My latest robotic creation is a 1lb fighting robot that competes under the name Algos. Algos replaces a basic 2 wheel drive wedge robot named Kobalos.
The concept for Algos came about when I decided that I wanted to switch to a robot with an active weapon.
I had a few key requirements in mind when designing Algos-
It must have an active weapon capable of damaging the opposing robot
It must be able to fight well even if the weapon is disabled
It must be capable of absorbing a great deal of damage
These requirements led to a few key decisions. First, the weapon needed to be light to allow for the damage absorption and disabled weapon requirements, so I chose to go with a small diameter, high RPM spinning disk. Second, if the robot is going to be capable of fighting even without the weapon, it needed a passive attack system, which led to the general wedged shape and the use of the same drive system as Kobalos, my old 1lb wedge robot. Third, and finally, being able to absorb large amounts of damage while being light enough to stay within the weight limit meant exotic materials would be needed, so I built the entire robot out of 6al-4v (Grade 5) titanium in 1/16 and 1/32" thicknesses. Luckily, due to the size the material cost for the chassis was only $100 for enough material to build two complete chassis.
With the initial concept determined, the next key thing in the development process is deciding how you will make it. I often see people design the parts, then figure out what tool they need to make them. I typically follow the opposite approach and decide upon the manufacturing method before the first component is modeled.
In the case of Algos, I have chosen to use an abrasive waterjet to manufacture all of the components that are not pre-fabricated. The decision to use waterjet from the start influences many of the decisions made when designing Algos. Using 2d part designs reduces potential tolerance stack-up issues as all of the machining can be done at the same time, by the same machine, on the same piece of material for any given part thickness. Using all waterjet fabrication also means that replacement components are relatively inexpensive due to the minimal material waste and speed at which the parts can be made which reduces both material cost for any given part and reduces labor costs if the parts are being fabricated by a company like http://www.bigbluesaw.com/ or http://teamwhyachi.com/.
Parts List:
Fingertech Robotics-
11:1 Silver Spark Gearmotors (2)
TinyESC v2 (2)
Channel Mixer
Lite Hubs with Lite Flite wheels
http://www.fingertechrobotics.com/
Hobbyking-
Turnigy Park 300 brushless outrunner, 1380kV
Plush12 Brushless ESC
R410 OrangeRX
http://www.hobbyking.com/
Robot Marketplace
Thunder Power 325mAh 3S LiPo - G6 Pro Power 65C Series
http://www.robotmarketplace.com/
Cad Files:
http://nearchaos.net/algosdrawings.rar
This contains .dxf drawings of all of the components made for Algos including both generations of center rails.
Algos has competed in three events so far. At Clash of the Bots 3 Algos went 2-2 with several axle failures. At Dragon*Con Robot Micro Battles Algos placed 3rd with the upgraded weapon system. At the Atlanta Mini Maker Faire in the first Atlanta Robotic Combat event Algos went undefeated to take the 1lb championship. Video of Algos can be found at http://www.youtube.com/playlist?list=PLa4DDbMiSUTUSAMtKEfMFhywgw-Bp6Rv-The concept for Algos came about when I decided that I wanted to switch to a robot with an active weapon.
I had a few key requirements in mind when designing Algos-
It must have an active weapon capable of damaging the opposing robot
It must be able to fight well even if the weapon is disabled
It must be capable of absorbing a great deal of damage
These requirements led to a few key decisions. First, the weapon needed to be light to allow for the damage absorption and disabled weapon requirements, so I chose to go with a small diameter, high RPM spinning disk. Second, if the robot is going to be capable of fighting even without the weapon, it needed a passive attack system, which led to the general wedged shape and the use of the same drive system as Kobalos, my old 1lb wedge robot. Third, and finally, being able to absorb large amounts of damage while being light enough to stay within the weight limit meant exotic materials would be needed, so I built the entire robot out of 6al-4v (Grade 5) titanium in 1/16 and 1/32" thicknesses. Luckily, due to the size the material cost for the chassis was only $100 for enough material to build two complete chassis.
With the initial concept determined, the next key thing in the development process is deciding how you will make it. I often see people design the parts, then figure out what tool they need to make them. I typically follow the opposite approach and decide upon the manufacturing method before the first component is modeled.
In the case of Algos, I have chosen to use an abrasive waterjet to manufacture all of the components that are not pre-fabricated. The decision to use waterjet from the start influences many of the decisions made when designing Algos. Using 2d part designs reduces potential tolerance stack-up issues as all of the machining can be done at the same time, by the same machine, on the same piece of material for any given part thickness. Using all waterjet fabrication also means that replacement components are relatively inexpensive due to the minimal material waste and speed at which the parts can be made which reduces both material cost for any given part and reduces labor costs if the parts are being fabricated by a company like http://www.bigbluesaw.com/ or http://teamwhyachi.com/.
Parts List:
Fingertech Robotics-
11:1 Silver Spark Gearmotors (2)
TinyESC v2 (2)
Channel Mixer
Lite Hubs with Lite Flite wheels
http://www.fingertechrobotics.com/
Hobbyking-
Turnigy Park 300 brushless outrunner, 1380kV
Plush12 Brushless ESC
R410 OrangeRX
http://www.hobbyking.com/
Robot Marketplace
Thunder Power 325mAh 3S LiPo - G6 Pro Power 65C Series
http://www.robotmarketplace.com/
Cad Files:
http://nearchaos.net/algosdrawings.rar
This contains .dxf drawings of all of the components made for Algos including both generations of center rails.
Step 1: Component Selection
There are a number of components that I have standardized across all of the 1lb robots I have built in recent years. These components are:
Fingertech Robotics TinyESC v2
http://www.fingertechrobotics.com/proddetail.php?prod=ft-tinyESCv2
The TinyESC is, in my opinion, the best insect weight esc available today. The high voltage capability, current limiting and light weight make it a great option for robots up to 3lbs.
Fingertech Robotics Silver Spark gearmotors
http://www.fingertechrobotics.com/proddetail.php?prod=ft-Sspark16
The Silver Spark gearmotors have proven to be incredibly reliable when combined with lite flite foam wheels. The gearboxes in Algos have been through 5 full tournaments across two robots without failure.
Fingertech Robotics Lite Hubs
http://www.fingertechrobotics.com/proddetail.php?prod=ft-lite-hubs
These are the lightest hub adaptors I know of for lite flite wheels. They're inexpensive and they work well.
R410 4 channel Spektrum compatible receiver
http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=11972
They're light, tiny, and work with DSM2 and DSMX transmitters.
Kitbots Nutstrip
http://kitbots.com/product.sc;jsessionid=60230E830CD664E6E707AD6291A8BCCF.qscstrfrnt04?productId=17&categoryId=1
It's available in a range of sizes and allows for easy assembly of waterjet cut panels.
For the remaining components on the robot, there was a good deal more decision making to be done. I knew I wanted a high RPM weapon, but with sensorless brushless motors there are occasional issues with starting torque, so a balance of torque and peak RPM was necessary. The style of weapon and motor mounting location also meant it would likely see some abuse, so low cost was a priority. When low cost is the goal, Hobbyking is the go to source for components. I eventually settled on a 1380kV outrunner from Hobbyking (http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=14398 ) for the weapon. At 11.1v, the rated voltage of the 3s lipo I was intending to use, the motor would theoretically spin at 15,318 rpm which works out to a tip speed of around 90mph with the 2" disk.
The motor suggests a 10A rated ESC for typical use, seeing as there was very little weight and size increase, I opted for the 12A Plush12 (http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=2161 ) brushless esc to give extra breathing room on that portion of the electrical system.
There's not really a set rule for battery selection, however with lipo batteries, one goal is typically to not drain the battery completely during a match. Matches for 1lb robots are normally 2 minutes long, so I wanted to make sure the robot would have plenty of battery to last a full two minutes. In combat situations, you have current spikes intermixed with low and intermediate draw periods. The TinyESC's limit each motor to 2.8A each, or 5.6A total. The weapon motor claims 7A peak, so when added together, you have a theoretical maximum draw of 12.6A. If through a strange string of circumstances you managed to stay at peak draw for an entire match, you would need to supply 12.6A for two minutes, which converts to a 420mAh battery ( (1000*Amps required) / (minutes of fight/60) ) however, as this is an extreme worst case scenario, a smaller battery can be chosen. I opted for a 325mAh battery under the assumption that 50% average draw would be an extreme case in actual use. I opted for a high end battery pack in this scale, a Thunder Power 3s 325mAh pack ( http://www.robotmarketplace.com/products/LP-TP325-3SPP65J.html ) rated at 65C continuous output, which translates to 21.1A.
The final piece of the puzzle was the mounting of the weapon disk. There are plenty of options on the market today, after looking at several options I chose to try the lightweight set screw hub from ServoCity (https://www.servocity.com/html/lightweight_set_screw_hub__3mm.html ) and have been pleased with the results.
I use a custom power switch for my robots, however there are several off the shelf switches that have been used successfully and removable links are common in the small weight classes. Fingertech Robotics has been developing a small power switch that should be an ideal solution in the future.
Fingertech Robotics TinyESC v2
http://www.fingertechrobotics.com/proddetail.php?prod=ft-tinyESCv2
The TinyESC is, in my opinion, the best insect weight esc available today. The high voltage capability, current limiting and light weight make it a great option for robots up to 3lbs.
Fingertech Robotics Silver Spark gearmotors
http://www.fingertechrobotics.com/proddetail.php?prod=ft-Sspark16
The Silver Spark gearmotors have proven to be incredibly reliable when combined with lite flite foam wheels. The gearboxes in Algos have been through 5 full tournaments across two robots without failure.
Fingertech Robotics Lite Hubs
http://www.fingertechrobotics.com/proddetail.php?prod=ft-lite-hubs
These are the lightest hub adaptors I know of for lite flite wheels. They're inexpensive and they work well.
R410 4 channel Spektrum compatible receiver
http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=11972
They're light, tiny, and work with DSM2 and DSMX transmitters.
Kitbots Nutstrip
http://kitbots.com/product.sc;jsessionid=60230E830CD664E6E707AD6291A8BCCF.qscstrfrnt04?productId=17&categoryId=1
It's available in a range of sizes and allows for easy assembly of waterjet cut panels.
For the remaining components on the robot, there was a good deal more decision making to be done. I knew I wanted a high RPM weapon, but with sensorless brushless motors there are occasional issues with starting torque, so a balance of torque and peak RPM was necessary. The style of weapon and motor mounting location also meant it would likely see some abuse, so low cost was a priority. When low cost is the goal, Hobbyking is the go to source for components. I eventually settled on a 1380kV outrunner from Hobbyking (http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=14398 ) for the weapon. At 11.1v, the rated voltage of the 3s lipo I was intending to use, the motor would theoretically spin at 15,318 rpm which works out to a tip speed of around 90mph with the 2" disk.
The motor suggests a 10A rated ESC for typical use, seeing as there was very little weight and size increase, I opted for the 12A Plush12 (http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=2161 ) brushless esc to give extra breathing room on that portion of the electrical system.
There's not really a set rule for battery selection, however with lipo batteries, one goal is typically to not drain the battery completely during a match. Matches for 1lb robots are normally 2 minutes long, so I wanted to make sure the robot would have plenty of battery to last a full two minutes. In combat situations, you have current spikes intermixed with low and intermediate draw periods. The TinyESC's limit each motor to 2.8A each, or 5.6A total. The weapon motor claims 7A peak, so when added together, you have a theoretical maximum draw of 12.6A. If through a strange string of circumstances you managed to stay at peak draw for an entire match, you would need to supply 12.6A for two minutes, which converts to a 420mAh battery ( (1000*Amps required) / (minutes of fight/60) ) however, as this is an extreme worst case scenario, a smaller battery can be chosen. I opted for a 325mAh battery under the assumption that 50% average draw would be an extreme case in actual use. I opted for a high end battery pack in this scale, a Thunder Power 3s 325mAh pack ( http://www.robotmarketplace.com/products/LP-TP325-3SPP65J.html ) rated at 65C continuous output, which translates to 21.1A.
The final piece of the puzzle was the mounting of the weapon disk. There are plenty of options on the market today, after looking at several options I chose to try the lightweight set screw hub from ServoCity (https://www.servocity.com/html/lightweight_set_screw_hub__3mm.html ) and have been pleased with the results.
I use a custom power switch for my robots, however there are several off the shelf switches that have been used successfully and removable links are common in the small weight classes. Fingertech Robotics has been developing a small power switch that should be an ideal solution in the future.
Step 2: CAD Design and Development
The original CAD drawing for Algos was lost to a hard drive failure and the design had to be redone from scratch. Even with the design being completed once, there still was a great deal to do with regards to design refinement. An additional issue that came up during the design was the original planned weapon motor went out of stock.
The initial CAD model primarily served to create the general shape of the robot. This was the stage in the design where most of the component placement and general geometry was determined, both for the chassis and the weapon system.
The second phase of the design was focused on taking this initial concept and removing enough mass from the chassis to guarantee that it would end up under the 1lb weight limit. During this process, the only area safe from weight reduction was the front wedge, as it was meant to be the primary impact surface.
Shortly after making the final adjustments to the design the motor I was intending to use went out of stock. With an event in the not too distant future, I had to find a new weapon motor and alter some aspects of the design, including the weight of the disk and some of the weapon mount hole locations.
The initial CAD model primarily served to create the general shape of the robot. This was the stage in the design where most of the component placement and general geometry was determined, both for the chassis and the weapon system.
The second phase of the design was focused on taking this initial concept and removing enough mass from the chassis to guarantee that it would end up under the 1lb weight limit. During this process, the only area safe from weight reduction was the front wedge, as it was meant to be the primary impact surface.
Shortly after making the final adjustments to the design the motor I was intending to use went out of stock. With an event in the not too distant future, I had to find a new weapon motor and alter some aspects of the design, including the weight of the disk and some of the weapon mount hole locations.
Step 3: Fabrication
The fabrication process was quite simple for Algos. The entire chassis is made of 1/16" and 1/32" thick titanium that is cut on a waterjet. The weapon disk is made from a piece of hardened 4130 steel that was used on the 2nd version of my 12lb robot, Apollyon. Using the waterjet meant I was able to cut the pre-hardened steel disk without altering the temper.
The two main waterjet resources I use are Big Blue Saw (http://www.bigbluesaw.com/ ) and Team Whyachi. (http://teamwhyachi.com/botshop.htm ) They both do very high quality work. One handy thing with Big Blue Saw is that they are set up to provide instant quotes. I do recommend contacting them directly for projects with a large number of different parts that are cut from the same thickness material, as there are likely more economical options than ordering the parts one at a time, as they would not need multiple setup procedures.
The two main waterjet resources I use are Big Blue Saw (http://www.bigbluesaw.com/ ) and Team Whyachi. (http://teamwhyachi.com/botshop.htm ) They both do very high quality work. One handy thing with Big Blue Saw is that they are set up to provide instant quotes. I do recommend contacting them directly for projects with a large number of different parts that are cut from the same thickness material, as there are likely more economical options than ordering the parts one at a time, as they would not need multiple setup procedures.
Step 4: Assembly
With the Nutstrip, chassis assembly went extremely fast, meaning the majority of the work was wiring. I have included a generalized sketch of the electrical layout of a robot with two drive motors and a single weapon motor. In the case of the weapon motor on Algos, it was brushless so instead of two wires it uses three, however the rest of the layout is identical.
Algos was a very dense robot and due to that, I had to shift several components from their intended positions. Initially the plan was to have the majority of the electronic components next to the battery, however with all of the wires there wasn't room for everything. The receiver was relocated to the weapon motor side of the chassis and secured to the side rail to minimize the risk of it coming into contact with the weapon motor. The drive ESC's are wedged between the back armor panel and the drive motors, keeping them in place and minimizing the amount of space they take up.
Algos was a very dense robot and due to that, I had to shift several components from their intended positions. Initially the plan was to have the majority of the electronic components next to the battery, however with all of the wires there wasn't room for everything. The receiver was relocated to the weapon motor side of the chassis and secured to the side rail to minimize the risk of it coming into contact with the weapon motor. The drive ESC's are wedged between the back armor panel and the drive motors, keeping them in place and minimizing the amount of space they take up.
Step 5: Shaft Replacement and Hardening
Algos went to its first event not long after completing the wiring. With all the armor attached and the systems working, it seemed like it would have a good shot at the event.
Algos went 2-2 at the event, which isn't a bad record for a new robot, however the weapon system just didn't work. The shaft of the brushless motor was steel, however it appeared to be unhardened and had a groove for an E clip cut into it near the front face of the motor. This meant that every good hit resulted in a broken shaft and a disk flying out of the robot. After replacing several shafts I eventually stopped using the weapon.
The question after the event was how to fix it. I decided to take a two part approach to fixing the problem. First, I would add a bearing to the far side to support the shaft on both sides. Second, I would make my own shafts that were longer, didn't have a groove, and were heat treated.
I found a nice length of O1 tool steel on McMaster-Carr for a few dollars and decided that this was a good place to start. Once the material arrived, I cut oversized lengths of O1 to allow it to be sanded to a nice snug fit in the bearings and outrunner can. After checking all of my fits, I picked up a steel can and some 30wt oil based upon the suggestions athttp://hocktools.wordpress.com/2011/01/31/diy-heat-treatment-of-tool-steel/ on DIY heat treating.
Each shaft was heated to a bright orange using a torch then dropped lengthwise into the oil bath. After cleaning off the shafts I did a quick test hit on a spare piece to see how hard the shaft had gotten relative to an annealed piece of O1. The annealed piece bent 90 degrees, the hardened piece snapped off very cleanly. The second stage of the process was tempering. The three shafts and test piece were placed in an oven at somewhere between 450 and 500 degrees F and left to soak for 45 minutes.
The new shafts were pressed into the motor cans and everything appeared to be working well.
Algos went 2-2 at the event, which isn't a bad record for a new robot, however the weapon system just didn't work. The shaft of the brushless motor was steel, however it appeared to be unhardened and had a groove for an E clip cut into it near the front face of the motor. This meant that every good hit resulted in a broken shaft and a disk flying out of the robot. After replacing several shafts I eventually stopped using the weapon.
The question after the event was how to fix it. I decided to take a two part approach to fixing the problem. First, I would add a bearing to the far side to support the shaft on both sides. Second, I would make my own shafts that were longer, didn't have a groove, and were heat treated.
I found a nice length of O1 tool steel on McMaster-Carr for a few dollars and decided that this was a good place to start. Once the material arrived, I cut oversized lengths of O1 to allow it to be sanded to a nice snug fit in the bearings and outrunner can. After checking all of my fits, I picked up a steel can and some 30wt oil based upon the suggestions athttp://hocktools.wordpress.com/2011/01/31/diy-heat-treatment-of-tool-steel/ on DIY heat treating.
Each shaft was heated to a bright orange using a torch then dropped lengthwise into the oil bath. After cleaning off the shafts I did a quick test hit on a spare piece to see how hard the shaft had gotten relative to an annealed piece of O1. The annealed piece bent 90 degrees, the hardened piece snapped off very cleanly. The second stage of the process was tempering. The three shafts and test piece were placed in an oven at somewhere between 450 and 500 degrees F and left to soak for 45 minutes.
The new shafts were pressed into the motor cans and everything appeared to be working well.
Step 6: Remade Components and Titanium Anodizing
At the first event, the center rails of the chassis had taken several large impacts and were bent badly. They also weren't sized for the 3mm ID bearing I was intending to use, so they were remade. In the CAD files on the first page I have both center rails included. The new rails had less weight reduction, as I knew how much weight margin was available. This meant they would be much stronger without risking putting the robot over the weight limit.
While I had the bot disassembled, I decided to try some titanium anodizing based off of the methods in this instructable- http://www.instructables.com/id/Anodize-Titanium/
I used a string of 12 9v batteries in series and some foam soaked in Coke Zero along with a bit of electrical tape to anodize the front wedge. Later I did the same with the internal rails, but instead submerged the titanium into a small pool of Coke Zero to get a more even coating.
While I had the bot disassembled, I decided to try some titanium anodizing based off of the methods in this instructable- http://www.instructables.com/id/Anodize-Titanium/
I used a string of 12 9v batteries in series and some foam soaked in Coke Zero along with a bit of electrical tape to anodize the front wedge. Later I did the same with the internal rails, but instead submerged the titanium into a small pool of Coke Zero to get a more even coating.
Step 7: Last Notes
It took a few events, but Algos is working how I wanted it to at this point. It went undefeated in 6 matches at its most recent event, taking the 1lb championship at the first ever Atlanta Robotic Combat event which took place during the Atlanta Mini Maker Faire.
For the moment, I'm intending to leave Algos as is. I'm sure at some point something will happen that will require another design tweak but until that time, I consider Algos to be complete.
For the moment, I'm intending to leave Algos as is. I'm sure at some point something will happen that will require another design tweak but until that time, I consider Algos to be complete.
No comments:
Post a Comment