Made it home from the Expo in one piece. I had a lot of fun and saw many neat projects. This was the first robot Sumo match and line following contest I've ever seen, pretty neat and lots of fun cheering on the contestants!
With taking apart and putting together the legs on the hexapod I have determined it is necessary to have some type of guide to help lineup the screw holes, the ones that are underneath the bottom plate are almost impossible to get started since I can't see the holes to line them up! I've got an idea to solve this that should be quick and simple, in theory that is... I also need to work on the plug-in panel for the motor leads. It is difficult to reach and align the plugs since it is on an angle from the leg. Ideally it would oriented directly behind the leg and there should be room for all the various sensor plugs also. This one will require some extra thought. It's getting very crowded in that area of the body.
Your project is truly amazing! The amount of thought, effort and attention to detail is admirable. I didn't know about the 1984 ODEX I before the show, and you introduced it to me. You are building a functionally equivalent version using modern materials. When your project is completed, I am sure it will amaze people and dominate any show in which it appears. I enjoyed your talk about designing and fabricating the project. Thank you for sharing your "in-process" project at the Propellerpowered Expo 2013. I am impressed!
I've been taking a bit of a short break from the robot and working on a short project that has been needed for a while but I've been too busy making robot parts to spend any time on it. The project is reorganizing the area of the basement shop where all the programming and electronics work is performed.
That area has been a disorganized mess for a while and adding the new lathe made it worse. So I have 2 new workbenches and full set of wall shelves to put in. This means that some of the older, smaller benches are moving out and repurposed or trashed. So there is a lot of cleanup and finding new places for 'stuff'. In the meantime the robot is still disassembled on the floor and will have to stay there until I create some space for it.
I have started to review Inverse Kinematics websites again. It's amazing that some of the concepts I was earlier having issues with understanding are much clearer now even though I haven't studied it much over the last year. Hopefully that translates into the ability to write some code!
Finished up putting in the new workbenches and removing the old. That really opened up the space that I'll need for testing walking gaits and terrain navigation later. Here are some photos of the hexapod with all six legs attached:
It's staring to look like a real robot now! The globe is not mounted, I only put it on there to see if it would be the right size for the robot, didn't want too small or large. The top deck in the photos will be the computer/electronics deck and that will be 6-8 inches tall and the globe goes on top of that. I will mount a pan and tilt mech with color camera(s) on it for remote viewing. The maximum height right now is about 35 inches from the floor to the top of the computer deck. Add 6-8 inches for the computers it will be 41"-43" tall with a 12" globe for a final height of around 54 inches.
I'll be wiring up one leg for calibration and testing again. Then I want to try out some basic Kinematics routines and see how it responds. That will have to wait a bit as I'm getting ready for a week of shipwreck diving in upper Michigan in the Lake Huron and Michigan with our local dive club. Tanks are filled and the boat is ready. Time for in-depth robotic programming after I get back!
It's bigger than I was mentally expecting also! I knew what the dimensions were but to actually see everything put together...
Diving is done, had some great dives on freshwater wrecks but its time to get back to wiring a leg up and start some programming tomorrow.
I need to run the calibration routines first and then run some basic movements to check for potential binding that I might have missed during assembly and make sure everything is tight. I want to try out some of the inverse kinematics routines and see how accurate they are in positioning the leg tip.
My wife heard the dreaded words, 'I'm going to re-design part of the robot'. She usually cringes and then asks how much it's going to cost... It seems she has had some experience with my redesigns. Anyway, I haven't been happy with the battery compartment and the way the power distribution is laid out. The setup is difficult to access and making changes requires quite a bit of disassembly to get to the parts. I've been thinking of ways to solve these issues and I came up with a new idea that moves all the power distribution down to the battery level, adds space for a second battery and provides much easier access.
The first photo is the original battery compartment. It was sized it to fit the battery exactly which doesn't leave any room for changes. The next photo shows the square box in the middle of the body that contains all the power distribution wiring. Its pretty full already and I've only wired up 3 legs; plus it takes up a lot of space. The third photo is the new battery compartment. It has a hex base and there is room for 2 batteries to stack on top of each other with room to spare. The 4th photo shows the battery placement with filter caps still leaves a lot of room for the power distribution relays and wiring. I'm heading down to the local boat surplus store in town and see what they have for 12v distribution modules. If I can find something already designed for power distribution it could look better than whatever I cobble together. I will mount a aluminum sheet in there vertically and put all the power distribution components on it so they are in one place and the board can be removed for modifications. There will be exterior panels all the way around the compartment to hide the inner workings.
The next item to solve was dealing with an issue I experienced at the Expo when I was attempting to put the legs back on the robot. The bottom bolts were very difficult to put in for each leg because the bolts are upside down and there is no way to easily line up the holes. This required some scrap 1/8 x 3 delron that was cut to provide alignment for the bottom of the leg. If needed I can do the same thing for the top bolts. The photos show the setup and how they work.
I got my 32 channel radio transmitter ready to go. This is a Q4 I got through the Quantum Robotics (http://www.quantumri.com/) Kickstarter project. The controller uses the XBee modules (I'm using the 60mW Series 1) and under the hood is a Parallax Propellor chip. So it is fully programmable for whatever you need with 4 joysticks (with pushbutton switch), 4 potentiometers, 6 toggle switches and 10 pushbutton switches. The only thing I had to add was a battery and the XBee module. I'm using a Prop Activity board for testing the transmitter and initial connections to the robot. I think this will be a great way to control the robot with plenty of ways to send commands to it. I'll be looking for some example codes to play with and get used to the setup.
I had to order some more stuff tonight, ran out of ring crimp connectors and 2 and 3 wire connectors. I was soldering all the motor leads, microswitches and potentiometer outputs for the computers and got most of them completed. I'll finish the soldering once they show up.
My wife heard the dreaded words, 'I'm going to re-design part of the robot'. She usually cringes and then asks how much it's going to cost... It seems she has had some experience with my redesigns. Anyway, I haven't been happy with the battery compartment and the way the power distribution is laid out. The setup is difficult to access and making changes requires quite a bit of disassembly to get to the parts. I've been thinking of ways to solve these issues and I came up with a new idea that moves all the power distribution down to the battery level, adds space for a second battery and provides much easier access.
The first photo is the original battery compartment. It was sized it to fit the battery exactly which doesn't leave any room for changes. The next photo shows the square box in the middle of the body that contains all the power distribution wiring. Its pretty full already and I've only wired up 3 legs; plus it takes up a lot of space. The third photo is the new battery compartment. It has a hex base and there is room for 2 batteries to stack on top of each other with room to spare. The 4th photo shows the battery placement with filter caps still leaves a lot of room for the power distribution relays and wiring. I'm heading down to the local boat surplus store in town and see what they have for 12v distribution modules. If I can find something already designed for power distribution it could look better than whatever I cobble together. I will mount a aluminum sheet in there vertically and put all the power distribution components on it so they are in one place and the board can be removed for modifications. There will be exterior panels all the way around the compartment to hide the inner workings.
The next item to solve was dealing with an issue I experienced at the Expo when I was attempting to put the legs back on the robot. The bottom bolts were very difficult to put in for each leg because the bolts are upside down and there is no way to easily line up the holes. This required some scrap 1/8 x 3 delron that was cut to provide alignment for the bottom of the leg. If needed I can do the same thing for the top bolts. The photos show the setup and how they work.
I got my 32 channel radio transmitter ready to go. This is a Q4 I got through the Quantum Robotics (http://www.quantumri.com/) Kickstarter project. The controller uses the XBee modules (I'm using the 60mW Series 1) and under the hood is a Parallax Propellor chip. So it is fully programmable for whatever you need with 4 joysticks (with pushbutton switch), 4 potentiometers, 6 toggle switches and 10 pushbutton switches. The only thing I had to add was a battery and the XBee module. I'm using a Prop Activity board for testing the transmitter and initial connections to the robot. I think this will be a great way to control the robot with plenty of ways to send commands to it. I'll be looking for some example codes to play with and get used to the setup.
I had to order some more stuff tonight, ran out of ring crimp connectors and 2 and 3 wire connectors. I was soldering all the motor leads, microswitches and potentiometer outputs for the computers and got most of them completed. I'll finish the soldering once they show up.
Bob Sweeney
Bob,
Are you going to do the receiver portion by youself. I thought that these guys where going to come through with a receiver, but as far as I can tell, they have not.
Are you going to do the receiver portion by youself. I thought that these guys where going to come through with a receiver, but as far as I can tell, they have not.
Jim
That was the first I've seen that project. I'm planning on using a propellor with a XBee as the receiver computer, probably using the activity board in the photo in the previous post above. I haven't worked through all the details yet on what the controller system will look like. There will have to be some experimentation with networking multiple Propellor chips and figuring all that out. I know what I'd like to have the robot be able to do, figuring out the best way to do it isanother task! i think using the new P2 chip would be great but it sounds like it might be delayed some more due to some issues they have discovered. In the meantime I'll experiment with the original prop and get a lot of testing of the leg programming completed. Hopefully the path forward will become clearer.
I was hoping to finish up the power distribution wiring this weekend but ended up working on the mechanical layout of the new battery box instead. I found a great boat 12v power distribution block that also incorporates blade fuses and will make the wiring easier to work with. So I tore apart the original distribution wiring and laid everything out on a aluminum sheet. After figuring out how to mount this piece I ran the mill drilling holes at 60 degree angles and then tapping all 36 holes. I made some custom aluminum standoffs for the sheet and got that mounted. The remaining holes are for the black plastic covers that will be the outer skin on the battery box. I need to find some fairly stiff black plastic sheets that I can cutup for this.
Outside of fabricating the outside panels, the battery box is put together. Should get some time this week to do some wiring. I want to get the power distribution done before I start playing with programming the transmitter and reciever. Does anyone have some working spin code for this type of transmitter, Q4, I can use as an example? I hate to have to re-invent the wheel if there is some usable code already available!
I'll get some photos posted of the distribution layout soon.
Finishing up the basic wiring for power distribution. Here is the distribution layout board and a closeup of the power control switches.
I cut out the top deck for the computer layer with some plastic panel I got from the local boat shop, its called starboard and is really nice and easy to cut and shape. It is cut out to be 0.7" wider on each side than the aluminum panel since the motor controllers extend out that much. Here is another view of the switch plate, it turned out pretty nice but I need to work on the lettering some more!
I came up with a design for the plugs for the motors and various sensor feeds back to the computers. I will be drawing this one up on the CAD/CAM program and then milling out 6 of these. The sides of the battery box will be covered with black plastic panels, and there will be a cutout for the switches. Next will be installing all the wiring for sensor feedback into the computer and finish wiring up the legs. I ordered and got the remaining 2 and 3 prong plugs needed to finish up that part of the build.
If I don't get distracted by designing the computer deck, I'll wire in a controller and start playing at programming one leg. I've been sketching out some interesting designs for the top computer deck and I may just jump straight into completing that first.
I haven't posted for a bit as my work schedule has increased by a lot during refueling at the plant, so I'm putting in a lot of 13 hour days. On my days off I am getting some work done on the robot, just not as much as I hoped. Here is an overview of the current progress.
This is the plug panel for all the 2 and 3 connector outputs from the leg. These are the 3 leg position potentiometers, Left and Right swing limit switches, the leg down switch and the leg impact sensor. I added several extra spots for future enhancements on the board. There were 6 of these created on the CNC mill, only took about 9 minutes to mill and a couple of hours on the CAD /CAM to come up with the design. I tried my hand at engraving but that didn't turn out as well as I hoped. I'll have to keep trying, I think I need a to order a diamond engraving tool in order to get what I'm looking for.
Next step was figuring out how to mount the panel. I added the motor controller panel to the same setup.
I wanted to be able to easily add and remove the legs so the panel needed to be mounted behind the leg mounts. I couldn't mount them on the bottom base plate as the swing motor was in the way. So that left attaching them to the top base plate so that they hang down. This should work out pretty well as the top base plate is also where the computers will be installed so the wires from the plugs will go up through the plate and get plugged into the computer boards. The motor controller panel is also easy to reach with its mounting also.
I got the plastic panels for the battery box so those were cut to size and drilled. Everything installed as expected and it looks pretty nice. Each panel is held in place with 6 screws and one panel has a cutout for the power control switches. Typing this up I see that I forgot to install the battery changer port... add it to the list!
Today I will be working on several miscellaneous little items that need to be done but are time consuming. I'll see how much I can get done as I'm back to work tomorrow. Still working out the design for the computer level. I want to have some type of RGB light display system that will be a visual aid for troubleshooting. I just haven't quite figured out what it should look like. I've done some experiments with a RGB light string and shining it through the edges of acrylic. I'm looking for a way to diffuse the light so it doesn't look like individual LEDs but more a general glow. Time for more experiments and see what I can come up with.
Diverbob
What I like the most about this message thread is that you are doing a complete mechanical build
before concerning yourself about actually programming.
The programming is what actually makes a project like this work,
BUT
The mechanical build is actually what makes it real.
Thanks and keep going
I look forward to every one of your posts.
Diverbob
What I like the most about this message thread is that you are doing a complete mechanical build
before concerning yourself about actually programming.
The programming is what actually makes a project like this work,
BUT
The mechanical build is actually what makes it real.
The mechanics are fun but with a hexapod there are a lot of parts involved! Everything is in multiples of six so a mistake cost a lot of time and money! So I try to take my time so I don't make decisions that will have unforeseen consequences down the road (already hit a few of those 'consequences' from earlier decisions). My previous attempt at a hexapod was filled with mistakes resulting in numerous rebuilds and starting over time and again. This one is much more deliberate design from the start and I think overall the process has gone well.
Since by profession I'm a software developer, I really, really want to get started programming! But there is a limit to my programming without all the parts installed and power available. If I jump into programming too soon then the mechanical parts that are needed for all the functionality would be delayed and I would have to revisit and redo software later. I still haven't taken the time to design the swing limit switch assembly. I built a space into the leg specifically for one, just need to prototype one and try it out. If it works then make 5 more! There are a couple of other items like that to go also.
Thanks for following the build! I knew at the start it would be around a 3 year project, almost 2 years are done and I can feel the end of this first phase finishing up. I don't think there will actually be an end to this project as the platform can hold all kinds of sensors and new ideas for expansion (how about an arm or three?)
Did some more work on a troubleshooting aid for the robot. What I've come up with is a light bar that will be located between the main body and the computer level. It will be about 3/4 inch thick and made with clear acrylic panel material. The acrylic will be frosted and will look like a white band going entirely around the robot base. The acrylic is being machined to 3/4 inches thick and 1.5 wide. There are 12 pieces, one for each face on the body. Behind each acrylic panel will be mounting for either 2 or 3 RGB LEDs. When lite, the LEDs light will be diffused and the panel will glow with the selected color. Each LED can be programmed to a specific color to signify whatever event or condition the robot is experiencing. For example, a leg impact can cause a red light on the side where the impact occurs, a blue light might signify the leg is down. Not only can this be used for troubleshooting merely by watching the lights but it would make a neat lighting system.
The CAD files are done and ready to be cut out next. I got some more starboard for mounting the light layer parts onto. I want to build this section as a complete module that can be easily bolted on and removed. I'll post photos as they become available.
Got about another couple weeks of extended hours before I get back to somethIng more reasonable. I should be able to start getting things done faster than it's been the last month!
Where or what are u using for the connectors on the Coxa motors. I have the same ones from surplus but with the lead screw. And I can't find the connector big enough to slide over the spade.
Where or what are u using for the connectors on the Coxa motors. I have the same ones from surplus but with the lead screw. And I can't find the connector big enough to slide over the spade.
I couldn't find a specific connector so I used 3/16 spade terminals and put heat shrink around the terminal. Not the ideal solution but it works until something better comes along.
I noticed that this thread is almost 2 years old with over 22,250 views and 291 entries! It's a lot larger/longer lasting than I ever expected. Although I started this project knowing it would keep me busy for a while, my initial guess was 3 years, I continue to be surprised at the number people tuning in!
The place where I work that pays for all this fun is winding down the refueling outage so I'm back to normal hours starting next week. Luckily the next long outage isn't until fall of next year. This week has started the move back to normal working hours so I have time to work in the evenings again. Tonight I finished up milling the top and bottom plates for the computer deck. I also started rough cutting the plastic panels for the sides on the computer deck. They are slightly oversized and I'll write a quick program to cut them to size and put a 15 degree angle on the sides so the 12 side pieces fit together snugly. I have the code for the acrylic panels done, will cut that this weekend. By that time I can put together some of the photos and post them here. I keep a camera in the shop just for documenting the creation process, so far there are several hundred photos and videos made. I try to pick the most representative ones for posting here.
I finally got time to work on the computer level mechanicals. I had some issues that required a lot of adjustments to the CAD drawings which resulted in some errors that showed up during assembly. That required more visits to the computer to see what could be done to fix the problem. This shot is of the main deck for the computers. They will all be mounted inside this location. this 12 sided piece is not even all the way around. Every other side is either a 'long' or 'short' length. This is because the width of the motor controller skin is wider than a leg. The 6 short 'slots' are for wires coming up from the motor controller level. The other holes are attachment points for the walls and the light level components (haven't started machining those yet). The large holes are access points to the screws that hold the motor controllers and legs to the top aluminum plate. It will all become clearer later, I'll put together a video overview on the level constructions.
Instead of using posts like in the rest of the body to connect the various plates, I wanted the outer skin to be the mounting point to maximize interior space. Since this is a 12 sided bottom, 12 side pieces of 0.25 thick starboard are used. There are 6 each of the 2 sizes on the body. Each one is cut to 4" height and a 16 degree angle is cut in the sides of each piece. Someone is now calculating that 12 sides on a 360 degree object comes out to 15 degrees. However, I cut the angle a little deeper to allow the corners to come together better when you take into account any inaccuracies in the side lengths.
I've had some experience with threaded holes in plastics. Anything that has to be taken apart and put back together a number of times will eventually get the threads stripped. To combat this I use small brass threaded inserts. These are 6-32 and only require a 0.188 (3/16") dia. hole. An arbor press is used to insert them into the drilled holes. This makes for a very sturdy connection that will last a long time.
Here is the sides connected to the bottom plate. The dimensions were right and everything fit together the first time without any filing or sanding! Granted the last side was tight but it fit!
Here is a view of the computer deck with the top plate installed.
The next item was prepping for the acrylic globe that goes on top. I cut a 2" collar of 5" dia cast acrylic and used the lathe to smooth the cut edges. On the side where the globe will rest the surface was cut at an angle to match the globe. I will be using acrylic glue to mount the 2 parts together. I will have to cut the opening on the acrylic globe larger to match the interior dimensions of the collar but I haven't figured out a way to do that easily. It may be a case of carefully using the Dremel tool and sharp cutters to get that taken care of. I didn't like how high the globe stood on the collar so I cut the collar height down to about 1" which I think looks much nicer. I used the mill to drill 6 holes in the collar to match the holes on the top deck. After threading the holes, screws were used to mount the two parts. I'll be working on the globe problem tomorrow. I will have to order some more acrylic glue, I seem to be out of it now. It doesn't take much for it to evaporate away from most containers.
Here is a shot of what the collar and globe will look like on the top of the computer deck. I took this with the 2" collar and you can see the neck of the globe that needs to be removed.
The next step tomorrow is to start cutting out the parts for the light level. The light level is only 3/4" thick and goes between the computer bottom plate and the top plate for the motor controller level. I need to re-examine the CAD drawings for this to make sure the changes to the computer base plate don't impact the light level design.
After several false starts I got a successful run at making the acrylic light conduits that go between the motor controller level and the computer deck. I got 4 made but had to stop when a major storm tore through the area with 80+ mph winds, rain and the occasional tornado to duck. No power now other than my DSL wireless modem that is on battery backup and an iPad. I'll give I bit more time for the next line of thunderstorms to come through and the start up the generator.
The parts finished so far are fitting together nicely. I will have some minor filing to do to remove the little nubs that keep the acrylic clamped down when I'm milling all around the part. Photos will follow later once I get some power again!
Power is back on but the high winds are still going strong. I don't want to be operating the mill under these conditions, if power goes out or flickers during a cut I. Could get some unwanted results. I'll continue tomorrow and finish the light bar. So tonight will be a relax night instead.
Got back in the shop this evening and finished up the light bar work. This photo shows one of 6 acrylic parts that are 3/8" thick
These show the bottom of the computer level and how the individual pieces are mounted. The large holes along the outside end are for access to the screws that hold the top of the legs and motor controllers attached to the top aluminum plate. The slots are for wiring to come up from the motor controller level.
This is what the light level looks like when mounted on top of the motor controller level. I decided to only go withr one layer of acrylic instead of 2 like I originally planned. I like the look of the single layer and it isn't really noticeable with the lights off. Plus there is more light available that I expected.
Of course I couldn't resist finding a RGB light strip and just wrapping it behind the light bar and applying power. The colors come out very well and the lights are well diffused. I need to locate some digital RGB LEDs that I can now mount and wire up. This looks really nice right now. I still like the idea of individually programming the lights as a sort of poor mans troubleshooting guide where different colors are used above each leg and motor controller to show what commands it is executing and show what sensors are seeing. In operation the lights could be flashing a variety of colors that mean specific things if you know the purpose for a specific color. Otherwise it will be a nice light display!
This is a view of the inside of the computer deck and the top cover with the acrylic shoulder mounted with 6 screws.
Lastly, here is a full view of the robot body without legs. Inside the globe will be the video cameras with pan and tilt units. I will build a little platform for those so that the cameras are centered in the globe area once I get the equipment and know the dimensions. I still haven't attached the globe, still debating on the best way to cut out the opening to match the acrylic shoulder. Since I only get one shot at this I want to be very careful not to make a irreversable mistake!
Next steps will include doing some wiring and getting power to the motor controllers and then the motors themselves. The plug panels need to be wired up also so I've got a lot of crimping and soldering to do shortly. There is still several relatively minor machining items to work on like the bottom impact and side swing limit switches. Both of them need delrin housing designed and installed but those should be fairly quick and easy. I'm doing a 'stay-cation' next week so unless I get really lazy I should get a bunch done. As always I'll keep you posted on progress. Getting closer to that day of applying power for the first time!
I took the week off from work and getting a lot done. Big moment today, flipped the switch on the power distribution system and no magic smoke appeared! All 18 motor controllers came to life at the same time, lots of pretty green lights. Spent the last couple of days crimping 204 ring terminals and 60 Molex connector pins. My hands are beat and sore.
Will be cutting the slots in the top aluminum plate for the sensor wires to pass through on the mill next. My RGB light strip arrived today from Adafruit. It's a bit wider than expected so I'm going to have to figure out how to fit a 1/2+ " strip into a 3/8 area. Since the computer base is 1/2" thick, I may mill a slot in it. That could be a good way to anchor the lights as a bonus.
Waiting for a Parallax order next, this is a pack of servo extension wires. These will be soldered to the sensor plugs so the sensors can use standard 3 pin headers for connecting to the computer.
It's been an exciting weekend so far! I took a break from the shop and went on our Dive Club annual Turkey dive (30+ years of tradition) in the local river. I was working my way upstream in 17 feet of water and discovered a body. The police were called and the person was identified only a couple of hours ago. Usually I just want to go play with the catfish in the river on this dive, they get kind of lethargic when the water is only 39 degrees and don't swim away. No catfish this time but plenty of other excitement. I marked the location, surfaced and had one of the other divers come over to verify that I hadn't just found a well dressed manikin. Unfortunately my first impression was the right one. By the time the police were done most of the day was shot and I only got a little time in the shop tonight.
Here is the progress on the light and computer levels. I cut the access slots in the aluminum plate so the sensor wires for each leg can be accessed on the computer deck. This photo shows all the sensor wiring that has to be hooked up.
The Adafruit RGB displays were wider than expected so I milled a 1/4 wide by 0.22 deep slot all the way around the bottom of the computer deck. This should work out well, the clear covers on the RGB lights will hold the lights in tightly place. I may use a dab of hot glue to keep them from falling out if needed. I had to cut up the LED strip into 12, 2 LED sections. For each leg 2 LED sections were soldered together with wires so the lights lined up correctly behind the acrylic. On the input end a 4 pin connector was soldered on to make for easy attachment to the computer board.
I also got 50, 3 wire servo extensions that will be soldered onto the ends of all the sensor wires. This should help tame that unruly mess of wires and make it easier to plug into the computer board. I need to find some good wire labels that I can use to mark each connector, I don't want to have to trace each wire back to a leg in order to make sure they are plugged in correctly. This is tomorrows job, doing all this soldering. This should keep me busy for a while!
The robot is slowly taking shape and looking like what I was thinking about. Once these remaining tasks are done I'll breadboard up a Propellor and start programming once again. Got a lot of routines to start developing, Looking forward to the challenges that will bring.
Comments
With taking apart and putting together the legs on the hexapod I have determined it is necessary to have some type of guide to help lineup the screw holes, the ones that are underneath the bottom plate are almost impossible to get started since I can't see the holes to line them up! I've got an idea to solve this that should be quick and simple, in theory that is... I also need to work on the plug-in panel for the motor leads. It is difficult to reach and align the plugs since it is on an angle from the leg. Ideally it would oriented directly behind the leg and there should be room for all the various sensor plugs also. This one will require some extra thought. It's getting very crowded in that area of the body.
Bob
Your project is truly amazing! The amount of thought, effort and attention to detail is admirable. I didn't know about the 1984 ODEX I before the show, and you introduced it to me. You are building a functionally equivalent version using modern materials. When your project is completed, I am sure it will amaze people and dominate any show in which it appears. I enjoyed your talk about designing and fabricating the project. Thank you for sharing your "in-process" project at the Propellerpowered Expo 2013. I am impressed!
Dennis
That area has been a disorganized mess for a while and adding the new lathe made it worse. So I have 2 new workbenches and full set of wall shelves to put in. This means that some of the older, smaller benches are moving out and repurposed or trashed. So there is a lot of cleanup and finding new places for 'stuff'. In the meantime the robot is still disassembled on the floor and will have to stay there until I create some space for it.
I have started to review Inverse Kinematics websites again. It's amazing that some of the concepts I was earlier having issues with understanding are much clearer now even though I haven't studied it much over the last year. Hopefully that translates into the ability to write some code!
Bob
It's staring to look like a real robot now! The globe is not mounted, I only put it on there to see if it would be the right size for the robot, didn't want too small or large. The top deck in the photos will be the computer/electronics deck and that will be 6-8 inches tall and the globe goes on top of that. I will mount a pan and tilt mech with color camera(s) on it for remote viewing. The maximum height right now is about 35 inches from the floor to the top of the computer deck. Add 6-8 inches for the computers it will be 41"-43" tall with a 12" globe for a final height of around 54 inches.
I'll be wiring up one leg for calibration and testing again. Then I want to try out some basic Kinematics routines and see how it responds. That will have to wait a bit as I'm getting ready for a week of shipwreck diving in upper Michigan in the Lake Huron and Michigan with our local dive club. Tanks are filled and the boat is ready. Time for in-depth robotic programming after I get back!
Bob
Diving is done, had some great dives on freshwater wrecks but its time to get back to wiring a leg up and start some programming tomorrow.
I need to run the calibration routines first and then run some basic movements to check for potential binding that I might have missed during assembly and make sure everything is tight. I want to try out some of the inverse kinematics routines and see how accurate they are in positioning the leg tip.
Stay tuned! Bob
The first photo is the original battery compartment. It was sized it to fit the battery exactly which doesn't leave any room for changes. The next photo shows the square box in the middle of the body that contains all the power distribution wiring. Its pretty full already and I've only wired up 3 legs; plus it takes up a lot of space. The third photo is the new battery compartment. It has a hex base and there is room for 2 batteries to stack on top of each other with room to spare. The 4th photo shows the battery placement with filter caps still leaves a lot of room for the power distribution relays and wiring. I'm heading down to the local boat surplus store in town and see what they have for 12v distribution modules. If I can find something already designed for power distribution it could look better than whatever I cobble together. I will mount a aluminum sheet in there vertically and put all the power distribution components on it so they are in one place and the board can be removed for modifications. There will be exterior panels all the way around the compartment to hide the inner workings.
The next item to solve was dealing with an issue I experienced at the Expo when I was attempting to put the legs back on the robot. The bottom bolts were very difficult to put in for each leg because the bolts are upside down and there is no way to easily line up the holes. This required some scrap 1/8 x 3 delron that was cut to provide alignment for the bottom of the leg. If needed I can do the same thing for the top bolts. The photos show the setup and how they work.
I got my 32 channel radio transmitter ready to go. This is a Q4 I got through the Quantum Robotics (http://www.quantumri.com/) Kickstarter project. The controller uses the XBee modules (I'm using the 60mW Series 1) and under the hood is a Parallax Propellor chip. So it is fully programmable for whatever you need with 4 joysticks (with pushbutton switch), 4 potentiometers, 6 toggle switches and 10 pushbutton switches. The only thing I had to add was a battery and the XBee module. I'm using a Prop Activity board for testing the transmitter and initial connections to the robot. I think this will be a great way to control the robot with plenty of ways to send commands to it. I'll be looking for some example codes to play with and get used to the setup.
I had to order some more stuff tonight, ran out of ring crimp connectors and 2 and 3 wire connectors. I was soldering all the motor leads, microswitches and potentiometer outputs for the computers and got most of them completed. I'll finish the soldering once they show up.
Bob Sweeney
Bob,
Are you going to do the receiver portion by youself. I thought that these guys where going to come through with a receiver, but as far as I can tell, they have not.
Jim
That was the first I've seen that project. I'm planning on using a propellor with a XBee as the receiver computer, probably using the activity board in the photo in the previous post above. I haven't worked through all the details yet on what the controller system will look like. There will have to be some experimentation with networking multiple Propellor chips and figuring all that out. I know what I'd like to have the robot be able to do, figuring out the best way to do it isanother task! i think using the new P2 chip would be great but it sounds like it might be delayed some more due to some issues they have discovered. In the meantime I'll experiment with the original prop and get a lot of testing of the leg programming completed. Hopefully the path forward will become clearer.
Bob
Outside of fabricating the outside panels, the battery box is put together. Should get some time this week to do some wiring. I want to get the power distribution done before I start playing with programming the transmitter and reciever. Does anyone have some working spin code for this type of transmitter, Q4, I can use as an example? I hate to have to re-invent the wheel if there is some usable code already available!
I'll get some photos posted of the distribution layout soon.
Bob
I cut out the top deck for the computer layer with some plastic panel I got from the local boat shop, its called starboard and is really nice and easy to cut and shape. It is cut out to be 0.7" wider on each side than the aluminum panel since the motor controllers extend out that much. Here is another view of the switch plate, it turned out pretty nice but I need to work on the lettering some more!
I came up with a design for the plugs for the motors and various sensor feeds back to the computers. I will be drawing this one up on the CAD/CAM program and then milling out 6 of these. The sides of the battery box will be covered with black plastic panels, and there will be a cutout for the switches. Next will be installing all the wiring for sensor feedback into the computer and finish wiring up the legs. I ordered and got the remaining 2 and 3 prong plugs needed to finish up that part of the build.
If I don't get distracted by designing the computer deck, I'll wire in a controller and start playing at programming one leg. I've been sketching out some interesting designs for the top computer deck and I may just jump straight into completing that first.
Bob
This is the plug panel for all the 2 and 3 connector outputs from the leg. These are the 3 leg position potentiometers, Left and Right swing limit switches, the leg down switch and the leg impact sensor. I added several extra spots for future enhancements on the board. There were 6 of these created on the CNC mill, only took about 9 minutes to mill and a couple of hours on the CAD /CAM to come up with the design. I tried my hand at engraving but that didn't turn out as well as I hoped. I'll have to keep trying, I think I need a to order a diamond engraving tool in order to get what I'm looking for.
Next step was figuring out how to mount the panel. I added the motor controller panel to the same setup.
I wanted to be able to easily add and remove the legs so the panel needed to be mounted behind the leg mounts. I couldn't mount them on the bottom base plate as the swing motor was in the way. So that left attaching them to the top base plate so that they hang down. This should work out pretty well as the top base plate is also where the computers will be installed so the wires from the plugs will go up through the plate and get plugged into the computer boards. The motor controller panel is also easy to reach with its mounting also.
I got the plastic panels for the battery box so those were cut to size and drilled. Everything installed as expected and it looks pretty nice. Each panel is held in place with 6 screws and one panel has a cutout for the power control switches. Typing this up I see that I forgot to install the battery changer port... add it to the list!
Today I will be working on several miscellaneous little items that need to be done but are time consuming. I'll see how much I can get done as I'm back to work tomorrow. Still working out the design for the computer level. I want to have some type of RGB light display system that will be a visual aid for troubleshooting. I just haven't quite figured out what it should look like. I've done some experiments with a RGB light string and shining it through the edges of acrylic. I'm looking for a way to diffuse the light so it doesn't look like individual LEDs but more a general glow. Time for more experiments and see what I can come up with.
Bob Sweeney
What I like the most about this message thread is that you are doing a complete mechanical build
before concerning yourself about actually programming.
The programming is what actually makes a project like this work,
BUT
The mechanical build is actually what makes it real.
Thanks and keep going
I look forward to every one of your posts.
Garyg
Since by profession I'm a software developer, I really, really want to get started programming! But there is a limit to my programming without all the parts installed and power available. If I jump into programming too soon then the mechanical parts that are needed for all the functionality would be delayed and I would have to revisit and redo software later. I still haven't taken the time to design the swing limit switch assembly. I built a space into the leg specifically for one, just need to prototype one and try it out. If it works then make 5 more! There are a couple of other items like that to go also.
Thanks for following the build! I knew at the start it would be around a 3 year project, almost 2 years are done and I can feel the end of this first phase finishing up. I don't think there will actually be an end to this project as the platform can hold all kinds of sensors and new ideas for expansion (how about an arm or three?)
The CAD files are done and ready to be cut out next. I got some more starboard for mounting the light layer parts onto. I want to build this section as a complete module that can be easily bolted on and removed. I'll post photos as they become available.
Got about another couple weeks of extended hours before I get back to somethIng more reasonable. I should be able to start getting things done faster than it's been the last month!
Bob
The place where I work that pays for all this fun is winding down the refueling outage so I'm back to normal hours starting next week. Luckily the next long outage isn't until fall of next year. This week has started the move back to normal working hours so I have time to work in the evenings again. Tonight I finished up milling the top and bottom plates for the computer deck. I also started rough cutting the plastic panels for the sides on the computer deck. They are slightly oversized and I'll write a quick program to cut them to size and put a 15 degree angle on the sides so the 12 side pieces fit together snugly. I have the code for the acrylic panels done, will cut that this weekend. By that time I can put together some of the photos and post them here. I keep a camera in the shop just for documenting the creation process, so far there are several hundred photos and videos made. I try to pick the most representative ones for posting here.
Bob
Instead of using posts like in the rest of the body to connect the various plates, I wanted the outer skin to be the mounting point to maximize interior space. Since this is a 12 sided bottom, 12 side pieces of 0.25 thick starboard are used. There are 6 each of the 2 sizes on the body. Each one is cut to 4" height and a 16 degree angle is cut in the sides of each piece. Someone is now calculating that 12 sides on a 360 degree object comes out to 15 degrees. However, I cut the angle a little deeper to allow the corners to come together better when you take into account any inaccuracies in the side lengths.
I've had some experience with threaded holes in plastics. Anything that has to be taken apart and put back together a number of times will eventually get the threads stripped. To combat this I use small brass threaded inserts. These are 6-32 and only require a 0.188 (3/16") dia. hole. An arbor press is used to insert them into the drilled holes. This makes for a very sturdy connection that will last a long time.
Here is the sides connected to the bottom plate. The dimensions were right and everything fit together the first time without any filing or sanding! Granted the last side was tight but it fit!
Here is a view of the computer deck with the top plate installed.
The next item was prepping for the acrylic globe that goes on top. I cut a 2" collar of 5" dia cast acrylic and used the lathe to smooth the cut edges. On the side where the globe will rest the surface was cut at an angle to match the globe. I will be using acrylic glue to mount the 2 parts together. I will have to cut the opening on the acrylic globe larger to match the interior dimensions of the collar but I haven't figured out a way to do that easily. It may be a case of carefully using the Dremel tool and sharp cutters to get that taken care of. I didn't like how high the globe stood on the collar so I cut the collar height down to about 1" which I think looks much nicer. I used the mill to drill 6 holes in the collar to match the holes on the top deck. After threading the holes, screws were used to mount the two parts. I'll be working on the globe problem tomorrow. I will have to order some more acrylic glue, I seem to be out of it now. It doesn't take much for it to evaporate away from most containers.
Here is a shot of what the collar and globe will look like on the top of the computer deck. I took this with the 2" collar and you can see the neck of the globe that needs to be removed.
The next step tomorrow is to start cutting out the parts for the light level. The light level is only 3/4" thick and goes between the computer bottom plate and the top plate for the motor controller level. I need to re-examine the CAD drawings for this to make sure the changes to the computer base plate don't impact the light level design.
That's all for now!
Bob
The parts finished so far are fitting together nicely. I will have some minor filing to do to remove the little nubs that keep the acrylic clamped down when I'm milling all around the part. Photos will follow later once I get some power again!
Bob
These show the bottom of the computer level and how the individual pieces are mounted. The large holes along the outside end are for access to the screws that hold the top of the legs and motor controllers attached to the top aluminum plate. The slots are for wiring to come up from the motor controller level.
This is what the light level looks like when mounted on top of the motor controller level. I decided to only go withr one layer of acrylic instead of 2 like I originally planned. I like the look of the single layer and it isn't really noticeable with the lights off. Plus there is more light available that I expected.
Of course I couldn't resist finding a RGB light strip and just wrapping it behind the light bar and applying power. The colors come out very well and the lights are well diffused. I need to locate some digital RGB LEDs that I can now mount and wire up. This looks really nice right now. I still like the idea of individually programming the lights as a sort of poor mans troubleshooting guide where different colors are used above each leg and motor controller to show what commands it is executing and show what sensors are seeing. In operation the lights could be flashing a variety of colors that mean specific things if you know the purpose for a specific color. Otherwise it will be a nice light display!
This is a view of the inside of the computer deck and the top cover with the acrylic shoulder mounted with 6 screws.
Lastly, here is a full view of the robot body without legs. Inside the globe will be the video cameras with pan and tilt units. I will build a little platform for those so that the cameras are centered in the globe area once I get the equipment and know the dimensions. I still haven't attached the globe, still debating on the best way to cut out the opening to match the acrylic shoulder. Since I only get one shot at this I want to be very careful not to make a irreversable mistake!
Next steps will include doing some wiring and getting power to the motor controllers and then the motors themselves. The plug panels need to be wired up also so I've got a lot of crimping and soldering to do shortly. There is still several relatively minor machining items to work on like the bottom impact and side swing limit switches. Both of them need delrin housing designed and installed but those should be fairly quick and easy. I'm doing a 'stay-cation' next week so unless I get really lazy I should get a bunch done. As always I'll keep you posted on progress. Getting closer to that day of applying power for the first time!
Bob
Will be cutting the slots in the top aluminum plate for the sensor wires to pass through on the mill next. My RGB light strip arrived today from Adafruit. It's a bit wider than expected so I'm going to have to figure out how to fit a 1/2+ " strip into a 3/8 area. Since the computer base is 1/2" thick, I may mill a slot in it. That could be a good way to anchor the lights as a bonus.
Waiting for a Parallax order next, this is a pack of servo extension wires. These will be soldered to the sensor plugs so the sensors can use standard 3 pin headers for connecting to the computer.
http://forums.parallax.com/showthread.php/151952-Another-Fine-Product-Discontinued
Here is the progress on the light and computer levels. I cut the access slots in the aluminum plate so the sensor wires for each leg can be accessed on the computer deck. This photo shows all the sensor wiring that has to be hooked up.
The Adafruit RGB displays were wider than expected so I milled a 1/4 wide by 0.22 deep slot all the way around the bottom of the computer deck. This should work out well, the clear covers on the RGB lights will hold the lights in tightly place. I may use a dab of hot glue to keep them from falling out if needed. I had to cut up the LED strip into 12, 2 LED sections. For each leg 2 LED sections were soldered together with wires so the lights lined up correctly behind the acrylic. On the input end a 4 pin connector was soldered on to make for easy attachment to the computer board.
I also got 50, 3 wire servo extensions that will be soldered onto the ends of all the sensor wires. This should help tame that unruly mess of wires and make it easier to plug into the computer board. I need to find some good wire labels that I can use to mark each connector, I don't want to have to trace each wire back to a leg in order to make sure they are plugged in correctly. This is tomorrows job, doing all this soldering. This should keep me busy for a while!
The robot is slowly taking shape and looking like what I was thinking about. Once these remaining tasks are done I'll breadboard up a Propellor and start programming once again. Got a lot of routines to start developing, Looking forward to the challenges that will bring.
Bob Sweeney
Travis.