The task required from the ROV was to collect a range of benthic species from multiple sites in polar conditions at a depth of 20-30m. The samples were quite fragile, and needed to be collected without damage, and were all approx. 7cm x 7cm. See this post for the planning phase, but the issue faced was that the stock jaws on the gripper were not overly dextrous at their tips due to the overlapping design. We trialled 3D printed alternatives (PLA) and found that all of the force applied when gripping objects was transferred to the point where the jaws meet the gripper shaft, and printing this area with strength was difficult to do with the printer available.
Our solution was to machine a metal “backbone”, which a range of different gripper “faces” could mount on to. Our design also used a printed sleeve at the bolt joins to lessen friction effects. We had great success with our flat faced jaws, but opted to use our “cup” design for the majority of our sample collection. Designing faces applicable for the users task is made incredibly easy by the open source CAD drawings provided by Blue Robotics, and is accessible to anyone looking to do something similar.
We were able to average about one sample per minute from the seafloor, the gripper was responsive, rugged, and functioned incredibly well! *sampling ease was increased a lot by button mapping – see controls section below.
The camera on the BlueROV has always been a limiting factor for me – and compared to the other ROV in the water which had 4k capability there was worlds between any opportunistic “cinematic” footage captured – the BlueROV camera was far less useful for photogrammetry work as well, due to the obvious resolution limitations.
I want to note however that the low-light capabilities of the camera blew me away – where attached Gopros filmed almost blackness, the BlueROV camera had perfect visibility (lacking colour) – this is without underwater lights due to flood. The camera functioned perfectly as a tool alongside the gripper for spotting samples, securing samples, and getting them back to our crate.
The batteries on average gave me about 1.5-2 hours runtime in -1.9 degrees Celsius. I used 3-4 batteries per day, running 30% Gain and constantly operating the gripper. Batteries were stored topside in a pelican case, our ambient temperature was -20 C and had no issues with them – I was impressed with the bottom time offered!
BlueROV Heavy platform:
We run the Heavy configuration on our ROV – This provided an incredibly stable platform to work from. I would tend to operate the unit in Stabilize mode (which I find much more stable than depth hold whilst pitched), I would pitch down about 30 degrees to pick up a sample, then level and drive to the crate for drop off. Other ROV operators were very impressed with how flat and solid the unit would sit in the water. I was able to collect some fist sized rocks with no problems, which speak to the lift capabilities of the unit.
Anyone operating in sub-zero conditions can take note that I would often store the unit in the water turned on, rather than on the ice as it was warmer (-2 compared to -20 C). I only had one situation with frozen props on a battery change, which freed themselves in short order once in the water. Each time I lifted the unit out of the water I would tilt it around to drain water from the props, and this worked fine. I had no issues with condensation and did nothing other than putting new desiccant bags in once prior to the trip. The internal computer ran at around 28 degrees C, and was not slowed or affected by the freezing conditions at all.
I use an Xbox elite controller for my system. At purchase, I had really liked the idea of a peli-control box with joysticks ect, but this deployment has completely changed my mind. The flexibility offered by the cable to walk to a dive hole and operate by sight, plus flexibility to operate in a range of different habitats (tables, tents, vehicles) really shined. Other users may be interested to know that utilising the Microsoft button mapping capabilities alongside the QGroundControl button mapping capabilities can be handy: Ie I used the Microsoft capability to link my rear triggers and paddle triggers to letter buttons on the controller, then QGroundControl to assign these with two uses (paddle controls were linked to “A” button, which was set as “shift”. Rear triggers were linked to “X” and “Y” with primary functions to operate the gripper and shift functions to change flight modes. I figure that changing flight mode shouldn’t happen by accident and should require a two button press)). Happy to upload a diagram of my layout for constant gripper use if anyone needs. I could comfortably operate both thumbsticks and use the gripper at the same time.
Issues / Learnings :
Unfortunately, I had a lumen light flood at ~20m. The lights were not really used as the camera was functioning better without them, so this did not present a big issue to the trip. This may be that the housing material did not agree with the cold and warped / shrank. This was the only part of the unit that seemed to have issues with the cold but was interesting with the rating of them being projected at much higher than 20m.
I think a big takeaway from my experience is that I would recommend users who purchased as a full system to take the housing off their BlueROV and understand the wires and systems involved – this can be really helpful when things go wrong and you do not have access to the forums or help – obviously do this at your own risk, but the wiring is very understandable and will help you immensely in understanding how your system works. Users should also take the time to save their vehicle parameters every now and then (I learnt this the hard way).
When our light flooded the problem presented itself as follows:
- A complete loss of communication with the unit in QgroundControl.
- Unit still turning on, lights turning on but flickering occasionally.
- I could connect to the companion through the web UI and read it as OK, but could not see the Pixhawk in the relevant page.
- I opened the unit and could connect directly to the Pixhawk and get parameters in QGroundControl
- I couldn’t reflash Companion due to file size (poor internet connection in remote field location), I reflashed Pixhawk which was clunky through Qgroundcontrol on a poor connection. I note that this can be done with a saved file through the companion, but I had the issue of not reading Pixhawk on the companion. It would be good if the ability to flash from a saved file was possible in Qgroundcontrol when directly connecting to the Pixhawk. Reflashing did not fix the issue. Users should keep an image for both of these locally when deploying in remote environments for just in case.
- Through visual checks, I noticed our rangefinder did not have a light on its control board, and so disconnected this. This returned our video feed and the Pi and PixHawk started communicating again (lights still connected at this point, working but fluttering)
- We now had a working unit with flickering lights, and assumed the rangefinder was the issue and had messed with the Pixhawk. I started to probe the issue of the flickering lights, and found that one of the lights had electrolysis on the interior and was damp.
- The lights were disconnected from the system, and we were back to perfect working order. I tried reconnecting the rangefinder, but this resulted in immediate loss of connection between Pi and Pixhawk.
- Result: It appears a lumen has shorted, which has fed back into the Pixhawk, through to the Companion, and messed up our rangefinder (both systems failed at the same point which indicates they are the same failure), which resulted in a loss of connection between the Companion and the Pixhawk. An unexpected presentation of the root issue, and a good one to keep in the back of your mind when troubleshooting – the systems in the ROV are not completely isolated from each other when a failure point presents.
Summary of my tips for remote field operation and/operating in freezing conditions.
- Newton gripper is a great product. 3D printed jaws work well but require metal mounting support.
- Xbox controllers offer great flexibility in non-standard topside operating environments.
- Batteries hold up well to sub-zero conditions – 1.5hrs runtime with medium workload.
- BlueROV can deal with sub zero conditions well and maintain internal temps. Lumens may not like the cold.
- Low light camera has average imaging resolution, but really great in dim conditions.
- Auxiliary attachments are not completely isolated, and issues may blur in how they present. Consider disconnecting them when issues present as a way of stepping through the issues.
- Open your housing, understand your system.
- Save your parameter files.
- Locally store images for Pixhawk and Companion.
- Locally store the Ardusub guide – very helpful.
- Connect to your Companion, connect to your Pixhawk. Understand how to do this while the unit is working so you can do it when it isn’t.