UROP Day 5

I woke up to a nasty headache, so I took a paracetamol and slept until noon, had lunch and then came to the lab.

Today, the main task was to understand and set up a decent test rig with some relevant obstacles.

Cave Rigging 101

Every cave is vastly different. Some are really tight, and some are just really deep. Many have crazy winding passages and acrobatic overhangs to be able to clip into the rope going down.

Surveys from two caves in Yorkshire Dales. Bull pot has many small pitches with various deviations, rebelays and reaches. Gaping gill on the other is just a massive long pitch (basically 60 meters straight down) with some treacherous dangling traversals. Surveys taken from Council of Northern Caving (https://cncc.org.uk/caving/caves/)

The figure above shows two rigging guides: these are single sided summaries of the pitches and shows the team leader where the ropes have to be set up as they go down the cave. The non-rope sections are usually not shown in this maps. The circled numbers show the length of rope needed, and the other numbers show the length of the pitches.

Basically each section of the cave can be classified into one of the following:

1. Pitches: A long straight section going down
2. Deviation: A carribena is clipped into the rope and pulled to a side to redirect the rope
3. Rebelay: A point where the rope is connected into the rock wall. At this point one needs to unclip all their gear and clip it into the next section to be able to continue
4. Traverse: A series of short sections that are usually generally horizontal and often follows a ledge (but not necessarily)
5. Rope-less sections: There are also sections of the cave that can just be walked or crawled though.

My cave robot should be designed to be able to pass the first 4 types of sections, but the fifth would be really difficult as there is no rope to guide it. When humans cave, they solve the path finding process just by eye and intuition — something really difficult to program a robot to do.

Rack Pot

I attempted to recreate a sample of each of these different knots in the lab to be able to test the robot. Unfortunately, the ceiling of the lab isnt too high, so I could only do really basic sections, each with little tension on them (often in long pitches the weight of the rope exerts a significant tension within the rope).

Rack Pot: The robot testing cave created on a rack in the ICRS lab. (a) shows the overall path (ropes have been highlighted) and (b) shows a schematic. (c-f) Various sections of the rope system: (c,d) The rebelay after the initial pitch. This rebelay is free-hanging and is not near the walls. (e,f) the traverse sections with a deviation-style clamp in the middle.

Rack Pot, as I shall hence forth call it, consists of initially a traverse, a rebelay and then a 1.9m long pitch. The deviation-style carribena in the middle of the traverse is not usually possible in a cave system (as it would become load-bearing) but will be used for testing. It is only possible when the traverse is on a ledge, such that the cavers would be able to stand in this section and the rope is just for added safety.

While Rack Pot does not begin to encapsulate the various types of rigging found in caves (see below), it does serve as a suitable test best for initial practice. The endurance of the robot will be tested later once a functioning robot is built. Beyond that, I feel real life testing in a cave would be the next step — this would let me properly understand the issues faced in a real harsh environment rather than the safe (and dry) lab.

Collection of various rigging styles. Adapted from ‘Vertical’ by Al Warild, 2007 (http://cavediggers.com/vertical/)

With the rig infront of me, the complexity of challenge I laid out for myself has become much clearer than expected.

Sub-Challenges

Using Rack Pot, we can define some key sub-challenges that I can use as a guide to creating my full robot:

1. Be able to clamp onto the ropes. I would like to stress test it with a design load of 1 kg, and a stretch load of 2kg. No idea how close I am to being able to acheive that.
2. Be able to crawl along a rope. Initially, I will try the horizontal sections, and then the vertical sections. And then with a load.
3. Be able to climb up a long rope. This is an endurance test to be able to determine how long a pitch it can climb — theoretically there shouldn’t be any difference between climbing 2 meters and climbing 60m, but I will attempt to try as long a pitch as feasible.
4. Be able to detect a deviation and climb around it.
5. Be able to detect a rebelay and climb around it.

Acknowledgements:

I would like the thank Tanguy from the Imperial College Caving Club for lending me ropes and carribenas to rig Rack Pot and Tom and the rest of Imperial College Robotics Society for welcoming me into their workspace to build and test their robot, and Imperial College Advanced Hackspace for funding this project.