BLUEsat’s ERC Ventures and Adventures

BLUEsat’s ERC Ventures and Adventures

The NUMBAT Rover arm picking up a tool against a background of Red "Mars" soil at the European Rover Challenge.

Our trip to Poland was one hell of an adventure, with a lot of on-the-spot engineering, things breaking and being fixed (or “fixed” in some cases), panicked runs carrying the rover and our groundstation gear to the European Rover Challenge (ERC) competition grounds and much more. After over 24 hours of travel, of two planes and a scenic car ride thanks to accidental detours, we arrived at our 10-day home in Starachowice, Poland. On that same day, we reassembled the rover, excited for the days ahead. The next few days were filled with a lot of testing of subsystems, calibrating of optical flow cameras, 3D-printing, quite a bit of soldering, and last-minute modifications and additions to our design. Some of our testing even went overtime and one night we spent some time with 3 phone lights following the rover as we practiced picking up sand and placing it into a small container on the rover.

Night time rover training
Night time rover training, picking up soil in the dark

ERC Day 1

Navigation Task

Our first task in this year’s ERC was the Navigation task. Here, the rover had to try to reach certain checkpoints without looking at camera feed. Unfortunately, almost as soon as we put down NUMBAT (our rover) on the Mars yard, our radio signal kept disconnecting due to interference and being under-powered. This left the rover continuously following the last command it received, mostly just driving straight and digging the wheels deeper into the red dirt for the duration of the task. The organisers of the ERC were really awesome and let us complete the task again later in the day, after letting us increase the power of our signal and fixing that issue.

This time, the rover drove well and made it within 16cm of the first checkpoint!!! Unfortunately, as we noticed later in the task, our Lidar was accidentally switched off on the USB hub we were using, so the driver, Simon Ireland, had to rely on our black and white rendering of the map,  where shade variation depended on height, as well as distance estimations from our 4 optical flow cameras (also luck and intuition) for navigation. Because of that, time constraints and a massive hill the rover had to get over to reach the second checkpoint, we stopped there, still achieving one of the best results for that day for navigation!

Science Task

The other task we completed on the first day was the science task, in which thin silver rods marked 3 locations where we had to collect soil samples from, store them on board of the rover and try to find information about them; for example, moisture, temperature, density and whatever else your heart desires. The first issue we encountered was that the rover seemed to really struggle with driving along the track. About 10 minutes into the task, our driver, Thomas Renneberg, noticed that one of the wheels was 180 degrees out of alignment (it was driving backwards when we tried moving forward), leading to a quick fix and much smoother driving afterwards. The next problem was that it was very difficult to see the thin silver markers to find the locations of the soil, mostly because the camera quality of our main camera prevented us from seeing the thin markers that camouflaged in with the red sand. However, by changing to a different camera, we were able to locate one of the markers! By that point, our time was running out, so after a very quick scoop of soil was taken, we put it on-board the rover in a small sealed container in the last minute or two.

ERC Day 2

On the second day at the ERC, we had to complete the Maintenance task. This involved the pushing and turning of dials and switches, opening a compartment with more switches and plugging in some power plugs. One part of this task was measuring the voltage across two terminals of a power plug. Because this was implemented the night before the task, it hadn’t been tested with the arm and wheels in operation until about half an hour before we were meant to begin. We’re still not 100% sure exactly what happened but it seems like the voltmeter publishing across the whole network screwed it all up and stopped the arm and wheels from working. This led to us quickly try to ignore the rising panic and  re-flash both steering modules and the arm module, essentially resetting the code back to its previous state and getting the rover operational again with just minutes to go before the task.

NUMBAT at the maintenance task

Maintenance Task

During the task, we had some issues trying to crab steer (where all 4 wheels moved in one direction, essentially trying to translate the rover to a point instead of turning normally) with some of the treads on the wheels coming off. This didn’t slow us down too much though, since we had our trusty four-wheel drive system too, which got the rover close enough to the maintenance panel to complete the majority of the task.  Despite some of the smaller switches being hard to grasp, we managed to turn and push several dials and switches but couldn’t reach one of the switches that was just a little too high up. As the rover started dancing in circles with the arm fully retracted to a degree beyond what I had ever seen before, we realised that we lost our connection with our groundstation team again because of a mechanical connection and had to stop the task there.

ERC Day 3

Retrieval Task

On the last day of the ERC, we had to do the retrieval task. In this task, 3 green tool-shaped objects were placed, partially hidden, on the Mars ground and we had to find them, pick them up, store them vertically on the rover and try to bring them back to the retrieval location. We made a last minute modification to the claw system, using two semi circles that meet up together instead of the normal grip. 3D-printed funnels with paper taped to them to help guide the retrieval objects into the holes were also added in Poland (did I mention how glad I was that we took the 3D printer with us?). This task went well, with the first two retrieval objects picked up and stored on the rover with little setback. The third one was more difficult to find and we first mistook a white rock for the object, but after a minute of trying to pick it up we saw a flash of green on the screen! We were able to get close to the third object but couldn’t pick it up as one of the acrylic panels used for linkages on the arm broke, preventing the claw from closing all the way, but still, an amazing result overall!

Rover at the ERC putting tool in place
NUMBAT at the Retrieval task, with the last minute modifications

We were all pleasantly surprised and ecstatic to have come 8th out of 35 competing teams, especially as it was a better result than our 9th and 15th place in previous years! It was also our first time competing with NUMBAT; a rover we started building from scratch in 2017. We spent our last two days in Poland disassembling the rover, packing bags and exploring some castles and gardens before heading back home to Sydney to try catch up on uni work and lost sleep.