Tuesday, 22 December 2009

2009 Round Up.

This is our last day for 2009, and a good point to finish on, we have the leg running under control of the HexEngine, currently the leg is still missing the metatarsus ram which arrived today, the coxa ram is fitted but needs to be plumbed up. Finishing the hydraulics for the test leg will be our first job in the new year. I'm starting to wonder if we actually needed low friction seals on our rams, it seems there is a fair amount of leakage across the piston seals, that may be causing some undesirable issues. TBC.

The load pin has arrived, so we will also look at fitting this upon our return. I have yet to decide how I am going to wire this in to the HexEngine, I think I will need to modify the FACU box to take auxiliary analogue inputs, or create a new box that will also sit on the 485 bus which will act as an analogue/digital I/O module.

There is much work to do on the FACU control unit, safety features to check, PCB to finalise etc. Also the data stream mode needs to be added to the HexEngine, along with the position/status report packet back from the FACU.

Tuesday, 15 December 2009

A fine adjustment..

After all that beautiful Laser and Waterjet CNC cutting.. it always comes back to a power file :)

Having run the leg for a while, we noticed a little chafing on the hoses, so a fine adjustment with the power file for now, and some retro fitted rollers should sort things out.

Second Cylinder Delivery

Our second delivery of cylinders arrived this morning, we now have the two 40mm MIL type cylinders, one for the tibia, and one for the coxa. Unfortunately we still can't mount the coxa ram as we still don't have the ram mounts which were accidentally left off the order. We will get the tibia ram piped up and commissioned today, we have Jerry coming in tomorrow for his first visit since the project started. It would have been nice to show him a complete working leg. If we had had the complete Parker order on time, this would have been the case!!

Thursday, 10 December 2009

The leg lives!! Very EXCITING :)

Yes, today we have the leg moving under hydraulic control.. Ok, so its only one joint of one leg, but it feels like a major mile stone, and also slightly terrifying :)

We started with simple open loop control of the valve at 50bar, so that we could drive the cylinder from one end to the other. This was working pretty well, although a little underpowered on the annulus upstroke when lifting the weight of the leg. However, we were only running at 50bar. At first we had the tibia down in contact with the floor, I drove the leg in the wrong direction, eg, towards the floor. To our surprise even at this low pressure the leg started to move our test rig which is bolted to the floor and wall.. that's the terrifying part :)

Next we got the leg running under closed loop control and at a higher pressure of 75bar, it's all going pretty well considering its our first hydraulic test, however, we have quite a bit of air in the system which isn't helping. We will spend more time on Monday bleeding the system and give the leg a power test. Josh has an idea of using our engine hoist to check the femur power.. kind of hoist versus leg.. to be continued.

Wednesday, 9 December 2009

Manifold Assembly

Today we assembled the test manifold, to start with we are going to test the cheaper 4WREE valves, which we have two of. We will then install and the 4WRPE valves when get the rest of the cylinders and need to test the complete leg.

We asked a few hydraulic experts how tight we should do up the fittings, most of which are 3/8" and 1/2". The answer seemed to be tight.. chuffing tight! hence the big spanner!

Tuesday, 8 December 2009

The first hydraulic cylinder arrives!

We have just received delivery of the first of our hydraulic cylinders from Parker. Our order has been some what delayed, and sligtly mixed up, so they have shipped this cylinder ahead of the rest of the order to get us going. We have the Femur cylinder, the largest of the cylinders on the Mantis.

We picked up some garden hose so that we can lay out our hose routing and measure our hydraulic hose lengths, we will then order them from hosemaker, a very useful on-line service with a very helpful chap Roger who has been dealing with our endless questions. :)

Friday, 4 December 2009

Test leg assembly together :)

Here it is, our first look at a near complete leg assembly, without the rams and the metatarsus/foot, but most of it is there! Pretty exciting to see this together and get an idea of scale, You can see from the images of Josh and I sitting in the approximate driving position.

We must be Bonkers!!

Thursday, 3 December 2009

Test Leg Frame is Finished!

Josh finished the test leg rig today, this means we will hopefully be able to assemble the coxa and mount the leg by the end of this week :) Looking forward to seeing it come together!

Wednesday, 2 December 2009

Servo Libraries

I have spent the week writing code for the HexEngine servo drivers. I'm trying to make the servo library for both the RX10 and Hydraulic FACU box interchangeable. Hopefully this means I will be able to switch the engine back to the 1/10 test rig to confirm new features, without too much hassle! Well at least that's the plan.

Got the EDU licence dongle for Webots this week, fortunately it is running our simulation fine, so I'm glad I didn't fork out for the Pro version, although no doubt I will need it at some point in the future.

Tuesday, 1 December 2009

Another Setback!

I've been chasing up the hydraulic rams from Parker this week, we were hoping they would arrive by the end of November! Having spoken to Glen our rep from Parker, they think the rams will be with uds by 15th December, a little disappointing as we hoped to get the hydraulic rig running before Xmas. Also it turns out that one of the rams and two brackets won't be in the first shipment?? Apparently they were not ordered, even though we have paid for them! Another virtual item on the "Others Shelf Of Shame"!

God bless Ebay!

I can't quit remember how I stumbled upon these, but I'm sure glad I did! I picked up seven hydraulic pressure transducers for £20 each!! :) well happy with that. We are not sure if we will definitely use all seven yet, but I needed at least one to measure the system pressure, and possible one on each femur ram. I got a good price from SICK for their pressure sensor, of £80 each, so another good saving by picking up these brand new Parker units for £20 each :)

Monday, 30 November 2009

Spot the difference!

We were somewhat surprised at the weight of the four station cetop 3 manifold we ordered from Bosh. The unit weighs 6.7Kg, and as we have seven of them on our machine, that's 46.9Kg of manifold! So I started to look for an aluminium alternative. Our Bosh rep told me that they don't stock alu versions because the don't work so such high pressures as the cast iron version, however, seeing as we are only at 160 bar, this shouldn't be an issue. I managed to find an aluminium version of the exact manifold we have made by Macabor. I also found a pretty good price of £89 each, however, after some further investigation I found the Uk distributor for Macabor, and picked up seven of these aluminium manifolds for £47 each! Well happy with that, saved over £200 pounds and the combined weight of the alu version is 18.9Kg, saving 28Kg from the machine weight.. not to be sniffed at! :)

Thursday, 26 November 2009

Can-Tainer Lid

I machined up a lid for the Can-Tainer today, looking pretty good I think! I struggled to find a good chassis mount RJ45 connector, so I had to use this somewhat oversized bucaneer type socket, however, it would be pretty good for the final version, as it is IP67 rated when used with the matching connector. On the final version I will loose the DB9 and VGA connector so the whole unit should be IP67 rated.

Wednesday, 25 November 2009

3 Phase installed

3 Phase power has been installed today, now we can get on and test the pump :), although we can only run it for a second or two, as we have no oil in the reservoir yet, only a bit left in the pump from the manufacturers test.

Waterjet CNC cut aluminium parts.. Attemp Two!

Josh takes a look at the re-cut parts that have just arrived, they are in a different league to the first batch, some of which we still have and will use. The edges on these are within 0.1mm as was originally suggested, and they are parallel! very nice, our guess is that they re-calibrated the machine??

Take a look at the finish on this piece, its very good for water-jet!

Tuesday, 24 November 2009

Pump Wireing & Control

I have started to wire up the DOL and control panel for the hydraulic pump. I have added an Start/Stop button to the control panel, this controls the pump to tank solenoid valve which shorts out the supply in its off state. Once the motor is started, the solenoid can be switched on using the start button which ports the pump to the main output connector. If there is a loss of power or the stop button is pressed, the solenoid switches back to the off position until start is pressed again.

Should be getting 3 phase installed later in the week so I can test things out!

Friday, 20 November 2009

The Mantis Brain

I have the PC-104 stack built with the Vortex86DX core for the brain of the Mantis. I have added an 8 port RS485 transceiver board with baud rates of up to 921.6Kbps, I don't need all eight ports, at least no yet, but it was the only board I could find that had the high baud rate capabilities, at a reasonable cost. The Vortex board will only go to 460.8Kbaud.

Installation of the serial baord in Linux was a little fiddly, but thats brobably due to my lack of Linux knowledge. Currently I only have a single port running on the 8 port board, well actually I can have any of the eight ports, but only one at a time. So that's good enough for now!

At the bottom of the stack is a PC-104 PSU, that takes 12 to 30V and delivers 5V to the stack.

I have placed the stack in an old CanTainer I had, but I will purchase a new CanTainer for the final unit, and machine the case tops with the appropriate connectors.

Thursday, 19 November 2009

Hydraulic Power Pack Arrives

Ahead of schedule our three phase hydraulic power pack arrived today :) This power pack will be used for the test leg, and no doubt will come in very usfull during hydraulic commisioning. It could also be used to run the hexapod without the on-board engine, just for body moves, but wouldn't have the grunt for any serious walking. The pack is a 7.5Kw three phase unit, delivering 150bar at 29lpm. My next job is to hook up a DOL starter for the motor, along with an E-stop and interlock unit to control the on-board dump valve. This will allow for off load starting of the motor, along with the additional safety.

Next week we will get a single 32amp three phase socket installed in our unit, the pump only requires a 16A socket, but we may as well get a 32 installed just in case.

P.S. I know it looks like Josh is always doing all the work.. but some one has to pick up the camera, and its usually me, honest!

Wednesday, 18 November 2009

Test leg coxa chassis

The coxa chassis is close to being finished. This is one corner of the chassis, which we are using to mount our test leg to. For some reason it seems a little small when you think about sitting at the fron with a leg so close either side!!

Our first hydraulic valves

We have just received the first of out hydraulic valves from Rexroth, very nive they are too! We have two 4WREE and two 4WRPE type valves. We are hoping the 4WREE valves will be good enough, as they are a bit cheaper then the latter type.  These four valves will be used on our test leg rig.

Tuesday, 17 November 2009

Waterjet CNC parts... somethings not right!

It appears that there is a problem with all of our water jet CNC cut parts :(

We had the 25mm 5083 aluminium and 1.75" steel parts cut with a dynamic head water jet CNC machine. The dynamic head is supposed to keep the sides of the material within 0.1mm of square to the face. Without the dynamic head there is a much bigger draft angle.

However, it seems some of the edges on our parts are within 0.1mm, while others are up to 0.5mm! Having taken a close look at the parts, it looks as though the cutting head was misaligned in either the X or Y axis, causing this error throughout the cutting process. This error was further compounded by the steel material being poorly clamped during the cutting process and not being held flat.

Having spoken to the water jet company, they have agreed to re-cut the parts, so today much to Josh's disappointment we packed up all the parts and sent them back. We need to start a new shelf of shame, labelled "Shelf of Others Shame!". :)

Monday, 16 November 2009

The first movment..

Well, the first movement on the Mantis just happened, and for a start I guess it wasn't too bad! Admittedly this is under electrical control, not hydraulic, but still, its good progress.

The PID filter seems to be working pretty well, although as I found out its not a good idea to leave power switched off to the motor amplifer, and leave the FACU running, allowing the integrator to wind up.. when power is restored to the motor amp, you can guess what happens.. and boy does it happen fast!! ahem, partly operator error, and partly badly tuned PID.

The back to back actuator arrangement works ok, but as you can probably guess they tend to get out of sync with each other, which doesn't help, so I may have to go back to single unit.

Electrical test leg complete.

Josh finished the electrical test leg today, so now I can get on and test the FACU controller, however, there is allot of slop in the Igus bearings on the joint.. still waiting for a reply from Igus on this matter, but as it is its not good enough. There is also some miss-alignment of the rotary encoders, but this could have been a drawing error?

Tibia joint with electrical test actuator in place.

Friday, 13 November 2009

Terrain simulation

On a slightly brighter note, after two days of gloom and staring at my HexEngine code, I have made some progress on some additional reflexes in the terrain adaptation of the hexapod, which I have been simulating within Webots. Still have a long way to go.. but it is progress!

Also I reduced the simulated weight and torque of the machine by half, which has helped the sinking issue.

Just been taking a look at MCA2, looks like an interesting machine control/interface system, may take a closer look to see if it would be suitable to convert/port the HexEngine, however this would probably be a massive task that I should probably leave to a later date.

Sloppy bearings :(

Josh has been using our new boring tool today, to bore the holes for the Igus bearings to H7 tolerance. First of all it turns out the cutting tool supplied is the wrong type, so the cutter is chattering and making a mess of the H7 finish!

Then when we do get one of the bearings in place, nice tight fit.. the 30mm anodized shaft also supplied by Igus, has about 0.1 to 0.2mm of slop around it. In fact the hole looks as though it is oval shaped, which we were told would be squashed into a tight round fit on the aluminium shaft once the bearing was pressed into an H7 fit.. hmmm, not the case! Hopefully we will get an answer from one of the tech. chaps at Igus next week.

Well I guess it is Friday the 13th!

Tuesday, 10 November 2009

Slow sinking robot..

I have continued to update the Webots simulation model, including setting the correct joint torques and weight for leg components and body. I thought it would also be good to simulate the ball joint foot at the bottom of the metatarsus.

I have this all running now, including contact sensors which are fed back to the HexEngin via a TCP packet, so that terrain adaptation can be simulated. This worked surprisingly well, until I added the ball joint linkage, it seems to have slowed the simulation down to running at 0.2 x real time. This causes problems with contact feedback to the HexEngine, and the hexapod gets into an oscillating shuffle! I have experienced this in real machines before.. and its usually to do with contact sensor problems or leg speed tuning.

I can remove the ball joint at the end of the metatarsus, but its a bit of a shame, as I wanted to use the angle of the joint to simulate ankle joint position errors. I will post on the Webots forum to see if there is a solution, without purchasing a faster machine ( Currently dual Xeon 2.66Ghz, 4Gb )

The sinking issue: Having made the robot the correct weight, approximately 1600Kg, it now sinks into the floor by about 100mm. II have tweaked various settings in the simulator, including making the feet cubes instead of cylinders which apparently helps, however, this is apparently a feature of the physics engine, so I will have to put up with it, or reduce the weight of the machine for the simulation.

One encouraging point is that the machine was able to stand up at full weight :) which confirms our previous calculations and power requirements. We are aiming for a body weight of approximately 800Kg, the simulation seemed happy with a weight of 1200kg.

Monday, 9 November 2009

PID Valve driver (FACU) test rig..

I need to test my PID valve driver (FACU unit) and it will be a while before the hydraulic components and power pack arrive for our test leg, so we decided to rib up an DC linear actuator to the tibia joint of the test leg. I have taken the 4-20ma output of the FACU unit used to drive the hydraulic servo valves, and converted it into a voltage suitable for a 6/12 amp servo driver I had in my spares box. I will fix the feedback of the servo driver into mid voltage 12ma position so the driver behaves only as an amplifier in open loop mode, then the FACU will close the loop with the absolute encoder on the tibia joint.

Fortunately I had a couple of ultra motion bug actuators left over from an old job. These are powerfull units, with 6" throw each, so to get a bit more range out of them, we join them, back to back and run two together to give 12" throw. We did this once before on one of the Harry Potter films for the Dragon animatronic, usually you would buy an actuator with a long throw, however, we had to use old stock, so this was the best soloution. The two ultra motion actuators back to back are almost the same length and throw as our tibia hydraulic ram, so this should give us a good representation of how FACU behaves!

Josh is working to get this electrical test rig up and running.. as you can see by his work area :)

Friday, 6 November 2009

Waterjet CNC cut aluminium parts..

Our delivery of waterjet cut 25mm 5083 aluminium turned up today :) This was cut using a dynamic head system which nearly eliminates the flare angle of the cut often associated with this cutting method.

The finish is really good, although some of the holes are a little baggier than they should be. We were told if anything the holes would be under size, but that isn't the case with all of them. For the next batch we will alter the drawings to reduce the critical holes size so they are definitely cut under size, and finish them ourselves.

The top and bottom aluminium femur parts are a pretty good fit into the main steel femur section.

Thursday, 5 November 2009

Tourmach up and running..

The Tourmach is installed and running. Now we have the workshop set-up pretty well, I suppose we had better start making something ;) In fact we would have started today, if it wasn't for the fact that our 25mm water jet aluminium parts were sent to the wrong address!

Webots simulation update.

I have taken a look at the webots simulation I started 2 months ago. Its been a while, so it took a bit of head scratching and many cups of tea to get back to where I was, however, the last thing I remember I needed to do was to figure out if I could use the standard £300 version or would I need the Pro £3000 version! Fortunately I have found I can use the standard edition, which is great! I will purchase a copy as soon as my demo version runs out in 30 days or so.

When I first built the hexapod model, I created it as a supervisor, instead of a robot model. I think the reason I did this as because I thought I would need supervisor functionality, which requires the pro edition. As it happens I don't, and I found a simple way to convert my existing world file with supervisor model into the robot model, by editing the world text file :)

I have also got contact sensors working on the simulation hexapod, this means I can send contact information back to the HexEngine so I can properly simulate terrain adaptation. I need to check if the standard version does allow shaped terrain??

Wednesday, 4 November 2009

Installing the Tourmach CNC Mill

We have spent the day removing Josh's Tourmach Mill from his workshop in Chesham. Pretty hefty bit of kit, we tried lifting it on our trust engine hoist.. on the 1/2 tonne position, Josh thought it weight under 1/2 tonnes. Engine hoist wouldn't budge it, turns out it weighs 600kg!. Well at least now we know. Bit of a faff getting it into the back of the van, but no damage done and we both still have all our toes!! must get some steel toe caps.

The most useful bit of kit in our workshop so far.. the engine hoist :)

Monday, 2 November 2009

Back to work!

Having had last week off, both Josh and I are back to work. The first of our hydraulic components arrived today, nothing to exciting: A manifold and a check valve, however, the size of the check valve was a bit of a surprise! gulp!

Wednesday, 21 October 2009

Femur, Tibia & Josh..

Josh with this weeks creations, the steel femur and tibia sections, placed in their approximate locations once the aluminium joint plates are in position. Hopefully our water jet cut, 25mm aluminium plates will be with us in a couple of weeks. Looking forward to see how they turn out!

Tuesday, 20 October 2009

Boxford is a runner..

So we got the boxford running today :) It took a bit of fiddling, as we weren't sure if the problem was the 3 phase converter or the lathe.. it turns out it was a bit of both.

The output of the converter supplies two phases at 415V and one simulated phase, the trick is to make sure the two main phases are the two which are a) reversed on the lathe direction switch, and b) the same two used for the 110V transformer used for the motor contactor.

It took a bit of fiddling, but we got there in the end, Josh is a happy chap.. so is Jimmy who lent us the converter, as he thought it was broken :)

Test Leg Frame

Josh has nearly finished the test leg frame which will support our coxa frame for our initial test leg. We will probably weigh down the frame with concrete blocks, and/or fix it to the wall.. tbc

Shelf of Shame exapnds...

Our shelf of shame expanded somewhat today.. we received the laser cut 10mm steel metatarusu ram mount plates we had to have re made due to a drawing error, the previous parts are already taking up room on the shelf of shame, but the new parts take pole position, and rather encouragingly this time it was not our fault! We ordered 10mm and 5mm steel parts, we received 10mm and 5mm Stainless Steel parts... that's and expensive mistake for the laser cutters.

They have of course apologised and are already re-cutting.. lets hope the next parts are corrrect, I don't think our shelf will take much more :)

Valve PID Control Unit

This box currently called the FACU will close the loop on the four actuators of each leg. The FACU takes four encoder inputs one for each leg, and sends 4-20ma signals down to the VAPS unit for distribution to the appropriate valve (See my earlier post for further info on the VAPS).

As mentioned previously, this design is probably going to change, but this configuration and hardware should surface for our initial test leg.

The FACU has a dsPIC5011 micro controller, and has been programmed in C using the Hitech compiler. The 4-20ma signals are generated using a four channel 10bit DAC, with a 4-20ma loop drivers on each of the outputs. Although the DAC is only 10bits, this only effects the flow rate resolution of the valve, the actual loop resolution is 16bit, effectively reduced to 15bit with out 15bit absolute encoders.

If higher resolution flow control is required, there is a 12bit version of the DAC.

Friday, 16 October 2009

Femur box section almost complete

We welded the femur box section together this morning, it went together really well, the curved steel slots & tabs all fitted neatly together.. which is a good thing, otherwise there would be something seriously wrong with Alibre Design.

We did find out that the metatarsus ram mount brackets are wrong, this was due to an incorrect sized parker ram we had used for our 3D model, so this will need to be re-cut, and our first mistake takes up its position on the shelf of shame...