Rough Terrain Rover

Odrive S1s are open source, well-documented, ROS compatible, and relatively cheap ($150 each)

• Their screw terminals are also a lot easier to interface with than the old Maxon controllers, removing the need for a dedicated controller harnessing board

• Had a GUI that allowed us to do proof-of-concept testing very quickly

  • Controlling them remotely was a whole different beast -- faced hardware issues with the CAN (like can0 and can1 switching on our Pi at random)

  • Switched from using ROS service calls (from example on documentation) to ROS2 control

Mechanical team massively expanded our chassis, making room for electronics to slide in and out on a green acrylic board

Rover drove well, was very dextrous on joystick control, and manipulator motors were able to be controlled off Odrives too in preliminary testing

Got compliments from other CIRC 2024 teams who'd heard of Odrives and were thinking of implementing them themselves

We are looking into wago connectors, bullet connectors, etc to clean up our harnessing further

Might switch to ODrive Micros in the future given our low power consumption and need to conserve space

First attempt was to just make minor changes to last year's report. This did not fly AT ALL -- I had no idea just how much more strict the rules were for officially competition teams

• CIRC also cracked down on safety because alledgedly one of the team's rovers blew up on the way to competition

Had to replace a lot of our off-the-shelf power converters and busbars with ones that had datasheets

Used fuses to limit the current going through each wire below the max safe continuous current-carrying capacity of each wire

• Our previous approach was to have fuses allow each circuit to get as close to the circuit's theoretical  max continuous current. 

  • This ended up having our total system max be around 80A -- which was extremely difficult to size wires and connectors for.

  • In practice, our drive motors (which account for the lion's share of the current) only drew a few amps. So we were (safely) overestimating the sizes of our fuses anyway

• Solution: added fuses on the joining points between multiple circuits that were smaller than the sum of the fuses on the subcircuits.

  • We weren't going to let all subcircuits be drawing max power at once, but we would still allow each subcircuit to have current spikes

  • New system maximum 30-40A (In practice, on the field drew about 10A max)

• Used 8-14 gauge THHN Cerrowire for our connections.

Rover passed inspection on the second day (first day spent fixing up rover from the drive up to Canada) and offically competed safely for the first time

Omindirectional Ubiquit antenna broadcasted Wifi from our router to a 120 degree sector Ubiquiti on our rover

• In-rover Raspberry pi was ethernet-wired to an access point on the rover antenna

• Antennas set to 2.4GHz with 40 MHz channel width

• Base station antenna mounted within line of sight of rover antenna, hoisted as high as possible (on a post outside)

Terrible connection testing inside and outside on Caltech campus, kept losing connection and getting kicked out of SSH

• Moving antenna farther away from rover helped a little (signal stregth in the -20 dbi range was too strong)

• Solution (from software team): use tmux, which is supposed to help you recoved entire terminals and keep running scripts even if ssh crashes

Through our RealSense camera video feed, the rover operators were able to remotely control the rover with no communication loss during the competition.

In the future, we want to be able to send ROS commands over RF as well. Whether the Ubiquitis work or not almost seems to be a matter of blind luck, and many other CIRC teams had trouble with their Ubiquiti systems too, so it's good to have a backup system.