Journey into ROVR - TarantulaX-what, how and why

Hello everyone, welcome to the second article in our Journey into ROVR series—TarantulaX: What, How, and Why. This article will provide a detailed overview of everything related to ROVR's first hardware product, TarantulaX.

First, what is TarantulaX ?

TarantulaX is ROVR's first hardware product, featuring a dual-band GNSS antenna that supports RTK services, an encryption chip, and low-power Bluetooth connectivity to smartphones.

Here are the specifications for TarantulaX:

Dual-band GNSS (L1/L5)
NMEA and RTCM3.2 Messages (RTK/PPK)
Positioning Accuracy: ±2 cm
Hardware Data Encryption for Blockchain Applications
Dual Patch Antenna: 35mm x 35mm + 25mm x 25mm
6-axis IMU
Waterproofing Grade: IP67

TarantulaX utilizes the globally covered RTK network from GEODNET to provide centimeter-level positioning information (special thanks to our strategic partner GEODNET, which offers three months of free RTK service with each TarantulaX device). Each TarantulaX device generates a unique encryption key through its encryption chip, signing and encrypting all data transmitted by the device.

On the mobile side, we offer the ROVR Network App. Users do not need to install an additional dashcam; instead, they can leverage the phone's camera and computational capabilities, combined with TarantulaX's centimeter-level positioning, to achieve real-time mapping. The ROVR Network App is compatible with both Android and iOS systems:

iOS version: Available for download by searching for ROVR Network in the Apple Store.
Android version: Due to software architecture upgrades, it is currently still in public testing. You can obtain the APK through the download link on the ROVR official website or in our Discord.

Second, how dose TarantulaX work ?

Part 1, Workflow:

Many community members, partners, and potential investors are curious about how TarantulaX works. Below, we will provide a detailed explanation of the operation and data flow of TarantulaX.

The data flow diagram for TarantulaX and the ROVR Network App is as follows:

In the diagram, each box represents an independent hardware or functional module, while each line represents a data flow (distinguished by different colors to indicate the communication methods). The text on each line indicates the content of the transmitted data.

Signed GGA(before corrections)
1. TarantulaX receives satellite-broadcasted positioning information and calculates the output as GGA messages. These messages are signed and encrypted using the built-in encryption chip of TarantulaX and then transmitted via Bluetooth to the ROVR Network App.
2. what is GGA?
  GGA(Global Positioning System Fix Data) is a data format from NMEA (National Marine Electronics Association) used primarily for providing GPS positioning information. It includes details like UTC time, latitude, longitude, fix status, number of satellites used, horizontal dilution of precision (HDOP), altitude, and geoid height.
  $GNGGA,123456.00,3751.65,S,14507.36,E,1,12,0.7,10.0,M,46.9,M,,*47
  - Data Type Identifier: $GNGGA
  - UTC Time: 123456.00 (12:34:56)
  - Latitude: 3751.65,S (37 degrees 51.65 minutes, South)
  - Longitude: 14507.36,E (145 degrees 07.36 minutes, East)
  - Fix Status: 1 (1 indicates a valid fix)
  - Number of Satellites: 12 (Number of satellites being used)
  - Horizontal Dilution of Precision (HDOP): 0.7
  - Altitude: 10.0,M (10.0 meters)
  - Height of the Geoid: 46.9,M (46.9 meters)
  - Checksum: ,,*47
GGA before corrections
1. The data content is identical to step 1. The ROVR Network App sends the uncorrected GGA messages via the internet to GEODNET's NTRIP server to request real-time correction data for the corresponding region.
RTCM
1. After GEODNET's NTRIP server receives the uncorrected GGA message sent by the ROVR Network App, it returns the corresponding RTCM messages.
2. what is RTCM?
  RTCM(Radio Technical Commission for Maritime Services) is a standard format used for differential positioning, primarily in real-time kinematic (RTK) systems. It provides correction information to improve the accuracy of GNSS positioning by transmitting data such as pseudorange corrections, carrier phase corrections, and satellite status.
  $RTCM,0,1,1,1234,500.0,1000.0,1.0,2.0,3.0,4.0*6A
  - Data Type Identifier: $RTCM
  - Message Type: 0 (0 indicates RTCM 2.x version)
  - Message Sequence Number: 1
  - Message Check Digit: 1
  - Satellite ID: 1234
  - Pseudorange Correction: 500.0 (in meters)
  - Carrier Phase Correction: 1000.0 (in meters)
  - Other Correction Parameters: 1.0,2.0,3.0,4.0 (e.g., frequency, signal strength)
  - Checksum: 6A
RTCM
1. The data content is identical to step 3. The ROVR Network App sends the RTCM messages to TarantulaX via Bluetooth. TarantulaX then uses the RTCM messages received from the ROVR Network App, along with the uncorrected GGA messages received from the satellites, to apply corrections and obtain more precise positioning information.
GGA(after corrections)
1. The GGA message can be corrected using real-time or non-real-time correction data to improve the accuracy of the positioning information. The corrected GGA message will differ from the uncorrected version in several key aspects:
  1. Positioning accuracy: The corrected GGA message typically shows a higher HDOP value, indicating improved positioning accuracy.
  2. Altitude information: If RTK services are used, the corrected GGA message may provide more accurate altitude data.
  3. Number of satellites: The corrected GGA message may utilize more satellites, which helps enhance the accuracy of the positioning.
Video Stream and Intrinsic&Distortion parameters
1. The ROVR Network App accesses the phone's camera to capture real-time video streams and extract images for 2D detection, which includes mapping features such as lane lines, lane boundaries, traffic signs, traffic lights, cones, and also detects pedestrians and vehicles to assess for potential obstructions.
2. While acquiring the video stream, we also read the camera's intrinsic and distortion parameter information from the phone.
3. What is Intrinsic Parameters?
  Camera Intrinsic Parameters describe the optical characteristics and imaging geometry of a camera, primarily used to convert 3D world coordinates into 2D image coordinates. The intrinsic matrix typically includes focal lengths and the coordinates of the principal point. The focal lengths determine the magnification of the image, while the principal point represents the center of the image.
  # Camera Intrinsics Example K = [ [f_x, 0, c_x], # First row: focal lengths and principal point coordinates [0, f_y, c_y], # Second row: focal lengths and principal point coordinates [0, 0, 1] # Third row: homogeneous coordinates ]
  - f_x: Focal length in the x-direction (in pixels)
  - f_y: Focal length in the y-direction (in pixels)
  - c_x: x-coordinate of the principal point (in pixels)
  - c_y: y-coordinate of the principal point (in pixels)
4. What is Distortion Parameters?
  Distortion parameters describe the image distortion caused by the physical characteristics of the camera lens. They mainly include radial distortion and tangential distortion. Radial distortion causes straight lines to appear curved at the edges of the image, while tangential distortion arises from misalignment between the lens and the sensor. These parameters can be used to correct images, making them more representative of the actual scene.
  # Distortion Parameters Example distortion_params = { 'k1': k1, # Radial distortion coefficient 1 'k2': k2, # Radial distortion coefficient 2 'k3': k3, # Radial distortion coefficient 3 'p1': p1, # Tangential distortion coefficient 1 'p2': p2 # Tangential distortion coefficient 2 }
  - k1: Radial distortion coefficient 1 (affects straight lines at the image edges)
  - k2: Radial distortion coefficient 2 (affects straight lines at the image edges)
  - k3: Radial distortion coefficient 3 (affects straight lines at the image edges)
  - p1: Tangential distortion coefficient 1 (caused by misalignment of lens and sensor)
  - p2: Tangential distortion coefficient 2 (caused by misalignment of lens and sensor)

2D Detecion Result
1. By utilizing the acquired video stream and the camera's distortion parameters, we perform a de-distortion operation on the camera. This process aims to achieve higher precision in the 2D detection results (image coordinate system).
2. what is 2D detection result?

{
  "detections": [
    {
      "id": 1,
      "type": "lane_line",
      "polyline": [
        {"x": 100, "y": 250},
        {"x": 150, "y": 200},
        {"x": 200, "y": 180},
        {"x": 250, "y": 160},
        ...
      ],
      "attributes": {
        "confidence": 0.98,
        "category": "white_dash"
      }
    },
    {
      "id": 2,
      "type": "traffic_sign",
      "bounding_box": {
        "top_left": {"x": 300, "y": 400},
        "bottom_right": {"x": 350, "y": 450}
      },
      "attributes": {
        "confidence": 0.95,
        "category": "stop"
      }
    },
    {
      "id": 3,
      "type": "lane_line",
      "polyline": [
        {"x": 400, "y": 500},
        {"x": 450, "y": 460},
        {"x": 500, "y": 440},
        ...
      ],
      "attributes": {
        "confidence": 0.97,
        "category": "yellow_dot"
      }
    }
  ]
}

Here is our detection results video:

Local Mapping and GGA after corrections @ 30s
1. After the 2D detection, we construct stereo image pairs with a baseline of approximately 3 meters using consecutive frames to perform 2D-to-3D conversion, enabling local mapping. Every 30 seconds, we encrypt the results of the local mapping along with the corresponding corrected GGA statements and send them back to the ROVR Backend.
  1. Street view images are not ROVR's target data product; ROVR only provides processed map data to customers. ROVR will not return images in any form and will not sell images externally.
  2. To comply with the legal requirements of different countries, data from various nations and regions will be stored on servers located in those countries within the ROVR Backend.
2. what is local mapping data?
  1. Local Mapping data is a type of map data stored in a local coordinate system. Compared to a true HD Map, Local Mapping data does not utilize a world coordinate system and has lower accuracy. The ROVR Backend aggregates and corrects multiple instances of Local Mapping Data for the same route, creating HD Map data that meets accuracy requirements (absolute precision of 50 cm and relative precision of 20 cm).
  { "type": "FeatureCollection", "crs": { "type": "name", "properties": { "name": "urn:ogc:def:crs:OGC:1.3:CRS84" } }, "features": [ { "type": "Feature", "properties": { "color": "WHITE", "type": "LANE_MARK_DASH" }, "geometry": { "type": "LineString", "coordinates": [ [139.679376803516902, 35.278183263389849, 40.226988258117977], [139.679334626060893, 35.278211092616715, 40.233198178000748], [139.679291870153975, 35.278239407648982, 40.24589875061065], ... [139.676890385264528, 35.280014796428532, 40.821298616938293], [139.676854620429367, 35.280049015147227, 40.829007665626705], [139.676829537519211, 35.2800729331978, 40.835873213596642] ] } }, ... ] }
3. Here is a sample image of the Local Mapping Data:
Download HD Map Sample Data by TarantulaX

Part 2, TarantulaX Installation:

Where to install TarantulaX:
1. We recommend placing the TarantulaX on the roof of the vehicle, as close as possible to directly above the phone's position. Since TarantulaX comes with a GNSS antenna, it is essential to ensure there are no metallic obstructions within a zenith angle of 75°.
2. Provided the same GNSS receiving antenna and integrated navigation software are used, placing the GNSS antenna on the roof of the vehicle will yield better positioning results than placing it inside the vehicle. This is due to the principles of satellite positioning—satellite positioning is based on carrier wave communication for differential measurements, and the vehicle body can interfere with the carrier signal strength, sometimes completely blocking it. Currently, all mass-produced vehicles integrate their positioning antennas on the roof.

How to install your phone:
1. In the recording interface of the ROVR Network App, we provide guidance for aligning the horizon and forward direction. Please adjust the phone screen accordingly.

Third, why is the TarantulaX designed this way ?

Why is an additional positioning antenna and RTK service needed instead of directly using the phone's built-in positioning antenna?
1. It is difficult to keep a phone placed outside the vehicle for an extended period, and placing it inside the vehicle can affect GNSS signal strength, leading to reduced positioning accuracy.
2. Even if the phone can be placed outside, standalone GNSS positioning accuracy typically ranges from several meters to tens of meters, which is insufficient for constructing an HD Map. In contrast, RTK services can provide centimeter-level positioning accuracy, which is crucial for building HD Maps.
Why is an encryption chip needed?
1. The built-in encryption chip in TarantulaX generates a unique public key and encrypts all data from the device. This ensures the security and integrity of the data during transmission.
2. We firmly believe that anti-cheating measures are essential for DePIN projects. This guarantees the availability of data and fairness for all users.
Why communicate via Bluetooth?
1. Using Bluetooth allows support for both iOS and Android devices, enabling a wider range of users to participate in the project.

In conclusion

Thank you very much for taking the time to read our article.

Since we officially started collecting data at 00:00 GMT on Monday, September 2, 2024, our testing users have accumulated over 365k km of data. We greatly appreciate everyone's help.

Meanwhile, the TarantulaX device is now publicly available, and we hope more friends can join us.

We look forward to your participation in accelerating ROVR's growth and achieving greatness together.

🔸 ROVR Network (***https://rovr.network/***)

💬 X: ROVR (***https://x.com/ROVR71776***)

🗯 Discord (***https://discord.com/invite/RjV3E3u4F2***)

Thank you!

ROVR Team