VaSIMR200
Joined: 12 May 2020
Posts: 5
Location: Wollongong NSW
|
Holonomic Beetleweight - Scale
The Scale project is not my first design, but the first machine I will have built. Initially this idea was pitched for my friend to build as we were both designing featherweights, but I liked the idea so much I had to make it in smaller
Scale
.
Holonomic robots have the advantage of being able to move in all directions. There are many kinds of holonomic drive. Common ones are H drive, 4 wheel omni, 3 wheel omni, Mecanum, Butterfly and Swerve drive (though almost exclusively in FRC). Rotation is also possible at the same time as translation, thus one can keep a weapon facing an opponent at all times while moving freely. The ability to move without turning is also ideal for vertically spinning weapons, as the gyroscopic forces are not a prominent. While it is nice to be able to drift around an arena there are several inherent flaws.
1) More wheels mean more weight. At minimum three independent drive systems must be used, thus adding 50% more weight to a standard tank drive. More common drivetrains use mecanum or quad omni, requiring four independent systems and thus doubling the drivetrain requirements. It is often possible for a standard drive system to 'limp' once some wheels have been damaged, but this is not possible for most holonomic drivetrains. If a wheel is significantly damaged all controlled movement is lost as the functionality of the drive system is dependant of the mathematical relationships of the vectoring wheels.
2) Omni wheels are kinda bad. They have comparatively low traction, are fragile, heavy and expensive. Without the knowledge of moulding or metal printing they cannot realistically be constructed in a combat-ready manner so they must be purchased, at the mercy of the vendors size range and pricing. Tri and Quad omni drives do not have full speed in translation, only in rotation. Mecanum can go full speed forward and back but has the left and right translation + rotation halved. H-drive is the realistic best here as it has full speed in all but rotation, and Swerve drive is the theoretical best with four independently powered and rotating standard traction wheels
3) A good combat robot is almost always a rectangle. All holonomic drivetrains use up large amounts of space and sometimes are not rectangular. Electronics become difficult to fit and manufacturing complexity increases. Weight goes up and durability comes down.
4) These drivetrains are complex and require extra software and hardware to be effective. The most software-complex varieties are mecanum and tri-omni, but even H-drive and quad-omni need a decent amount. Swerve drive is in another league of complex, thus is rarely seen in robot combat (see Radioactive from BB 2015).
All of this leads to a holonimic drivetrain being a rarity in robot combat where low weight, low complexity and reliability are key. While this post has so far spent most of it's text explaining why not to go holonimic, they are a interesting and fun drivetrain with challenges to overcome. For a first machine mistakes will already be made, so why not make
all
the mistakes?
For robot combat the best options out of the holonomic set are Mecanum and Tri-omni. Mecanum can function as standard tank drive, be fairly high in pushing power, have a rectangular chassis, have full speed in forward and backward directions and is not particularly complex mechanically. Tri-omni requires the least components and thus the lightest, omni wheels are cheaper than mecanum ones, traction goes out the window so all power can but put into speed and triangular robots are cool. I have built a 20kg mecanum robot test base as part of FRC, so in order to expand my portfolio a tri-omni setup is the better option.
As mentioned the wheels selection is limited to what the few vendors that sell the decide are profitable. Ofter some digging through the ozrobotics website 38mm PU omni wheels were located for $4 a pop <https://ozrobotics.com/shop/38mm1-5-inch-double-plastic-omni-wheel-with-8-pu-rollers-14184/>. Due to exorbitant shipping costs, an alternative source on Ali Express was the actual vendor for me. At this point the design of the machine was decided: an invertable drum spinner with tri-holonomic drive. Enclosures around the wheels are essentual due to their fragilty and necessity.
The first step in designing the chassis is a base triangle of 180mm side length. Despite the Fusion 360 model reporting that it is in its 306th version, that original triangle is still used for various references and features.
https://imgur.com/a/uZR6wZd
https://imgur.com/a/2Yzr9kN
I have access to a 6040 3-axis CNC mill, and there is 3 and 10mm HDPE sheet for cheap on ebay (free shipping!). Therefore this is the method of manufacture for the chassis. Holes that are not made by the CNC are drilled by hand. Three pieces stacked together permits the replacement of those parts individually, so the whole chassis is not written off with slight damage. Both top and bottom plate are bolted to these parts at various points creating a very rigid chassis. The complex shape ensures a decent amount of material around the entire machine and looks really good. The top plate does not need the machined recessions, but again it looks sick. Add extra machining time to taste (decal) and the design is done. Coming off the CNC the finish is not perfect, so hand sanding is required to make it look the part. As of writing this the chassis has been machined and assembled, the next step it to sand and drill holes.
https://imgur.com/a/p9QHEUO
https://imgur.com/a/c8oFm4g
Armour is fairly light. Unlike sidewinder, all sides of this machine are flat thus relying on the holonomic capabilities to get hit in the right places. 2mm 6061 aluminium armours the wheel pods due to it's low weight and high strength. With the HDPE behind it the intention is to absorb rather than deflect. The front of the machine has two blocks of 10mm 7071 aluminium with the edges machined off. These large ablative blocks can get hit all they want as it is supported by the thickest portion of the chassis. As of writing this, the blocks have been machined and the armour material acquired.
https://imgur.com/a/dJf5tZV
Initially a trio of 40W Turnigy brushless motors were used for drive, but this was upgraded to three Sunnysky X2212 980kv motors with a peak output of 400W. While adding a large amount of power, after this decision the fitting of components became quite a challenge. Each motor also weighs in at 57g, so with four of them 17% of the robot weight is motor. These motors were initially used solely for the weapon, but having all motors be the same simplifies repairs. Attached to the X2212 motors were 9.6:1 'Far Along' gearboxes for their low price, small size and because I thought they were planetary (they are not, currently under trial for durability).
Electronics wise the ESC was a fairly easy choice. My RC experience comes from quadcopters, so the quad-30A controller that weighs in at 10g is a very appealing choice. Not cheap but a decent option for my space and weight restricted machine. I will likely go for a separate ESC for the weapon given weight spare. Fitting 4S 1000mah into the tight space was only just possible, so to play it safe a 3S 1050mah Tattu battery is planned. A fingertech power switch is a placeholder, I plan to make my own of CNCd HDPE, two copper plate and brass bolt switch. The holonomic drive requires some fancy maths, initially I intended for all calculations to be done using trigonometric maths and matrices which require some decent computing power. Back down to earth, a formula that takes the x, y and z component vectors (forward/back, left/right and rotation) can be created that only needs a bit of extra support to make work (scaling values for when the motor power requirement is >100%). An ESP32 or arduino nano will be the processor of choice, taking a PWM signal from the reciever and sending adjusted values to the ESC. With this custom ramping curves can be done as well for the weapon.
https://imgur.com/a/M7q8FRu
One can view scenes of modern films in which someone gets a stick or similar object through the rear wheel of a motorcycle at speed, causing them to rapidly decelerate/crash. This is inaccurate as it is nearly impossible to put an object through a fast spinning spoked wheel as the object will not realistically progress into the wheel faster than the time between spoke gaps. Inspired by this, I wanted to make a 'kinetic reflector' in which small weapon teeth act as the spokes with a hard body for the purpose of deflecting any weapon hits.
The drum has been an interesting design experience, inspired by Robert Corwan's US ant Sergeant Cuddles a hub motor design was chosen. This is the only method I could have chosen for driving the weapon as space available for electronics within the chassis is required. If one were to make more space all the wheels must be equally moved further out along the original radiating lines from the centre, drastically increasing size for little space gain. With a 2350kv motor the top rpm will be very high, intended to be faster than any vertical spinning weapon of an opponent, thus this is the weapon of choice to deal with verts. The first revision had one end supported by a bearing with the other by the motor alone. Weighing in at 250g, the Grade 12 Titanium drum was too heavy for the 0.125" motor shaft of the X2212 as highlighted by everyone I depicted the design to. After actually doing some research into hub motor design, the CAD was changed to have a 6mm hardened steel shaft run the length of the drum, a bearing on either end to support the drum, the shaft be press fit inside the motor stator and the rotor be fir to the drum and have it's mounts turned out so that it is entirely supported by the drum. The bearings are of a strange size, as the end one is imperial (lying around) and a pair of drums have already been machined.
https://imgur.com/a/FvJ4jnk
For more destructive purposes a new weapon had to be designed around the proper hub motor design. An egg-beater style bar was an obvious choice as the main limiting factor of the drum is the bite - only about 2mm. A pair of matching metric bearings ride the drum on the same 6mm hardened shaft with the same hub motor setup. The middle portion has a much deeper tooth to catch corners or deal with horizontal spinners. Despite being lighter than the titanium drum it has a higher MOI, thus can still be destructive. This part is also very complex, requiring a 5-axis mill to machine without multiple setups. Similar to the drum the 6mm axle is flanged at one end and takes a split pin at the other. This axle is a dead one and rides on the 2mm plate sandwitched between the chassis parts.
A question for anyone reading this, would S7 tool steel be a good material? Some form of carbon steel was obvious but I am not an expert on steel grades.
https://imgur.com/a/eHat7zd
https://imgur.com/a/lQlyu6V
A fork configuration was designed almost purely for dealing with D2 kits and the various gas flippers being developed. Six articulating forks ride on the ground mounted to a large manifold block, held in place with 8mm shafts. I hope for these to be made out of 4mm tool steel but that can be changed due to cost. When the robot is inverted the forks will fall into place, causing no change to their effectiveness. A version of the forks without tines will be made to handle horizontal spinners in the rare occurrence they are required.
https://imgur.com/a/JVDNTSZ
https://imgur.com/a/wUTiXbm
Ever design something, spend the money and machine it only to realise an inherent design flaw? The chassis has two points in the corners of the weapon area that have too little material: any large horizontal hit to the armour blocks would have split the chassis in two. This is also bad for retaining the weapon as the drum mounts require fitting to the chassis around it. In order to rectify this issue, the 'split brace' was made. Short on weight, these are 6.5g pieces of that 2mm 6061 that mount in three points per side and hopefully keep the chassis and my sanity together. Also an opportunity to make a cool looking part. Currently there is only one on the top of the machine, but having a second on the bottom will be done given the spare weight.
https://imgur.com/a/zC9l0vh
There are a few potential issues with the design that any future opponents should not read into. Firstly, there is only 2mm of HDPE top and bottom protecting the battery. I may add a small amount of extra plate or padding given spare space, but a powerful axe-bot is my worst nightmare. Due to the addition of the split brace there is only around 1mm of ground clearance at those front bolts. This is potentially OK as the rear wheel can get one out of a stick, but time will tell if any issues arise (the drum misses the ground by 0.5mm so it may be more of a concern). To combat that battery protection issue, there is a configuration that sacrifices the side armour for a 'social distancing' apparatus. Similar to tornado of S6, this should keep some potential punctures further out but increases the chance of getting stuck.
https://imgur.com/a/GM7QZYw
Apologies for the long post, but I aim for this to be a record of the development of my first machine. Ill add photos and progress in this thread when it occurs. If anyone wants access to the CAD model or BOM it will be in this folder of the Cadalogue: https://drive.google.com/drive/folders/1Uz__W2U0FsFKdMOW7KiQD1TgXGmEOzCb?usp=sharing
The Cadalogue aims to be a database of models for anyone to peruse and learn from. Please take a look at the structure of the folders and add anything you see worthwhile. Link: https://drive.google.com/drive/folders/1JaNyuZplEx9Qh7DyDrCMpfummbGhvkWr?usp=sharing
https://imgur.com/a/m0vOrCK
A question for anyone who reached the end: what should be the robot's name? My money is on Slippery Customer but am absolutely up for suggestions.
//
Last edited by VaSIMR200 on Fri Oct 23, 2020 9:12 pm; edited 5 times in total
|
RoboWars Australia Forum Index
-> Builders Reports
