Trustworthy Positioning for Next Generation Intelligent Transport Systems Ahmed El-Mowafy

Transcription

1 Trustworthy Positioning for Next Generation Intelligent Transport Systems Ahmed El-Mowafy

2 Contents Background on ITS and C-ITS Requirements Challenges RAIM Test and Results Utilisation Workshop, Sydney, 6 Feb

3 ITS objectives Make transportation safer, more efficient and reduce emissions Safety Efficiency Emissions Utilisation Workshop, Sydney, 6 Feb

4 C-ITS V2V and V2I Source: Austroads, 2010 Utilisation Workshop, Sydney, 6 Feb

5 C-ITS Communication using DSRC. DSRC-based range-rating measurements can enable GNSS/DSRC cooperative positioning. VANET (Vehicular ad hoc network) Source: Internet Utilisation Workshop, Sydney, 6 Feb

6 Satellite positioning accuracy requirements Road level (few m) Lane-level (< 1m) More accuracy is needed Where-in-lane level (sub-m) Current systems mainly use SPS (Standard Positioning Service). SPS gives 1-5 m accuracy - not suitable for lane-level precision. Use of augmentation techniques, such as SBAS has the potential to offer the required accuracy. Most imported C-ITS uses SBAS technology. Utilisation Workshop, Sydney, 6 Feb

7 But: Is it only accuracy we are interested in! Utilisation Workshop, Sydney, 6 Feb

8 Ex: Curtin University Driverless Buss Utilisation Workshop, Sydney, 6 Feb

9 Accuracy VS Integrity Alert to driver/user PL AL Accuracy Integrity Utilisation Workshop, Sydney, 6 Feb

10 Challenges Standards? need to be set based on performance requirements. Complete map of risks (e.g. collision types, faults, etc.) and vulnerabilities (system errors, interference, jamming spoofing, etc.) and identify their probabilities. Integration of sensors (GNSS is a main component but not alone) Cost Communications Application dependence. Technology. Utilisation Workshop, Sydney, 6 Feb

11 Ex: Vulnerabilities due to the work environment Urban environment: Loss of lock Heavy multipath Non Line of Sight (NLOS) Frequent cycle-slips Example of NLOS Mitigations: Multipath mitigation at the antenna 3D city-models ray-tracing algorithms SNR monitoring Non-Gaussian error models VANET CIM concept Use 3D citymodels Utilisation Workshop, Sydney, 6 Feb

12 Characterisation of errors (ex: clock corrections) Histograms Q-Q CDF PRN16 PRN 29 PRN30 Utilisation Workshop, Sydney, 6 Feb

13 Accuracy GNSS Positioning Methods 1m 80cm 60cm DGNSS SBAS 40cm 20cm 10cm 4cm 10km RTK PPP 100km 1,000km 10,000km Baseline length Worldwide Utilisation Workshop, Sydney, 6 Feb

14 SBAS Improved positioning: sub-m (DGPS L1, L1/L5) deci-metre (PPP) Integrity monitoring: error bounds for PL Vulnerabilities linked to hardware, software and data link with the satellites GEO Satellites SBAS Ionosphere Troposphere SBAS Reference stations Utilisation Workshop, Sydney, 6 Feb

15 Integrity monitoring Advanced RAIM (ARAIM) Fault Detection & Exclusion based on statistical hypothesis testing PLs computation based on estimated impact of faults on position solution Determine Protection Levels (PLs) as safety bounds to positioning errors Take into account risk of anomalies/faults PLs must be smaller than the Alert Limits (ALs) to guarantee availability Utilisation Workshop, Sydney, 6 Feb

16 ARAIM For standalone vehicle Multi-sensor V2V and V2I. AT CT Multi-sensor integration Utilisation Workshop, Sydney, 6 Feb

17 Testing Ex. Kinematic test in Tokyo (with TUMSAT) Trimble RTK (10Hz) GPS, GLONASS and BeiDou a Bosch-consumer grade MEMS IMU The heading error of this IMU ranged from -2 o to 5 o, can accumulate to 10 o after 30 min if left uncalibrated. Speed sensor: s = 5 cm/s GNSS-Doppler: s = 10 cm/s. Reference : PPK & POS/LV Utilisation Workshop, Sydney, 6 Feb

18 Testing: challenging environment Utilisation Workshop, Sydney, 6 Feb

19 Integrity prediction Identify critical locations on the map, at different times of the day Integrity unavailable: red points (PL>AL) Along Track Cross Track Utilisation Workshop, Sydney, 6 Feb

20 Actual data: Flow chart of sensor fusion (RTK, IMU, odometer) 1 RTK available Yes Computes positions using RTK No Doppler available Yes 2 3 Meets GNSS Velocity Conditions No No Meets GNSS heading Conditions Yes Computes heading from GNSS Doppler No Yes Compute velocity from GNSS Doppler Compute positions from GNSS velocity Update heading of the IMU (GNSS & ZUPT) Computes positions from Speed of Speed sensor and IMU heading Output position Utilisation Workshop, Sydney, 6 Feb

21 Protection levels RTK PL AT,i = K fa, i σ δat,i + K md,i σ AT,i + (cos θ a 1 T S i b o ) 2 + (sin θ a 2 T S i b o ) 2 PL CT,i = K fa, i σ δct,i + K md,i σ CT,i + (sin θ a 1 T S i b o ) 2 + (cos θ a 2 T S i b o ) 2 IMU+odometre PL AT,i = K md,i σ AT,i + (cos θ a 1 T S b θimu b v ) 2 + (sin θ a 2 T S b θimu b v ) 2 biases PL CT,i = K md,i σ CT,i + (sin θ a 1 T S b θimu b v ) 2 + (cos θ a 2 T S b θimu b v ) 2 Doppler PL AT,i = K md,i σ AT,i + (cos θ a 1 T S b ve b vn ) 2 + (sin θ a 2 T S b ve b vn ) 2 PL CT,i = K md,i σ CT,i + (sin θ a 1 T S b ve b vn ) 2 + (cos θ a 2 T S b ve b vn ) 2 Utilisation Workshop, Sydney, 6 Feb

22 RTK Results * b = G+R+B AT CT G+R G+B G Utilisation Workshop, Sydney, 6 Feb

23 CT Combined * b = PL_AT (All) Time(Doppler) (sec) PL_AT Time (sec) Utilisation Workshop, Sydney, 6 Feb err_ct (RTK) err_at (IMU+odo) Time (sec) Time (sec) PL_CT (RTK) err_at (Doppler) PL_AT (IMU+odo) err_ct (All) PL & Error (m) PL & Error (m) PL_CT (All) 0.5 PL & Error (m) PL & Error (m) PL & Error (m) err_at (RTK) 0.3 PL & Error (m) IMU+odometer Doppler PL_AT (RTK) Time (sec) 0.5 RTK err_at (All) PL & Error (m) All sensors PL & Error (m) AT Time (sec) PL_CT (Doppler) err_ct (Doppler) PL_CT (IMU+odo) Time (sec) err_ct (IMU+odo) Time (sec) 23

24 Summary ITS / C-ITS might be the norm in the near future. Real-time safety related applications in ITS/C-ITS require highly trustworthy positioning: i.e. integrity monitoring. The technology might not be the problem: cost and interoperability might be. Integrity Monitoring (IM) is challenging IM can be achieved, but which standards? Applications? Utilisation Workshop, Sydney, 6 Feb

25 Thank you Questions Ahmed El-Mowafy Spatial Sciences 25