one-hour flight, and deliver maps at ground sample distances below 1 cm/pixel. Vertical take-off and landing (VTOL) offers hands-free operation and a smoother ride for onboard sensors as well as greater coverage than comparable multi-rotor UAVs. PPK computes ultra-precise geolocations for each image by combining the GNSS data with correction data from a nearby reference receiver. It offers a root-mean-square (RMS) error of 1.3-cm horizontally and 2.3-cm vertically without any ground control points.

    Wingtra, www.wingtra.com

    Counter-UAV aircraft

    Radar used to mitigate threats

    DroneHunter is a fully autonomous UAS airspace defense solution. The intelligent robotic aircraft is enabled with TrueView radar designed and engineered for physical remediation of intruder or threatening drones. DroneHunter is an autonomous UAS perimeter detection and protection solution designed to quickly detect, classify and secure against drones and other UAS. When an intruder drone is discovered, DroneHunter can engage autonomously via artificial intelligence (AI)-directed detection, tracking and guidance. Once the rogue drone is identified and the threat level analyzed, DroneHunter safely remediates the threat day or night, at a safe stand-off distance, with no collateral damage. DroneHunter supports multiple drone platforms based on use-case requirements.

    Fortem Technologies, fortemtech.com

  • Research Online: Positioning integrity parameters for vehicle safety

    By Yanming Feng and Charles Wang,
    Queensland University of Technology, Australia,
    and
    Charles Karl, Australia Road Research Board, Australia /
    Presented at ION International Technical Meeting 2018

    Connected vehicle safety and traffic applications depend on communication, position and velocity information to function. However, road users may have different vehicle communicating and positioning capabilities. Further, the performance of communicating and positioning could vary from time to time and location to location.

    The vehicle safety system must be fully aware of the performance of vehicle positioning outputs and warn drivers when the positioning system cannot be used for the intended level of safety applications. Minimum operational performance standards about positioning have not been established in the road community.

    This paper reviews and develops the required navigation performance parameters for vehicle positioning capability in terms of accuracy, integrity, timeliness and interrogability of positioning solutions.

    It attempts to adjust the integrity performance parameters for vehicle safety positioning and provide the analysis for integrity risk, protection level and different alert limits. It introduces the error ellipse representation to visualize the protection bubble area of each vehicle on the road. Experimental results demonstrate how different capability levels meet different integrity alarm limits.

    Available online via www.ion.org/publications/browse.cfm.