Global heading sensor fusion

[Supernova1114]

We need a source for global (world-space, absolute) heading for the rover. This will allow us to orient and move to GPS coordinates.

Using 2 GPS on the rover to create a vector and therefore a direction. Then we can orient the rover to move to the target waypoint.

For a differential GPS implementation. We will also need a GPS module at the base station.

Another option is to use a magnetometer, though these can be very sensitive to electrical noise.

[Supernova1114]

8/31 - Tested our own implementation or differential GPS using 2 GPS on a meter stick. Angle of vector was very inaccurate and would some times not even change. We tested the 2 GPS much farther apart. With this we got better results, but still not accurate enough.

Note that one of the GPS was missing an antenna.

Would like to purchase another antenna just in case. Would like to also possibly purchase more accurate GPS modules.

[Supernova1114]

Some thoughts:

• We don't necessarily need the GPS vector angle to be accurate constantly. We just need it to be initially accurate to find out global angle, then can use the NavX (IMU and Gyro) module to act as a relative heading. We would indeed still want the GPS coordinates in general to be relatively accurate so to actually reach the target waypoint.

• Another option is to use a magnetometer for absolute heading. We were having trouble last year with the NavX where small electrical noise would mess up the output angle of heading.

• Note that the Zed2i stereo camera also has a magnetometer and may actually be a better option for heading.

[Supernova1114]

Update:

• Did some more research. Differential GNSS does indeed use a GPS module at the base station.

[Supernova1114]

Copy from Discord:

Ok so here's what I found so far that stays within budget:

• mosaic-go heading GNSS module evaluation kit ($804): https://shop.septentrio.com/en/shop/mosaic-go-heading-gnss-module-evaluation-kit
• CUAV New C-RTK 2HP Dual Antenna Centimeter Position GNSS Heading Module ($267): https://www.aliexpress.us/item/3256805754135938.html?spm=a2g0o.detail.1000014.2.d13172f4hJ2kUq&gps-id=pcDetailBottomMoreOtherSeller&scm=1007.40000.326746.0&scm_id=1007.40000.326746.0&scm-url=1007.40000.326746.0&pvid=a5f2f806-ec02-4777-895a-7570a92e0479&_t=gps-id:pcDetailBottomMoreOtherSeller,scm-url:1007.40000.326746.0,pvid:a5f2f806-ec02-4777-895a-7570a92e0479,tpp_buckets:668%232846%238114%23782&pdp_npi=4%40dis%21USD%21535.00%21267.5%21%21%21535.00%21%21%402103244816937027992932300e1390%2112000034945153820%21rec%21US%21%21A

Some notes:

• The mosaic-go module does not come with antennas nor antenna wires.
• The C-RTK 2HP module looks like it has everything we need. Supports CAN Bus, comes with all wires and antennas.
• Cuav website: https://www.cuav.net/c-rtk-2hp/

[Supernova1114]

Update regarding this:

• Recently got a visualizer working for directional vector of the two GPS
• Thinking of just buying some U-Blox GPS and better antenna to see if we get better results with closer GPS.
• Some of the U-Blox GPS have support for RTK GNSS.

[Supernova1114]

Created PO

• Not getting RTK GPS, too expensive SaddlebackCollegeRobotics/.github#17

[Supernova1114]

We ended up also buying 2 of the RTK GNSS modules.

Also members are currently working on SaddlebackCollegeRobotics/.github#84

[Supernova1114]

RTK Modules arrived. Members currently working on getting RTK up and running. #129 Will have to eventually choose between position precision or orientation precision. Position precision may be better because orientation is able to be given via gyro sensors and dead reckoning.

[Supernova1114]

Doing some research. best case scenario would be to have two RTK GNSS modules onboard the rover, and 1 at base station. Would give both precise orientation, and precise positioning.

https://navigation.ros.org/tutorials/docs/navigation2_with_gps.html

[Supernova1114]

Will be using dual GNSS on rover for global heading. One GNSS module at base station for RTK.